Anthony douglas log
ose developer test:
1)I did not install linux because my tiny computer cannot handle two operating systems or a virtual machine. I would need to buy a new computer to do that. I can do that when the time comes, however it is not practical right now and this is better than nothing. I can still do all the required tasks just as easily in windows, freecad in particular runs find in windows.
2) I am learning freecad anyway, and have surpassed the minimum requirements laid out in freecad 101. I have spent maybe 10 hours of actual productive time on it. I already know a lot of CAD, so that helps a lot. I have used onshape, fusion 360, and solidworks already. Solidworks I used professionally.
3) documenting the work did take some time, to learn to use kdenlive. It took maybe 1.5 hours to figure everything out. The audio especially. video:
Subtract some points for not installing linux I guess, however I did above average with using freecad, so I give myself 80 percent total.
I have made also a video in which I use the body placement parameters editing approach with multiple bodies in the part design workbench instead of the a2 work bench, and could edit and upload that as well, however that is not good practice for creating a CAD document though. It is better to use constraints, so things are exact, will update automatically in the event of changes, and there are no errors, and it also took half the time to make the assembly. Plus you can get motion and other benefits from the assembly. When I have a look at some of the other test videos, there are some serious problems with the way people are making stuff there, with the parts not fitting together exactly, etc. When you go to manufacture something, the parts have to fit together. I noticed that is happening a lot with people using freecad, they do not know that you need to use references etc. and work precisely when doing cad, because this is not included in the tutorials anywhere. I could stand to learn a lot more cad and then make a decent manual, but that will have to be in the more distant future. There are other people who would be much better suited to doing so, but they are not doing it.
My email address, for reference, is Anthony.firstname.lastname@example.org. I'm putting this up after seeing only now done stuff on the talk page for Anthony Douglas, which I only discovered by accident in the recent changes page of the wiki.
I don't think I recurve any kind of email notification for edits to the talk page, and there is a lot of funny stuff going on with the usernames. My username is actually Anthony, but Anthony Douglas, and if I go into the history of this log, and find my own edit, then click on the talk page, then there is no such page.
So yeah. The talk pages are not a good way to communicate.
The wiki works in pull mode, in which we go and find stuff on the logs of other people, but doesn't work in pull mode; of you have something to say to someone, they are unlikely to hear it I think through the wiki.
More critically, I have been reflecting on the enormous waste that we have suffered by not keeping an index of people, a rolodex, who are interested in ose. Networking like that is one of the only things that really adds up over time. I have said this many times over many years. I suppose that the user logs may allow for some of this information to be recovered.
May 25 2019
I was glad to be able to do some analysis of proposed router structures with the use of fusion 360. A powerful piece of software, but not at all open source, plus you don't even have control over your own data, and it is impossible to pirate. We really need to have more practical and powerful open source software in this domain. For now, we can still design open source things using the equivalent of closed source compilers, but it is asking for trouble. Software engineers generally use closed source hardware, I would note. There needs to be more cross talk in these domains.
The latest study is here: https://a360.co/2VXH4HS
You can browse around and see the details there. In one of the studies, which is one of the number one ones, the most realistic evaluation is done, using abs plastic blocks. However they are quite big, almost 40 cm long and solid abs, which could be quite expensive and time consuming to print. If it was 3d printed it would usually be an honeycomb or similar structure.
here is an older version https://wiki.opensourceecology.org/wiki/File:Unnamed.png which shows the result of an analysis with the universal axis freecad model taken from the wiki, simply scaled up so the rods are increased in size from 8 mm to 2 inches, and then arranged to form the structure as shown. All parts are simply assumed to be a steel block with a youngs modulus of 210 GPa. Different types of steel, hardened or not, have mostly the same youngs modulus. This indicates the force vs. distortion that the material exhibits (not the tensile strength, which tell you at what pressure it breaks).
3d printer statement
Ok, so I decided to write a statement about 3d printers. I am among the first to stay with developed technology, the immediately implementable things. And see there is a certain poorly explored golden zone there. However talking about e.g. new three d printer processes and foundational thought is a different ball game than making a whole project orbit around far fetched speculation while even the most rudimentary real hardware in hand right noe would be worth way more... I think moving around often along a broad spectrum from speculative all the way to on the ground hands on development and even improvisation is valuable.
Ok so clearly 3D printing is considered very important. I have learned a lot over the years of interest and it is important for people to share at least the most important bits when they see that they have something to add. It may not seem like it makes a big difference but it is one of the few things we can do that adds up over time.
CNC maching sort of sucks. The capital costs are vast and I have steered away from development of machine tools, including a project for a four axis printer subsitute, because the details really really add up. It was originally developed as an anti labor union move and in many ways is inherently unsuited to the purposes of working people in the current state our little world.
Frankly FDM type printing really sucks. We need better stuff. Despite the vast number of highly trained person hours that have gone into the body of work which is e.g. thingiverse, there is sparing little useful stuff on there.
Cnc machining is *sort of* like the big boy version. If you check out cnccookbook.com for instance, you can find many examples of people buying extraordinarily expensive machines to do essentially some fraction of what people want to but cannot do with printers, in terms of system output; that is the parts that are spit out by the system.
But if you investigate the input requirements, you will find one of the main motivations behind 3DP dev. And secondly the geometric limits.
Inputs are mainly location, capital cost, labor time (this includes so much expanding would be fair), stock material, and cutting tools. Uh. From whats on the top of my mind right now.
I have many times and at great length considered the possibility of a very small auto feed five axis mill turn system and process as filling the role of being able to produce useful things yet having the other main desired characteristics of 3d printers. Undoubtedly there is potential there. There is a five axis open source mill for $3k available. Add the other stuff and $5k is more like it for a complete system. And you'd have to pirate some five axis cam software as there is no other option for that whatsoever. Or develop some though that is quite a task. That is a very interesting peice and I may have gotten one if I was in a higher economic bracket.
I think it may be possible to use polymer concrete, integral fluid bearings and some other desigm elements to greatly reduce the financial expense, but that is another story.
In the end though the sheer amount of input the process requires still is many many times what 3dp does for the parts it can make. The geometry limits are still pretty severe, too.
Additive manufacturing can overcome some of the inherent limits. It is currently very limited though, just not fundamentally. Although volume change upon deposition is pretty close to fundamental.
I must have looked into thirty different printing processes at this point, and I think the most promising marginal options are, not in order:
1. Inkjet powder bed molds followed by vacuum polyurethane or aluminum casting. The guerillacnc site has a very good quality (honestly this is the sort of stuff we need) site on polyurethane casting. Combined with the unlimited general geometries (no support sructures etc needed), high tolerances, relatively good surface finishes, high strength and temperature resistance, looks right at hand.
Both processes are low cost and known to work fine. Just put them together. There is an open source powder bed printer but it only gets 96 dpi right now. Maybe the cartridge can do more but the developers haven't gotten around to adding that. Also simply hacking a normal inkjet makes a lot more sense than making your own like that in many ways. The peizo actuator inkjets are relatively uncommon but what you really want so you can use a range of binder fluids, iirc there was only one inkjet printer I could find that used peizo. The others use small heater elements to vaporize the fluid and produce pressure so if there is nonvolatile stuff in the fluid I don't know if that would work well. Maybe. For gypsum or salt pure water might be fine anyway and useable with most injkets.
There is one group that used salt as the powder and got good results that I recall seeing years ago. That no one has tried using such printers for casting seems unlikely but who knows, and I cannot find any evidence of it.
2. Tungsten cvd and lithography or milling and support material. The central unknown here is the chemistry and chemical engineering side of doing it yourself, because for a useable operating cost you would have to recycle the tungsten back to the gas precursor I think. Buying the gas does not sound practical.
There is some extremely promising stuff with nickel cvd. However the deposition rate is probably limited to something like 30 microns per hour, trying to increase the rate leads to powder being deposited instead of the nickel integrating with the already solid nickel crystal structure. Nickel cvd is widely used and there is a lot of pressure to increase the rate but this is the best industry has, from what I read, been able to do, so attempting to improve it on a shoestring seems unpromising.
Stress free tungsten however, I have read in patents, can be deposited relatively easily at mm per minute rates, I think it was. It was well within useable anyway. This makes it one of the hard (for me anyway) to find routes to stress free metal addition.
So you start with a blank build plate, add a layer of photolithography stuff, expose it where desired, wash away the exposed material (or the converse), deposit tungsten, including on the top surface of the mask. Then so called planarizing, you machine away the material above the mask top, and just a little bit of the mask's upper layer to get the desired layer thickness. Now you have a pattern of tungsten and photoresist where desired. Repeat for more layers. Complete functional devices fully assembled can be made.
This has already been done for microdevixes with nickel, including a fully assembled robotic gripper with quite a few parts to very high tolerances. The time is not a problem if the device is so small.
This type of process is obviously sounding very expensive and hard to develop. It pisses me off that there is not more done at the higher level on it. They can experiment with microdevices and tungsten for super niche applications like maki g rocket nozzles (the patent was related to that I think). But they seem to make no effort to put them together to make something that would really be welcomed by the people - of course as that is not who they serve. The ratio of effort required to value and applicability to the problems we face of the research done is orders of magnitude greater if directed to suitable subjects.
The development effort to prove the concept here would similar to the nickel microdevices stuff but have far more profound implications.
And granted perhaps the greatest value for nasa would be to develop fundmental stuff like the cvd process itsself which has broad applications, but they don't really do that much either actually. As an aside I have noticed how nasa research on e.g. water recycling and food production etc. is surprisingly applicable, perhaps because they must return to fundamentals much as we seem to in order to make progress, the usual approahes being so so far from what we really want.
3. Liquid metal deposition followed by stress relief, ceramic support material and milling.
There is a group at cornell I think it was that has been doing work on what thet call Shape deosition manufacturing, which is a higher level concept in which material is deposited crudely then milled into shape at each layer.
There are a lot of combinations possible, so after exploring quite a bit I have tried to get some more immediate progress by focussing on considering the particular combination of use of tool steel as build material, stress relief to a moderate degree using electropeening during the build process and or maybe ultrasound, ceramic support and three axis milling with a ceramic cutting tool. This seems to be a peak in the effort vs value diagram. Using steel instead of aluminum is not looking any harder to me because to mill aluminum you need lubricant, but not for steel, for example.
The support material is a central unknown. It must have certain characteristics to some degree, the more the better in some cases. Be a fluid and then become a solid. Have sufficient mechanical strength to stand the forces encountered from shrinking metal which can be relatively high. The metal must adhere to it to some degree so it is held against the milling and stress of deposition.
The materials used in ceramic shell casting meet these generally. They usually use some amount of not thermally stable binder though which might be complicating as it decomposes at the casting temperature. It also needs to be removeable though.
The only really promising options I found are friction bonded solids, in which particles of a size distribution all the way down to colloidal are mixed in a slurry deposited solvent evaporates and the actual friction between particles adds up enough to give a strong solid. Ultrasonic consolidation can be another or complementary. Sound waves causes relative motion between particles and the heating and abrasion between particles from friction leads to them becoming welded together. That might be impractical to remove. There might be some etchants that could remove it but leave the metal alone.
I don't know about the friction bonded solid removal. Ultrasound might in contrast shake particles appart I suppose. A lubricant like water scootched in there with the aid of vacuum maybe.
If a solvent soluble binder that can stand the heat can be found or if the outgassing that happens is not too bad with the commercially available.
Zircon is the usual material for shell casting steel.
The other is the post deposition stress relief as mentioned above in the tungsten cvd stuff. There is a contractive stress that forms as a result of the cooling of the metal on each layer just like fdm printing. Unlike plastic though we may be able to rlief the stress to a manageable degree at each layer because metal is malleable. Then at the end of the print process the object being embedded in the support material to prevent motion of long arms or anything we could anneal it all for nearly stress free parts.
On an almost personal or perhaps methodological note, I can now say with some reasonable authority that the milling side of this can work fine. However I remember many times thinking before I had gotten into milling that we knew that from the general characteristics and developed nature of milling as a process. In other words I did not need to know a lot about milling, nor would the developer of such a system. This seems to me to be a useful note on the reality that people can probably use tools around us like the process which is milling, in such an innovative manner without spending a ton of time on the details. The fundamentals and core realities do get you a long way probably.
Another point that I often see others will bump into after investing more time then they need to if I could just tell them now:
Productivity rate and per print reliability. I often hear people talking about producing quantities of things that you would probably never make with 3dp. It is only one process among many manufacturing processes and is ultimately not that special except its suitability to anticapitalist activity or making do with small amounts of capital. The amount of time it would take to print numbers of large objects makes the time it takes to build and use injection molding or extrusion etc look like a bargain.
Among the central values it has are the divisibility of capital (that is the amount of capital needed to make anything at all. $1 k vs $15 k for a mill for instance), low retooling cost, reduced input requirements, relatively low geometry limits in most ways, and the first hop from bits to atoms that it can provide.
Some random model of maker bot apparently gets 4.5 cc per minute. (http://makerblock.com/2010/08/whats-the-cost-of-printing-with-a-makerbot/ ,the second bullet point)
That's 103 hours per cubic foot. To not have a malfunction in that time seems highly unlikely from my experience.
This leads to something else that cuts right to the core of Ose, the maker movement and things like this: when people who are deeply interested in the subject and have spent a lot of time on it etc. focus on three d printers going into a role that makes no sense, it highlights at least two things: we need to understand the existing body of work of manufacturing techniques way better and our current understanding is extremely rudimentary.
Secondly there is a sort of paradox with the idea that on the one hand great projects can be completed well, like the development of a superior 3d printer, while at the same time a hopelessness around implementing things that are very conventional, like injection molding.
Perhaps a distant third is the reality that we really want and see that even basic versions that are fully done of certain systems, like a printer, would be worth so much more than the literally dozens of partially done ones... and yet we have little in the way of fully done stuff to offer others. Thank goodness for the reprap project, because the sum total of all supposedly open source printers out there would not, I am betting, be worth a damn to someone who wanted to move onto the next step like holding printer build days. They are nowhere near ready from what I have seen.
I just want to post quickly to apologize for missing the last webinar. I have every intention to come to the next one this Friday and have set my phone to remind me.
I did some research on open source extruders. I would post them here but they show up immediately upon googling "open source filament extruder", and I am on my phone here so it would be time consuming.
They sound reasonably straightforward, but I feel like it is important to express what I am thinking instead of sitting on it: I do not think it is the best use of development time for ose to produce another such design. They sound mature enough and it is a different ball game from the larger machines that OSE is best currently set up for.
Secondly I think it is important to recognize the secondary degree of utility, in contrast to the primary nature of the actual printer. The value of his very small scale plastic recycling is in truth very low except to reduce print costs, and even then I have read filament is only $5 per kg or so when you shop around and get the generic stuff. It takes a long time for a printer to extrude that much ultimately. Compared with the capital cost and other running costs od the printer it is not that bad.
The amount of waste material recycled is extremely small relatively. The idea of feeding poorly founded and shallow desires like the enthusiasm of the environmentalist for filament extruders from recycled plastic concerns me a great deal. I have long seen the strength of the OSE concepts as coming largely from the deeply well founded and close to the root issues they do. Recycling of plastic just for filament should be displaced by more critical things just as steel recycling is. More so because large objects which take a lot of plastic make no sense to print, the process being too slow whereas steel can be recycled and used in large amounts more sensibly.
March 3 (2)
After attending the webinar on Monday, I very much want to give an update on what I have been up to, and how it relates to open source enterprise. I am typing on a phone so please forgive the stilted language. also I should apologize for my apparent passivemess during the webinar, I hate to derail anything though.
I also would really like to talk to the others in the webinar some time.
Ultimately I have spent a great deal of time thinking and reading and puzzling on many of these subjects, and very little time talking with let alone collaborating with others on the subject. Talking may seem boring but I think a certain kind is actually thinking together. I hold the opinion that more good quality thought and knowledge integration will lead to greater traction. Yet it is also immediately at hand, unlike things like reliable funding sources and so forth.
Ok, after departing from OSE, I had resolved to pursue open source hardware entrepreneurship full time as an income source. I visited some maker spaces during my adventures, some in new york, then Amsterdam (Ifabrica, the Waag society, the one where TOOOL is based) and Vienna (Happylab and Metalab) by way of combining my travelling with a certain longer term adventure :).
In Vienna I tried to employ the cloud design local production approach as a modern equivalent to paying for ones travels by making bracelets. This took the form of laser cutting designs I got from combing the web, most of which were to be found on thingiverse. I did produce a variety of decorative boxes, domino sets, fruit bowls and some other things. They were the most interesting designs I could find. I since lost the pictures. I did not try 3d printing because of the severely limited potential for useful things.
I encountered some limitations and barriers of the sort I expected must exist in some form. Social barriers to machine use, the laser at Happylab malfunctioned and lasered my finger, then they almost blamed me for frying it... obtaining suitable stock material in the artificial consumerism centric environment of the city, modifying designs without the actual 3d source files, to laser precision issues and excess soot buildup. Despite high hostel prices that obligated long commutes, sleeping on rooftops and other things that ate into my productive time, I made a few things.
Ultimately it became clear the design base was too limited and the laser is also badly limited especially when the usual impositions of a shared machine show up, egoism and politics included. I would need to make my own designs, and milling looked more promising.
I vacationed a bit more then returned to Toronto and joined Site3, Hacklab and the east end tool library makerspace. I learned to use and gained access to several milling and routing machines, and Inventor, obtaining cutting tools and atock and fixturing and trieed to produce, gearshifting to increasingly rudimentary projects when the barriers in this context became better understood. I spent 3 months or so in Toronto, but had no open source project successes, my time going into working and navigating housing, a place for deskwork type stuff etc.
The way these things pulled me away in Toronto led me to decide to leave for Winnipeg, where Assentworks had several machine tools sitting around I hoped I could use in exchange for repairing, and find compadres to collaborate with, and hopefully a place to do some things on my computer. I had found that outside of the collaborative environment and nearby housing and reasonable basic hand tools etc at ose even minor projects seemed impossible. Even being in Canada is a serious drawback in comparison as sourcing parts is much more expensive and time consuming for instance.
At Assentworks I did find the people running the space had laid a good foundation in some ways, especially regarding large machines which they succeeded at getting grants for. A Tormach 4 axis mill which needed substantial repairs, a five axis shopbot only one person had ever touched just to test it in the year since they got it, and a seven axis robot arm based milling system that to this day no one has gotten working in the slightest, except for the few test runs I got in before leaving.
I spent a lot of time trying to get and learn the Cad and cam software, and the proprietary machine controller software, and not having a useable deskwork like environment. I worked to repair the Tormach, establish the toolchain and get the wrinkles worked out for it and the five axis shopbot, and the three axis shopbot working with hsmworks.
I tried very hard to stick with truly useful things, and not go down the hole that I commonly see others succumbing to, which is producing decorative etc. items. Frankly almost nothing was produced by anyone in that space. Almost all revenue came from consulting stuff the clients probably wasting their money in many cases or contracts for expensive decorative installations and deskwork stuff. Also the occasional print job.
The exceptions I knew of was a guy who engraved weed grinders on hos own laser cutters, and a few people made cutting boards. There was one character Andrew who was doing much like I had wanted to do in ottawa and having good success, effextively developing and implenting moderately sophisticated and utilitarian electromechanical systems, but singlehandedly not in collaboration with others aside of course from the important foundational collaboration to sustain the space.
I did learn a lot of Inventor, Solidworks, Mastercam, and HSMworks as well as trying out some other cam software. I decided to focus on becoming good at five axis milling and solidworks, sourcing parts and supplies, and the all important innovation glue logic of inventing with by and for the people. Clearly there is a lot you can make with five axis milling although the slowness and convoluted nature of the process claws back of the iniyial apparent excitingness, there is still a lot left.
I tried out the possibility of jetissoning a lot of that and just focussing on open source furniture etc. that can be made on the 3 axis shopbot, as it is surprising what you can make starting with just flat sheets of plywood. Tables, beds, cupboards, storage, chests. I designed and produced a few tables from plywood but the profit margins on the stock material were a problem we had encountered at OSE often too; the bed Andrew designed was going to cost more in materials than some crappy wobbly beds from amazon.com because we had to pay so much for even the cheapest lumber. It was going to be $30 or more just to make a reasonably sturdy table from half inch plywood. Clearly more of the supply chain needs to be captured. Obviously Ikea is not paying anywhere near that much for their input materials.
Add even the most minimal sanding and finishing at a modest hourly rate and the table wouldn't be a particularly good deal. Still it was good to take things from bits to atoms for once. I did not try to produce anything from opendesk.cc as the economics were far worse still for all their stuff. The only real success was a side project that I bashed out one evening, which is an invention that substantially improved my health at the time and this winter; a device that stores heat and water vapor upon exhalation and returns it to the inhaled air. Without it in previous winters I had to stop biking in winter. However, and this is a feature as far as I am concerned, you don't need a three d printer to make one.
I left after 2.5 months as things were going poorly on the project and my health. In Vancouver I had more success. The Vancouver Hack Space had social and political problems that rendered the mill there unuseable and the space less so, but I found truly good people and some embodied prior collaboration success in the form of some tools at the Makerlab. I made this my base of operations and succeeded at finding some collaborative inertia and using the 4th axis router system there, and did some solidworks drafting. I left regretfully only because of mental health concerns that looked like they would not remit in the environment, and because of the government barriers re relocating officially from one province to another finally catching up to me. The result is that I have hardly touched an x64 computer since never mind gaining skill with solidworks.
I took some pictures of some of the things I made there and put them on instagram: https://www.instagram.com/osmatterrearrangement/
I had to return to Ottawa, and suffered from some health concerns and injuries for some 2.5 months. Then I went to Montreal to help found a collective for about 1.5 months, got in touch with Echofab, offered to get the mini cnc lathes there working, but that's about it.
The collective was going very poorly, so I left to go to some gathering of old friends on sustainability, where I evangelized some re OSE. Then I went to an interesting collective in Quebec where I tried to contribute and improve my skills, learning some arduino C and algorithms. I held an OSE workshop there which went down well.
I at that point had decided I wanted to back to OSE and told my nearby friends as much. I was excited to see on the blog that the torch table was still recognized as important and fundamental, and a priority. But when I got back to Toronto it seemed clear to me that I would be doing the same thing I did two years ago, putting off getting a job and a life aside. Once was a good idea but I do not think my mental health could stand such deferrement again after two years of trying.
I must stay in the city and achieve a more normal income in the next months. Aside from that I continue to try to pursue open source hardware enterprise and system development. I have been keeping this in mind during my job search, looking for entry level cnc jobs hoping such skills gained would be useful later. However the capitalist system is heavily focussed on turning people into cogs whose skills end up javing very little application outside of a particular machine. All but two of I think the six machinists I have met are not machinists anymore or never were. I have also met people running machines that just jumped right into it and I have more experience than that already. There is some vauge hope I may go aid in he development and operation of electronic assembly robots in summer. However I frankly have started moving towards being a personal trainer as it appears more practical and available overall.
Also a technological development project though of a different variety, during the last four months I have had some success developing zero energy (zero calorie) foods, which have valuable properties for treatment of appetite disorders, and have given me some slight but otherwise unobtainable improvement. When I get a moments peace I hope to put this development effort to use by sharing the recipes as open source works on a collaboration inviting website, and perhaps selling the ingredients.
After the webinar I think I should write and share my status and situation on the subject, and with just a phone and data connection cannot find the document created during the workshop, so I will post here and perhaps copy it somewhere more useful later. Although the personal log system seems to work remarkably well.
For the 3D printer workshops initiative, the road seems relatively well cleared for me to hold such workshops. I am well prepared for the technical side re any troubleshooting, have Steamlabs as an excellent venue with much of the equipment needed as they often hold arduino workshops, some contacts in the maker community, and have hosted a few workshops of various kinds over the years. I think is important to use only the stable build of the Prusa, though, and not modify it at all, as that may complicate things more than it is worth.
It weighs upon me to be so flaky, like I had to leave Vancouver, but it is possible I will have to leave TO for employment, so we will see, though.
I think perhaps one of the lowest hanging fruits for aiding viral replicability, is time lapse videos. And indeed ideally a full length normal speed video of the whole workshop unedited with sound. These are a form of data that I think are poorly respected. Usually it is considered unpresentable or useless, but this is something I have seen again and again when it comes to information sharing in attempted open source enterprise: when you actually go to do it, you want this hard data.
That most of it may appear useless or repetitive is unimportant; such sections may have important things not apparent to would be helpful editors and can be skipped easily anyway by viewers. The main substantive concern is privacy for participants but I honestly think that would be minimal.
It is not as good as well distilled and evolved information, but it is orders of magnitude easier to produce. It is an approach different to the oft seen and ultimately poor approach of "do it then open source it after". The raw stuff, is the source code, not the marketing material, not the presentable edited video. Of course commenting is good. What does not ever work is some kind of higher level summary of the source code. Plus such summaries take extra work to produce when we have our hands full. Raw video is excellent bang for the buck, and as someone trying to hold such a workshop, it is one of the most valuable things that I think could be shared by previous workshop holders.
I will post separately re an update of what I have been up to personally, as it has been to a large degree pursuit of open source enterprise and system development.
Wow, did I not do any logging sing june 3? Hm. I have been using dropbox instead for my documentation. Honestly, the wiki is just too kludgy, slow, has too many limitations, and requires an internet connection, which we often don't have for one reason or another.
Dropbox also sucks, though, I have noticed it fails to sync, and this has been a real barrier in a number of cases.
I am there fore trying to switch to google drive.
The publicly accessible link to the google drive folder is here: https://drive.google.com/folderview?id=0B12wZ5LxhSYndWNpaVViMU5tNkU&usp=sharing
Google drive Sync sounds a lot like dropbox. We will see.
The torch table is now moving in a way that is quite accurate, less than 0.18% linear repeatable error, which can be reduced further by changing a file. Basically it will be more accurate than our best measuring instrument, the calipers.
It is highly repeatable, not missing steps or slipping. After 5 times around the table it was within 0.2 mm.
I got the z axis under control of the arduino yesterday, so it can be controlled on a relative position basis in gcode. It also has a manual control. The z axis is important even in this open loop fashion to aid in piercing, and to ensure the tip does not snag on slag piles.
I think it will be relatively easy for someone who knows C to implement the feedback loop for the arc voltage and z height from this point. I am learning C now but hope to be gone real soon, so it won't be me.
The toolchain for the files is working except for circular holes, which should be a minor problem, to the degree it might take about 3 person hours to process the Liberator's .dxf files fully, to account for kerf, lead in and lead out, and the pathing order is also ok. I have tested it on ld-3, one of the brick press parts, and may do more later today.
Processing for piercing is working, we can get reliable piercing. It requires that you open the gcode and modify it manually, but the process is reliable and really quite fast, it only takes a couple minutes of person time. Gabriel has modified the lasaurapp to insert the pierce gcode in the right places, but the gcode is hard codded in, so if we need to change the timings that is a bit of a problem, although we can ssh into the beagleboneblack and change the source.
Cut quality is still poor with oxy. Ultimately oxy is a poor technology for the application, it has a wide kerf, is messy and more hazardous, has very poor accuracy due in large part to variable kerf, and produces vast amounts of heat compared with plasma. These latter 2 in particular are real deal breakers re a serious fabrication machine. The former is nearly self explanatory, although you can make some things with poor accuracy, it is one of the fundamental enabling characteristics of manufacturing methods, and we really do need more than the 3 mm over 25 cm oxy is expected to get. The latter means that accuracy is even more poor for holes, because the heat causes excessive melting.
We should not waste any more time on oxy. We have to stop this feature creep stuff and actually decide what we want, then go do it.
Although the reader may hope to be able to get it working better by getting the parameters worked out, the fundamental limits I describe are widely accepted in the industry and will not be overcome. I am doubtful that it might be useful to someone somewhere because it supposedly has a lower operating cost, because those operating cost figures assume cheaper gasses than such people will probably have, and they are perfectly capable of doing this figuring out themselves. Even if we get oxy working as well as it can with our poor quality equipment, they be almost back at square one as they will have different equipment anyway, and we cannot describe many of the settings quantitatively, which means they will not be able to use it anyway practically. We will not be adding to the knowledge base by further work on oxy. And in any case, we should be spending our time on the *most* important things, not just random things that are kind of maybe useful.
The only reason to continue with oxy is to use it ourselves, and it makes a poor tool anyway, as mentioned.
The shielding is the main thing that needs to be done to use plasma. It is no minor task, but manageable. I would have been working on it, but I can't even turn on the plasma cutter without an air dryer, and Marcin is not willing to get anything other than a very particular $400 refrigeration based unit which is mail-order only from harbor freight.
So progress has essentially come to a halt.
Gabriel and Aidan and to some degree maybe Stephen and whoever they bring on board will take this from here.
My task is mainly to document and backup things before I go, make sure they are accessible. Fortunately almost 80% of the stuff I have done here is also useable for plasma.
As a final note, I need to talk to Marcin on what the reason for insisting on Oxy was, as it has been a serious mistake that has prevented progress, but is compounded by the failure to move on getting the plasma ready in time to keep up with development.
A machine that cuts metal parts to good tolerance, with a good cost to performance ratio, that is open source and passably well designed and built. That is what we need, and nothing else will do.
Here is the link to the dropbox folder: https://www.dropbox.com/l/UqQL7SkTXMTFwY38YEfrdo? . You should be able to download anything in the dropbox from that page, or all from a .zip file. This approach of a dropbox, plus a publicly accessible link, allows us to practically upload and sync files from our computers, and to a lesser degree phones, which is good for videos and pictures, while also ensuring anyone in the world has ready access to them for a relatively long way in to the future. Hopefully a couple years, I have no idea what dropbox's policy is. Within a year the info will be mostly obsolete, though, as the next version of the table should have been implemented, and the results published.
Ultimately it should all be on the wiki, for the longer term (OSE implements periodic wiki backups), but the wiki has a 50 meg file limit, is slow as molasses, limitations on file types, needs things need to be manually updated, and is otherwise impractical, in the end, to use.
I uploaded the backup of the dropbox folder today, to the wiki, for the torch table. The next step is to import things into the dev board. I haven't been doing that as I have been the only one working on it. The file should be named Dropbox_folder_for_torch_table_and_router_v2_may24_onwards.7z on the wiki when the upload completes.
Today I was looking into doing a repeatability test with the table, to see if we can detect any/when the steppers skip, basically, and what they skew of the table is. The thing is that it looks like it will be a lot easier to do with a working z axis. And that shouldn't take too long to get working. The main problem is that I don't know much C. So I spent about 5 hours today learning some more, and then I will need to learn about the arduino environment specifically, then it's a very minor matter to get the arduino to accept input from a switch and output to the drivers.
I have made some more progress on the torch table. The main thing right now is piercing. The oxy torch needs to stop briefly before it proceeds with each cut line, to allow time for the operator+torch to turn on the o2 lance and proceed with cutting. I was experimenting with the torch, and I think it may not be able to pierce while the o2 lance is on, but further experimenting with the gas flow rates is called for.
I have tried several approaches to get the thing to pierce: -Use a line segment of different color at the beginning of each cut, and tell it to cut very slowly there. Doesn't seem promising as the system cuts all non-black lines first, instead of going in the order in which they appear on the cut diagram. I also had to figure out where the lasersaur starts the cut lines in order to decide where to put the red lines, and need to add this discovered logic to the documentation (it's in my notebook presently). -Use the manual pause button in the lasaurapp interface. Impractical because lasersaur takes a few seconds to actually pause after you tell it too. Oi. -export the processed .svg as g-code, then modify the g-code and feed it back to the driveboard. A promising approach, except after modifying the g-code, which took a good part of my day, I discovered there is no way of actually getting the driveboard to simply read a g-code file. The g-code is part of the internal processing; the svg is converted to a g-code file, then the g-code is fed to the atmega from the bbb, which interprets it as it is streamed in. K. But there are no provisions for circumventing the svg-processing part and using g-code directly. This is a fundamental omission, I think, as during testing and hacking, you always want to be able to test a system piecewise, and the flexibility loss you suffer by omitting piecewise use of the machine is a real problem. As seen here; we want to do something ever so slightly different than the designers had in mind, and the system is relatively hard to shimmy into this role.
An important lesson to remember when designing future systems. Although if I knew more Python and linux it would still be manageable, modifying the source code is still a level of bother that is a lot greater than just having provisions to import and execute a g-code file, which the system almost inherently has. I may pull something off in this vein yet.
Remaining to be investigated are: -Get the system to read a g-code file that I put on the bbb via ssh. -use the stop mode to pause/resume the cut manually. Again, timing is a problem.
In other news, I repaired the interned connection in the shop by running a cable down there, after being stymied by not having the passwords for the router and isp, and therefore not being able to do it using a wireless connection.
I did not order the pcb or parts for the capacitance sensor for the torch table yet, because I found the arduino is capable of implementing basic capacitance sense capability with almost no additional hardware, and frankly I was going to but got dragged away to work on the microhouse, and manage personal stuff which takes a lot longer than it ought to under these conditions. However the range and signal to noise ratio might not be as good as the separate sensor. So we might have a problem there, or we might have a particularly elegant solution in the end. It's a small gamble.
A lot of my time has been consumed with helping with the microhouse, which is not so micro any more, having it's size tripled, plus a big deck added. We spent all weekend on it, and some time on Monday and yesterday.
I also really need to put a lot of this down to figure out my travel itinerary, and get tickets etc. Getting out of here is a real problem, as getting a ride out to a normal city is difficult.
I still don't know when I will leave, it depends on when can get a ride, and what my investigation of low cost travel options reveals. There are also still a few parcels that have things I need to take with me which are taking forever to get here. My time here has been epic and life-changing, as I expected, but I frankly started wanting to leave some time ago. 2.5 months is more than enough time here.
So ultimately the priorities right now are the torch table and the documentation thereof, and getting out of here.
Some more progress on the torch table has been made, but I have been able to spend quite little time on it, overall. Yesterday I had to go in to town, help plant the garden, etc. and today we are building the microhouse porch.
Bittorrent sync has problems regarding edit conflicts - it sometimes wipes out recent changes made to a file, so we are switching back to dropbox. To see the files in the dropbox, this link should work for anyone who wants to see: https://www.dropbox.com/sh/i28hx6enwcl9u8b/AAC7BtGtDiiY-BALI6rrYp-6a
There is nothing there at present. I need to copy the stuff from the bittorrent sync folder, but currently the computer that has the stuff on it is somewhere else, apparently not connected to the internet, because it is not syncing with me. I thought Kyle also had a copy of the folder, and this other computer did. Apparently not. So now I am stuck without the files I had been working from.
I have made some progress on the torch table; it now operates over the whole table. I spent quite a bit of time on attaching the limit switches. Got some progress on piercing logic. Was not able to find any metric bolts or allan keys in the hardware store, so we still cannot attach the z axis unit.
The plan for the z axis is to order the pcb and parts for the sensor, assemble it with one of the arduinos here, and one of the stepper drivers (we have a spare), and hopefully Skylar can program the arduino, otherwise I can hobble to do so.
The gases will be constantly on, for now. To terminate a cut you just move the torch head extra fast, which the controller will do automatically for us anyway, since it is just transiting from the end of one cut to the beginning of another. This might end up leaving a slight tail or notch in the piece, though. A way to control the O2 valve would solve this, but I don't know if the operator could control it, as you'd have to time things just right, such that you turn it off just as it finishes the cut, which I don't think you'd be able to do very well anyway.
For now, this will be a pretty crude cutting system, but it will make a solid platform for use as a plasma torch table later.
I have made much less progress on this than I could have, though, because I am constantly pulled away by a million other things here. Not distractions, just things like helping to take a fence down, trying to find a way to get into the city to do essential personal things, get some new shoes, stay clean and fed, stay in shape, find a place to work without constant disruption. All minor things here are much more time consuming, and this decimates the time I have to spend on the actual reason for being here rather than somewhere else.
I think this is a problem that can totally be solved relatively easily, but it requires changes in the culture and infrastructure, something too long-term for me to practically work on, given I am leaving soon.
Got the torch table x and y finally mostly working today!
Next, I need to change the working area of the system, attach the limit switches a bit better, speed up the steppers. try an dvg file down the tool chain, probably affix and hang the cables a bit better, then we need the z axis controller. Then some sort of logic for piercing looks like the next hardest issue.
It occurs to me that almost all the time spent thus far was on saving money, and not very much, either. Also, that much of that time, in turn, was spent overcoming limitations like a lack of electrical tape, etc. which are very inexpensive things. Clearly there is much room for increasing the rate of development work here. We could have been where we are now two weeks ago.
A video of the table moving can be found in the torch table stuff folder. It is currently stuck in a corner of the table, because I have the stepper motors turned down to 1/16 steps per pulse, and because the driveboard etc. software constrains the working area to what it things is the size of the lasersaur bed.
I have added some info to the documentation files in the torch table folder, as well. I think the approach of using more general pupose tools is turning out to be much better and more powerful already. Just basic things like inserting a table, making sure you don't loose your work, etc. is easy, by applying tools better suited for the task. If it takes 4 times as long to do things with a given tool set, then you will only get 1/4x as much done in any finite time.
I have been making some progress on the cnc torch and router table. Here is some documentation on wiring and parts I took an hour to make today, for anyone doing something with it in the future.
There is a copy of everything in the following bittorrent sync folder, too, view only key:BXKD32CVOIQBVYLHYXNPCJLAKPQWWBMYU.
It takes forever to upload anything, it should be named Torch_table_stuff_summer_2014.zip so you should be able to find it that way. I may or may not have the chance/remember to put the file url here later.
One thing that looks like it might hold us up is the z controller, I haven't heard anything from skylar about it at all.
Things have been going well on the torch table. I have been seriously under the weather, but hopefully will be doing better soon. These are hard conditions to operate under, even after I have had time to set up some.
Tyler has got the mechanical drawing of the existing table done. We have made a lot of progress in picking apart and understanding the driveboard from the lasersaur, and in overcoming the usual huge number of minor barriers, and I made a wiring diagram for us to follow:https://docs.google.com/drawings/d/1tT_dwtOEHOU2znsg3k915PcJW_YMGrmoXdlzTLb3FdY/edit . We have an ongoing problem with the conflict between the advantages of google docs for sharing vs. the severe performance limitations it has. I think a good approach is to use google drive, or better yet, bittorrent sync (although version control and edit conflicts are a problem for that). Then we can use practical, general purpose tools, but still share the files. Granted you can't edit them in real time like this, but for a lot of stuff you don't have more than one person editing it anyway at one time. However the documents can still be readily shared, including withe world.
Google docs is good for performance requirement documents, etc. but it is too slow and limited for a lot of stuff.
There have been a lot of false starts, unfortunately, but basically we are focussing on getting the xy robot working from .svg files, and then tacking on a separate z axis controller and a bunch of manual intervention to cut to a useable oxy-fuel torch table of sorts. You will have to stand there and supervise it, ignite it etc. though. There are no gas control valves, ignition, pierce detection or flame-out detection, etc.
This is largely because plasma is a more accurate technology anyway, so what we really want is plasma. And integrating the z axis with the driveboard controller is also an important thing to do, so it can be used for milling, to reduce manual intervention, etc.
Ok, so we made some progress on the torch table today. Tyler is drawing the existing table in Solidworks, Devin is up to some stuff, and I was assembling a list of parts, and things we need to do, with a focus on the things that we are most sure will need to be done sooner or later. Doing these things first helps to ensure that the time spent is not going to be thrown down the drain later; it provides a sort of known-good way of getting at least a bit of traction at the beginning of the project, which will be built upon.
Now I need to sort of save my work by doing something someone else can build upon, if I were to disappear tomorrow, for example, and also to communicate with the rest of the team what I have learned and whatever news I have.
Things to do: There is a document on the kitchen table we are using to add todo things and things to acquire as we think of them. It would be nice if we could have both a high level of accessibility of the kitchen table document, while also the benefits of a google doc, but ultimately if we have to pick one it seems like the paper is still the better choice. Google docs just are too slow and take too long to access. I noticed that they are usually forgotten and remain unused, with the result that we fail to keep track of things or use documentation altogether. Having a shopping list on paper that we actually use rather than one on google docs that is not accessible, forgotten about, or for whatever reason not actually used, is not as good, ultimately.
Ok, so after having a look at the lasersaur manual,
Very basic system diagram thus far, it is far from detiailed or complete. I don't know yet if we need a 24 v psu.. The stepper motors are actually 4.5 volts, 2.5 amps. It depends if anything on the driveboard or the motor controllers actually needs 24 v, or if it is only used for other things, i.e. switched by but not used by the drive board.
The current set resistors should be about 26.1 kohm, or something close to that, according to the equation in the lasersaur manual.
We may be able to use the same driver to drive both x axis motors. otherwise, it looks like just parallelizing 2 drivers should be fine.
I looked up the motors, and they are probably compatible with the lasersaur software. The steps per mm of travel will need to be adjusted, hopefully that won't be too hard.
a toolchain example is .svg file (vector graphic from e.g. inkscape) -> bbb, then streamed to the atmega on the driveboard, which does the gcode interpreting, and handling the state of the robot.
Looks like we can defeat all interlocks and the door switches easily enough. The main safety interlock can double as a master power switch, apparently.
There is a spare stepper motor in this box of electronics, here. We can probably use it fine, although it occurs to me generally that we have a problem at OSE with a poor stock of standard useful things, which really slows things down and holds things up, sometimes. The just in time thing doesn't work well here. It's another story, but it would probably not be a bad idea to have a couple extra arduino, stepper, screwdriver bits, 5/8 inch bolts etc. just handy as a matter of course.
For z height control ultrasonic looks promising still, but making our own ultrasonic sensor is a small project of it's own. An arduino avr or pic, stepper controller, we can use the power supply from the lasersaur (5v) and the same stepper and gecko drive looks like the most promising approach. We could use a different, smaller and cheaper stepper driver, but we have a spare or even two of the geckos, so I don't think it makes sense for now to try to use a different driver. My main standing concern is the beam path of the ultrasonic sensor, and the supposed 1mm resolution of the sensor. The positional envelope we want to stay in is +/-1/16', only 1.5 mm, so you want higher resolution than that, really. The commercial ultrasonic based controllers don't do anything about the beam, I noticed. They just measure slightly to the side of the torch head.
link to edit and view the document:t
Standing questions, which I hope to improve on, tomorrow, include: -brushes for the flats to keep out debris? This could end up being important to avoid jams, although easy to improvise it may need to be done wone way or another. -power supply situation? -how easy is it to set the steps per inch -use a compensation table with the lasaurapp? -we need to get on with ordering the stuff like limit switches etc. Assemble part numbers, consult the others and combine their lists, do some shopping and email MArcin a list of stuff, distinguish between what will be needed sooner or later for progress, and what is nice but optional, or based on the current proposed way forwards. -acceleration and deceleration rate of the steppers needs to be addressed to avoid missed steps -general mechanical fit of the system needs to be investigated, and what needs doing identified
I read a bit on existing models and projects, and found the Torchmate growth series table, which is a commercial version that is a lot like what we are trying to do here, apparently. It is a table with replaceable toolheads that can do milling and routing, too. Not accurate milling the way a bridgeport, for instance, can, but it could do the rough machining and bulk removal which would really help a ton.
The CEB press was completed in the nick of time, and Scott, the buyer, will test it further before leaving for far off places with it. Retail price is $9k, plus you need a power cube to run it, 2 for 12 bpm. The accuracy of the blocks coming from the machine is excellent, we got it to within 1/16 on all dimensions for a couple of test bricks.
Interestingly, the accuracy on the width-wise dimensions, which stems from the width of a plasma cut plate of steel, is within 1/64th, which comes from the accuracy that the plasma cutter has. So basically the accuracy of the cutter was successfully leveraged there. It seems clear to me that it can be elsewhere, too, and that this will cut the labor requirements of manufacture a great deal.
The Microhouse structure and roof is also done, with only some more interior stuff like putting the the very last of the laminate flooring, ladder for the loft etc. in left. Now the main focus is the torch table.
I haven't actually bought a ticket yet to anywhere, for better or for worse. The thing is that it changes so much here depending on who else is present, what the current focus is etc. that I might as well almost be in a different place. I had thought that I had really had my fill and it was time to go soon. But I want to see how things go in the next week, as I am tempted to stay further, maybe another month even, if I find I am contributing a good amount and also learning a lot.
But today I really, really need to get in to Cameron to do some personal things. One of the major problems I have had here is the accumulation of personal things that I cannot get done because of the lack of access to the things present in a city, and secondly because of the increased time it takes to do normal routine things like cooking and cleaning that inhabiting in a group environment of this nature (which is not like some other group environments) entails.
I hope to get together with the guys and get some performance based design criteria down for the torch table, and place some orders for parts we know we will need, etc. We also need to ask Marcin about the history of the table etc.
Today was the last day of the Microhouse (structure) build. Tyler, Kyle, Devin and myself went down to the workshop to have a look and start the torch table development. We started a folder system using bittorrent sync to share files, hoping it will be more reliable than dropbox. We are going about making the design requirements, but still need to bring Marcin on board so things are congruent with the larger scale plans. From the look of things like now from what we know of the large scale movement of the enterprise, it doesn't look like it makes any sense to use the oxy acetylene torch. Instead, we should skip straight to working on the plasma torch head approach. Also, it may make a lot more sense to get this existing table working first, before we try to reach out to the advantages that gridbeam might bring. More of a bird in the hand worth two in the bush sort of thing. Aligning this table looks a lot easier than building a new one.
We had a great day yesterday; the ceb press was breaking down in a variety of interesting ways, difficult enough to figure out to be interesting and educational, but not enough to stop us. We have 5 different types of easily spotted jams, plus some software bugs which are harder to see happening unless you already know the software. Also it is not clear what the origin of some of the jams is; it may be that they start as software bugs but then become something else, which we would see as mechanical clogging, for instance. They can all be smoothed out relatively easily when we build the next press, I think. We stayed up late, I stayed till about 10 pm, pressing bricks. We figure we need about 3000 to 4000 bricks in total.
The drawer cloggage, bridging, bashing and jamming against the drawer roller,
Other bugs we spotted include the misalignment of the rear piston and the whole structure back there, the design of the top of the hopper area, etc.
The importance of an actual pulverizer became clear; a substantial fraction of the soil dumped on the grate would normally not go through, and we would brush it off or manually pulverize it, joking about how you might someday build a machine to do this (and some people of course chiming in with the old "zat would never work!" refrain while unable or unwilling to say even vaguely why they thought as much. It's particularly funny when people say that about something that you already are confident works pretty well and is widely used).
Hm, I've been pretty bad about filling in my log recently. Largely because there wasn't much to tell. We've been getting the place, mostly the hab lab, ready for the influx of 24 people for the microhouse workshop that starts on Friday. Tonight will be the inaugural dinner.
We planned for and bought food, did some more trim and painting, assembled bunk beds and moved things around.
After the microhouse workshop there is another one, and then we are going to be advancing the torch table project until mid May.
We assembled bunk beds, put in carpet, moved things around, etc. Kyle and Chris have arrived, so with James Slate it is more exciting here. We still have to do something about the food for 20 people, and dinner for 22.
Pretty good day today. We got what we expected to get done re preparing the infrastructure for the upcoming workshop. I volunteered to use the paint sprayer, a dirty task, but I was happy to do it as I am a fan of the use of labor multiplying machines like this. I discovered the spray nozzle was very poor, though. It, I gather, is supposed to put paint in a line. But it applies a great excess to the end of the line. The spray nozzle thing is fixed and non-adjustable. I don't know how they managed to do things this way, as it leaves a line on the wall of extra paint. I tried to compensate by moving the nozzle around some extra.
I also ran in to the problem of the paint being too translucent to cover the wall beneath sufficiently in one coat. The main limiting factor in how much you can put on in one coat is the formation of drips of paint. Particularly a problem for a ceiling, of course, especially one with areas that encourage flow to a point, as this one does. I was conservative for room 1, but on the next room I did I, in trying to find the optimum, put too much on, and got dripping. Nasty state of affairs. I reduced thickness for the next room, but not quite enough, still it wasn't nearly as bad. But I knew a second coat was almost out of the question, certainly with the sprayer, as the stuff needs to be put back in the room. We could cover things with plastic, though. Multiple coats of paint has always struck me as an absolutely abominable divider of labor power, though. I mean it nearly doubles the amount of work to two coats, and sometimes 3 is needed. Can we seriously not just invent a paint that has a better opacity to viscosity to drying rate performance, here? What if we add a thixotropic agent to the paint - thixotropic substances are liquid when they are under relatively high shear forces, but become solids when they are left mostly alone. So the paint goes on as a liquid, but it becomes a weak solid - sort of a gel - after application. That would certainly allow an increase in layer thickness. Or we could load more pigment particulates per liter of paint to increase opacity. It needs to be nearly double, though. What if we use flakes instead of spherical particles? If they aligned in the right way maybe that would improve the pigment per unit volume to opacity ratio. Especially if we could get them to align after application to the surface somehow. Make them electrets, or embed magnetic particles within them in the right orientation... or there may be some approach using surface physics or something. A paint that saved you from needing a second coat would sell like hotcakes to painters the world over.
Hey, where did my entries until april 12 go?? Well I ain't re-writing them now. Basically what has happened since the ninth is that we poured the foundation for the microhouse, planned what we need to do to be ready for the April 18 workshop, and got on with some of the ordering, and shopping. Dorkmo arrived yesterday, so that is great to have another person here helping. James slade is supposed to come down some time soon, too. There is a lot of painting, cleaning, fixing up etc. to do to ensure we are ready to house and feed 20 workshop participants for 5 days, then about as many for 3 days more after, for the CEB workshop.
We encountered an interesting situation in which we designed a bunk bed to build, needing 4 in the near term and 6 for the longer term, but found that the BOM would cost $125 at Menards, and a bunk bed on amazon.com would be $150 after shipping. It uses a metal frame, instead of wood. I think we may have been able to slash the price of ours by using less wood, or maybe we would find that metal pipes ended up cheaper. I think it is an important case study that should not be forgotten. We had a similar situation with the paint sprayer. We should be able to make stuff like this by now. When we look at the things we want made, we are not able to make them. Also, the things we want done, our equipment cannot do. Why? And why should we focus on ourselves, though? Well, if we look at the example of FOSS, I have read that the main reason it came into existence and was built to it's current point is that software engineers like it better, it making their jobs and lives a lot better and easier. The benefits that accrue to non-software engineers are mainly a form of collateral benefit. But we should expect such benefit.
Hopefully today we will finish up and power up the mixer. I have collected a lot of notes that need to be input and acted on, for the documentation of the next version of the source of the mixer. Then we need to do some stuff on the CEB press.
If I could only say one more thing here, it would be that I think it is also very important that OSE specify a set of standard fabrication tools, input materials such as stock, welding wire, etc. and training states for the people in it. By tools I include the building, safety garmets, drill bits, everything. The tools can be used for the training, I am currently quite persuaded that as long as safety is paramount you can learn over time how to do things effectively, with access to the internet and, ideally, a teacher here and there.
Things are really picking up, the weather is more hospitable, I think the next month and a bit are going to be very, very interesting and productive :).
I have been trying for a long time to learn both python and Inventor, and keep wishing I had learned it already. It's the nature of capital, of course, much like how many people who rent homes take a very long time to make a down payment on a house; accumulating capital is often pushed out because of the everyday press of other things, preventing long term progress.
We are very much expecting to finish the pulverizer build tomorrow. There is some trouble with the shaft, and the tooth bars just need to be bolted on, and that's it. The bucket also serves as a front end loader. Then we need to integrate the output of the build process into the product development, as enabling others to efficiently produce another unit is central to the enterprise. We have the time lapse as a record of the process, and also the time logs we have been keeping. I really wish, and I said this at the beginning, that we had a camera or several, documenting the whole process. The camera misses a lot. Even just now, I went in and asked if there were any shots of the shaft, so I could look at it instead of taking it all off to look at it. No. The shaft has been kicking around for many days of build time, and there is still no clear picture of it anywhere?
The real test, IMO, will come some time after the build process, to see how well we integrate the experience from the completely untested alpha into the tested once then overhauled beta version of the source code for the mixer, as well, of course how well the mixer works, and compares with units you can obtain in other ways.
Pretty good build day. I replaced the propane tank regulator this day, I think. I have been forgetting about my log to some degree, and will try to fill it in at the end of each day. When we are not actually at our computers the value of filling it in during the day is really too low.
Hoping to do just a little more today on the build, but I have a backlog of domestic work too, cooking, laundry, all that sort of thing that really adds up.
BTW for anyone who is not already aware, my plan is to stay till mid may, from the beginning of May the plan is to work on the torch table. But the soil mixer must be done for the upcoming workshops. I still need to install ubuntu on this computer, too, and there are some other things.
The hose for the hot water heater leaks propane, so that is a problem that we need to try to fix real soon, as it must be 2 or more pounds a day lost. We can turn it off except for showers, that will slash it severely. The welder was taking up a lot of time by malfuctioning, a problem that seems almost trivial to design around. The wire feed was incapable of overcoming the resistance to the motion through the liner and tip, is all. The liner is a ferromagnetic steel, so it collects filings etc. I think this was an important factor, as I disassembled and checked the resistance to motion on a range of counts, and it was only the liner that was unusually high, although still not that bad, it is kind of hard to push a wire. Sheer variety of things that can go wrong yet could be easily designed away seems to highlight to me how the tools around us are designed with criteria very different than what the user's desires are, which the users often blame on incompetence of the manufacturers, or just assume that it's harder than it is to design things to meet our needs better, even in the more obvious ways. I think the truth is that this sort of thing is mainly caused by the design criteria being so different than what the role and manner in which we want to use the tool.
In a way this is really important, because although the equipment designed at FeF may not seem super sophisticated, merely the congruence of what we are designing for and what the tool will be used for can get us enormous gains in the cost to performance ratio of the tool. The question is often asked of open source projects: how do you possibly expect to improve on the cost to performance ratio of the stuff from closed source, conventional manufacturers, when they have so much capital and experience at their disposal? I think this aformentioned reality is part of the answer. Ultimately the best argument is the success of prior open source projects, though, at doing so.
update: Good day indeed, today. The freedom of working on the weekend was actually very helpful, just being more free to use my time as I found suitable. Although I started later and stopped for an hour or two there, I got about as much and maybe more done, I think, and also got some domestic stuff done, too. I put the holes in the tine modules, then marshalled the nuts and bolts and got some kinks worked out there, got the welder #4 mostly working again by replacing the torch liner, finished welding the tooth bars, ground them, ground the angle plate in the bucket, tacked it in place, got the bearings fitting on the shaft and in place ok, got some progress on the motor mount, fixed a missing hole problem, and investigated some potential build and development process improvements.
It's the weekend! Yay! I want to go back to the workshop and do some more stuff. We try to take the weekend off, which I think is a very good idea, and enlightened. Enthusiasm and the energy that a project team has is not in opposition to making good decisions regarding a division of one's time.
The thing is that this is the sort of thing that I would love to be able to do on the weekend at home.
The wiki just destroyed all my edits, and my retrospective edits for the past 2 days, which included the build process. The edit window indicated "view source", and had the entries in the window, indicating they were part of the source of the page. Still, I highlighted the area and pressed control+c for basic disaster mitigation, so I would have a copy on the clipboard in case. Then I pressed f5 to reload the page to get an actual edit window, and while it was loading, right clicked, to be extra sure it copied it by clicking on the right click context menu. The copy etc. options were greyed out. The page reloaded and everything is gone, including in the history. I don't have time to add it back now, as it is 10:30 and we need to proceed with the build process.
I input my time sheet and updated my log for the last 1.5 hours. Time to go! File:TIMESHEET INPUT.ods
We have finished the dozuki guide today, I just went over it to check for bugs etc. now.
In the afternoon I finished limewashing 2 more hotel rooms, we will see tomorrow if they need another coat.
We worked on the fabrication instructions today. There was some backtracking due to some co-ordination issues. The illustrations for the dozuki are also being made. My completed fab instructions need to be integrated with an older, incomplete version for which we have some illustrations, otherwise the illustrations would need to be re-done.
I did some limewashing got about half a room done, much faster than usual. The trick I developed was to splash limewash on, then spread it into an even layer with the paintbrush, because otherwise the rate of material transfer from bucket to wall is a major bottleneck. Not much limewash comes with the bucket. A super soaker type squirt gun would work very well here, greatly increasing the rate of limewashing, with the current mix. I doubt it would clog, and if it did you could unclog it periodically and still end up with a large improvement factor. Ultimately this shows the importance of actually going ahead and trying things out on the ground. There is a lot of gold to be found in both armchair thinking and research and discussion, but I think that what is often left off the table is the sort of exploration that just gave me a 3x speed improvement; it includes a certain skepticism regarding the essentiality of sticking with the usual approaches. Our situation is never exactly the same as what others face, and that opens up room for doing things differently. A sprayer may seem attractive, for instance, but it wouldn't help much at the perimeter of the walls anyway, because it would get a lot on the roof. Secondly, we only have a 1.5 rooms left for now, so investing the time and effort in inventing a sprayer makes little sense.
source note: to reduce cracking, spraying the wall enough that it darkens, then allowing it to sit for about 5 minutes helps a bit, to reduce cracks by maybe 35%. It is important with this method to start at the top and work down, or you get splashes and drip marks. Set up the ladder, fill the bucket only a bit full so it is comfortable to hold, paint near the ceiling carefully. Then you can start splashing. Holding the rim of the bucket close to the wall, use the brush to sort of paddle out limewash, to get it on the wall. Proceed in a line, and stack the lines as far as you can reach. Get an idea of how much a splash covers after spreading after a few full repetitions of the splash and spread, and spread your splashes accordingly. Then spread with horizontal brush strokes, followed by long vertical stokes for the desired texture. Any drips or splash marks should be gone over with the brush within less than a minute or two, or they may partially dry and leave an undesireable mark, although going over repeatedly with the brush can help reduce these. The brush abrades the wall and gets mud in the limewash if you scrub hard, so try to press lightly. We want a layer as thick as we can get without dripping (which with this mix, 1:1:0.2 by volume of lime, water and salt, does not appear to increase cracking over a thin layer), and to reduce the mud contamination in hopes on not needing to do a second coat, which would greatly increase workload. If you do need a second coat, try to use uncontaminated limewash.
disassembled tine module, for production of a 3d exploded view of the tine module SoilPulverizertines_disassembled.skp
A load of gravel came first thing today, I took a humble video and put it on youtube. (currently sideways, I'm hoping youtube will have the capacity to rotate it after upload, http://youtu.be/0BfR3y66ATk) Order from Sweiger shop for the steel placed ( based on what the BOM, accessible through the dev board through dozuki, specifies, minus what we had in stock, plus some extra while an order is being placed) . Got the hydraulic diagram up, too Soil_Mixer_-_Overall_Machine_-_Hydraulic_Diagrams. The only thing there really is the cushion valves, and it would be nice to know more about what lies within the lifetrac cabin, but this sort of system diagram, which specifies inputs and outputs, without bothering with the actual details of what is in the box, has many advantages, too.
In the afternoon we spread some gravel, I finished the interior painting and put the carpet and beds back, and put some carpet from the silo to put in the hab lab while the original tiles are awaiting washing. Whoever's idea it was to use carpet tiles deserves a thank you, as they have proven more suited to the hab lab than normal carpet many times. Some weak adhesive on the back of them would be good, though, as they stick up sometimes and you can trip on them.
Starting today at 9. The main thing is the bom and then the order from swieger (Sweiger?) shop, then the hydraulics diagram with google Draw. Andrew is doing the fabrication procedure.
google docs has a bit of a problem: when I go to the workshop, I can't edit the spreadsheet anymore, because google docs can't handle not having an internet connection. I cut and pasted the BOM into Calc for now, then will paste it back into the bom.
got he hydraulics diagram done, will put them up tomorrow
I worked on the bill of materials, then went and searched the workshop to determine what we did and did not have in stock here, and so what we need to order. I'm getting a bit more familiar with the shop, but I still missed some stuff that we had in stock, thinking we did not.... anyway, it was good news in the end, we have all the hydraulic parts, and many of the others, too.
Then we did some painting in the rooms, including our bedrooms, so we slept outside them tonight, because of the fumes. There is still more painting tomorrow, so it will be the same tonight. In the evening Marcin stopped by, and we did some review of the bom and what we had in stock, and what parts/materials need to be ordered for the mixer build. I'll review the bom again, knowing where the stock is kept this time, produce a derivative file that is less verbose as a list of stuff to submit as an order from swieger shop, then email Marcin the list, make any changes needed, and fax it to swieger shop using the faxzero.com service (why swieger doesn't accept orders by email I don't know).
Friday, we spent today on the mixer changing the DXF files in an attempt to greatly reduce the amount we'd have to pay the fab shop to cut the steel for us. By keeping the big plates and many of the strip pieces aside to cut ourselves from plate with a torch by hand or stock strops with the ironworker, we slashed the bill from $2000 to $750 or so. I made some progress on installing Ubuntu, too, and installing, freecad, openscad and librecad, mainly a matter of overcoming error after error.
We spent the afternoon limewashing and preparing a room for limewashing, then we started a narrated historical tour of FeF, which I'm hoping we can do more of on Monday.
We spent this morning, mostly, fixing a couple holes and propagating the changes through the dependent files for the mixer. We got 1.5 rooms of limewashing done, in the afternoon. I got eh ubuntu live usb working, after finding the computers would not boot from a cd, and I made some progress installing it. I was reminded that Linux installation always seems to have a problem with partitions; the installer is incapable of resizing partitions, and gparted is fritzy, so I boot back into windows to free up space on a partition, then will reboot to ubuntu to try again. Minor progress on the bom and also the fab drawings.
Starting in the design stuff. First, the review of the 3d model, then we can export to dxf for and, check the scaling, combine the dxfs into a single file in librecad so they can be cut from a 4 by 8 sheet. Andrew is presently increasing the polygon count, too, so those circles and stuff are a little more circular, as I noticed some of the holes on the CEB press are clearly polygons after cutting, although I suppose this is small fries, it is something that it would be nice to address for the future, to improve the cost to performance ratio in some cases. Actual round holes are a useful thing.
Export from sketchup to to-scale DXF file is now complete!
The orthographic views of the weldments are exported too, the next thing for them is to dimension them. That shouldn't take too long in Inventor. LibreCAD doesn't snap to the lines in the drawing, so the dimensions come out with long strings after the decimal point based on how closely you could get your mouse cursor, and you can't edit the text after, it just becomes lines.
But now we move to infrastructure stuff, mainly whitewashing that hotel room.
Design stuff in the morning was mostly spent getting Inventor working, looking into importing Sketchup files, reading about Geometric dimensioning and tolerancing, and investigating libreCAD for use in dimensioning the outputs of the sketchup to DXF converter tool. The afternoon infrastructure stuff entailed removing the rest of the adhesive from the front windows, cleaning up around the microhouse, painting much of the interior of the MH, then we did some consultation with Marcin on the design, and turned to domestic labor, getting food, firewood, etc. Later in the evening Andrew and I had a closer look at the 3d sketchup model and found a number of errors that we were glad to catch at this stage, but it means we should go over the model to double check things. I suspected this would happen, but it's not really regression, just part of the development process, I think, here. Not like loosing a file or something. We should be checking each other's work anyway, and expecting errors, especially when trying to shimmy a program like sketchup into the role we want it to fill here, and with all of us relatively inexperienced users of the software. Certainly not in all cases, but I think much more so than is usually assumed, teamwork of this sort can overcome the need - and mean real need, of course they would be able to do things faster, but as long as we get there in a reasonable time frame - for more experienced people.
My log entries keep disappearing, for reasons that are unclear. Some kind of conflict with the multiple tabs open. Anyway, today we cleaned the hab lab and got personal stuff done. One issue that has come up is the presence of CCA wood in the wood fuel pile, and therefore probably arsenic in the ash of the furnace and campfire. According some research just now, http://www.noccawood.ca/docs/ccawood.pdf indicates that a 12 foot long 2 by six contains enough arsenic to kill 250 adults. So I had to think for a second when I saw Andrew vacuuming up ash around the fireplace, releasing a cloud of it into the hab lab. It has probably happened before, though.
Otherwise, I have gotten Inventor mostly working, about 6 days after I first started trying, we cleaned up the hab lab, set some mouse traps, collected non cca wood etc. I have been trying to encourage recovery from my cold as much as I can, too. The time consumed by the tiredness is infuriating, indeed. I have been reading on dimensioning and tolerancing, and hope to learn and do some of that for the soil pulverizer this evening.
Turns out I actually have a cold, I realized last night, probably accounting for the voice and unusual level of tiredness recently. Great timing. From prior reading on the subject, apparently one of the only known orally consumed things that helps is 500 to 1000 mg vitamin C, if taken within 3 days of the onset of the cold, which going by the onset of the voice symptoms, is already passed. Normally I take 500 mg per day of Vitamin C for that very reason; so that the times that I do get a cold, I have been taking it, before it is too late, not having known whatever was wrong with me was a cold. The other thing is zinc nasal spray, but it has the side effect of a relatively high risk of destroying your sense of smell, permanently. Echinacea, the other preparations in the pharmacy, don't actually help address the tiredness, actual viral infection etc. I went into the pharmacy once and looked up the ingredients on the labels of all the nostrums. I suppose a cough suppressant might reduce transmissibility. I am coughing very little anyway, though. They usually resolve in seven to ten days but some can last for up to three weeks. http://en.wikipedia.org/wiki/Common_cold#Management
Done the final review modifications to the sketchup file, and exporting all parts to .dxf files. They are uploading now, the file named "Pulverizer anthonyd mar15 8 40 pm 2013.zip "File:Pulverizer anthonyd mar15 8 40 pm 2013.zip(the wiki can only upload a single file at once, apparently, and indeed give errors if you try more at a time by opening multiple upload windows) Next we need to dimension it. We take the weekend off, but I wanted to do this as an exercise, partly; the thing is this is the sort of thing I normally do on my weekend, learning things and trying to check things off my todo list. It only took about 40 minutes so not too bad; otherwise I focused on advancing things that I can't do during the weekday 9 to 6 hours. I went to the shop to see if I could repair one of the bicycles for quick trips into maysville, and found that there is a bike in nearly working condition, just the front tire needs to be pumped up. Couldn't find a bike pump, though. Will have another look tomorrow. I discovered exercise equipment in one of the side rooms of the hab lab, and moved some of the equipment from the grain silo in there, so we have a compete kit. It brings to mind something someone printed out and left on the table here, the integrated human page from the wiki, advocating for, among other things, a balanced life style. I have not been exercising explicitly since I got here, really. Our bodies are a reflection, in part, of the lives we live, our history, so this heralds regression in my fitness levels, unless I can get a good exercise routine worked into my days here. The weights will really help, certainly, but in practice there is nothing like an actual gym, I know from experience. The intensity of the exercise can't practically be matched, it is mainly a matter of the geometry involved with applying an optimal level of resistance to the various muscle fibers over their range of motion.
Inventor is downloading yet again. Error after error, data corruption, incompatibility, their silly download manager complicating things by breaking down, amazing. They should just use bittorrent.
Something has malfunctioned with this wiki page, as my definitely saved entries for today have gone missing. I will have a look at the history later to figure it out. Anyway, in summary, we spent till about 2 on domestic labor, including cleaning and buying food, and taking a break to say goodbye. The afternoon was spent on whitewashing, painting, removing tape adhesive and other infrastructure stuff. Then we did a final design review of the latest soil mixer/pulverizer prototype. The slated changes are still slated, but I have gone ahead and extracted dxf files from the sketchup file for the parts that will not change. I will then go and export the changed parts, it shouldn't take long to change things and export them. Then we proceed with the dimensioning, to produce a set of drawings for fabrication here on the farm. The drawings will not be super fly, as it is for use here in the prototyping process, mostly. For example, even the plate parts should have edge views, but that would take twice as long to export the files, and it took me about 2 hours just now, and there is very, very little use for drawings of the side of a part made from a plate of uniform thickness when you have the face drawing and thickness.
Announcement to all others at OSE: Just know that my voice seems to be giving out. I'm pretty sure it is from all the talking, and near shouting. I noticed we tend to have to speak loudly due to background noise. I have probably put out more sound energy in the last week than any week of my life, I think! I seem to remember that it is very important to rest your voice when this happens, or you risk permanent damage. So I absolutely must talk less. When I do speak I will have to speak more quietly, too. Also I will try to use signals and gestures instead. I noticed I make the gesture for "I can't hear" a lot, the hand by the ear...:)
Last day for most of us present here. Yesterday Andrew, Marcin, Catarina, Chris and I talked about continuing to volunteer here longer, because Andrew and I do not need to return to school as the others do. The plan is for the two of us to continue to accomplish things here on the same schedule, of CEB press in the morning, and infrastructure expansion in the afternoon, until the end of April, which is absolutely great. I was uncertain and didn't want to seem clingy, probably I went too far with that, but it is a fascinating, incredibly worthwhile and rich project here, and I was very much hoping I would find there were opportunities to continue helping to advance it, and all the learning by doing that goes along with that.
todo: Finish the pulverizer/mixer fabrication instructions for the overall and rotor, this includes adding some pics and the note regarding the relative position on the tines on the tine modules and a few other things (if the tine modules are identical, then the tines will all line up, which is not really what we want.) Doing this gives me a head start on doing the other parts, so I have that on my personal todo list and hope I won't have to jettison it, due to things on the list that I have committed explicitly to doing. Also included herein, is to get the drawings done thus far up on Dozuki. Inventor is done downloading, so I may be able to do something useful with that, too.
The afternoon will probably be lime washing and painting, and progress on the microhouse.
links or list for documents I have added to thus far, to be linked to for easy locating later, for now these are place holdersj: The dozuki fab instructions guides and the wiki pages that are linked to from the master development board/spreadsheets for the soil pulverizer/mixer tool. http://opensourceecology.org/wiki/Soil_Mixer_Design_Rationale
I've got most of the parts exported to dxf, and will do the modified ones tomorrow. I am taking a moment to note here, though:
Got an error message when trying to install the 32 bit version of Inventor, which took since yesterday to download. The 32 bit version won't run on windows 8, so I need to download the 64 bit, back to square one. The importance of having your means of production with you or handy. Andrew had a similar problem, not having his SolidWorks with him. Probably because he never used his laptop to with it.
The morning was spent progressing on the fabrication plan for the soil mixer/pulverizer. This is particularly helpful as it helps iron out some off the design details, such as the manner in which the tine modules interface to the rotor shaft etc. They can be found linked to on the wiki through Dozuki under "build instructions" for the overall and also for the rotor module. I would greatly prefer to add some images, but the feature to add images to the wiki of the existing pulverizer to make things more clear to the reader, and also for Dozuki was not functional at the time, though it seems to be working now for the wiki, so I may revisit this later.
The afternoon was spent moving the swimming pool (parts), limewashing, cleaning at the microhouse, moving lumber and other small tasks that add up.
Very interesting discussion around the fire last night, as you might hope when people who are interested in a project like this come together.
Tuesday, March 11
Morning design. Eric and I measure the tines modules and the more critical rotor dimensions, so he can model it in scad. I document the process itself, in Process used march11 2013 to model an object for concept sketch in scad . My abilities in Inventor are too limited for now to do this in a reasonable time frame, I think, but I will download a copy now so I can try to draw things tonight. I decided to focus on other things until now, since there are others in the group who can draw well, to be the most effective team player I can. I still need to get the pictures I have taken up non trovebox, I was hoping to do so this morning, but the internet connection was down.
This afternoon I spent mostly on preparing the exterior wall of the hab lab, then limewashing it, and painting the interior with the yellow.
Monday, March 10
We made good progress in the design of the pulverizer/mixer, essentially settling on the high level design and finishing the 3d sketchup model. Going out to see the stuff was a great aid, and seemed to inspire and catalyze thinking in the group. In the afternoon I and some others focused on finishing the microhouse interior walls, with the chicken wire and plaster/pearlite mix. I returned in the evening and did another batch, timing myself it took 28 minutes to mix and apply a batch of the plaster mixture, we will certainly be done the lower layer tomorrow, then there is a final finish layer. I am going to bed now immediately after this, entering the log entry and eating, I think, as I have been getting increasingly sleep deprived over the past week. Yesterday and today I tried to remove the ton of malware on this computer with avg free, malwarebytes and the online trendmicro software, which have helped, but there is still something interfering with the browser severely.
Sun Mar 9, 2014
- Pulverizer and mixer design session in the morning.
- Afternoon work included cutting plywood sheets w/ table saw with Coltan, painting the boards, then helped mud up the cracks in the micro house, moving some scrap metal and helping with the dishes. One interesting thing was the difficulty of stuffing the mud into cracks and against the wall due to it's failure to stick well, then the discovery later in the day that mud near the fire was far stickier and would have been a great labor saver. It shows the importance of a brief practical exploration period. Another thing that made a huge difference, which I expected probably existed, was the technique Marcin showed us to get the mud in the walls; take a trowel with mud piled on it and force it against the wall with another, with the first trowel held underneath.