AbeAnd Logs October
- 1 Tues Oct 31, 2017
- 2 Mon Oct 30, 2017
- 3 Sun Oct 29, 2017
- 4 Sat Oct 28, 2017
- 5 Fri Oct 27, 2017
- 6 Thurs Oct 26, 2017
- 7 Wed Oct 25, 2017
- 8 Tues Oct 24, 2017
- 9 Mon Oct 23, 2017
- 10 Sun Oct 22, 2017
- 11 Sat Oct 21, 2017
- 12 Sun Oct 15, 2017
- 13 Sat Oct 14, 2017
- 14 Fri Oct 13, 2017
- 15 Tues Oct 10, 2017
- 16 Mon Oct 9, 2017
- 17 Sun Oct 8, 2017
- 18 Fri Oct 6, 2017
- 19 Tues Oct 4, 2017
- 20 Mon Oct 2, 2017
Tues Oct 31, 2017
Meeting discussed MT changes to be documented and re-edited in CAD. Also clarified various points on LT design constraints. Existing frame design should be good. cab needs to be narrowed to create a small gap between verticals, arms, and cab frame. Likely need to use 2x2 tubing for more space, but reinforce with bracing. Maybe 2x2 cab frame could be reinforced with curved plates cut on torch table.
Mon Oct 30, 2017
Figuring out some FreeCAD errors and file issues I created while editing the QA adapter. I nearly had the QA adapters done last night, but then likely due to the way I coped and edited file parts started disappearing when compounded and FreeCAD crashed.
Meeting with Roberto about errors and file workflow.
Very informative instruction and discussion with Roberto on FreeCAD. I recorded much of it with vokoscreen, unfortunately, my mic audio is garbled and Roberto's very faint. I'm unsure what audio settings are best in vokoscreen. From of his tips I can improve workflow and eventually script some more instructionals.
I finished editing the wider QA option for the LT following his suggestions.
Trying to add this to the bucket I just realized I did some math for the extension wrong somehow. So I will get more practice re-editing that.
I did not add a support cross bar on this version. It can be considered tomorrow.
Sun Oct 29, 2017
rechecking & refining math & changes on QA. I may have confused measures over the 45" to 42" difference in width. Also realized I needed to look through more file versions of the existing QA because the off the shelf plates are narrower than what was o n the MT anyway.
figured out the male QA tab width from diagrams. It also looks more like the photos.
I also did a quick measure and it looks like the 36" cylinders will fit vertically from the horizontal frame. They still only have a stroke of ~10" and much less movement and stress in that position. With the current arm config it looks like the top end woudl fit about half way up. If the bottom end is several inches back from the verticals the motion is minimal and nearly vertical in the raised position depending on starting angle. If boxing in the bottom of the arm only a small hole would be needed.
I kept trying to think of any easy way to make a universal adapter for a wider ALT width, but I finally realized the male QA side doesn't have to be adjustable at all. With enough added cross bracing I think adding extensions to the male QA plates might be strong enough even at 60". To get 57" for clearance past the outer frame it only requires a 7.5" extension plate plus 2" overlap for a good weld to a 2x2x13" spacer bar.
Now I see another option to keep the arm lift cylinders mounted vertically the verticals could move behind the existing cab with the 45's on the sides trimmed to allow the arms to sit over the outside frame. The male QA would require only minor extension.
Sat Oct 28, 2017
Continuing looking at arm width, QA mount points, arm shape and cylinder mounts points.
I was looking at potential arm shape. Obviously, the large arms have to be made from two sections. I was considering an S curve shape or a symmetrical elliptical curve shape that might be efficient to cut. Cutting curves by hand would be difficult even with symmetrical shapes and some geometry tricks though.
A more curved shape for the arms may lower the angle for attaching the front curl cylinders as well.
Looking at the BCQA the outer most holes for the male pins will give another 6.5" and the current MT female plate is only 42" where the BCQA standard diagrams show 45" max. So that gives a total of ~8" the arms can be moved out from the MT width. It looks like the male QA plates can work in that config. The pin attach plates on the male plates will have to be on the outer edge so the geometry will be close. The only problem is then the arms line up with the vertical supports. So maybe half the ~8", which will put the male pins close to the inner edge of the wider slot.
I was thinking the vertical supports couldn't be moved in because of the tensioner and motor plates, but actually, then the motor bolt plane lines up with the the outside of the tensioner. The 45deg braces would have to be welded to the sides of the verticals, but I don't see any interference with anything else. I am concerned about arm alignment with the max QA plate width or at least taking into consideration the 1" pin attach plates.
Examined geometry and measurements better. There is overlap between verticals and arms if trying to push the width to 42". The Pin attachment plates being 1/2" also pushes the LT arms in a bit more. The pin attachment plates don't have to be the same though. The verticals have to stay put. The arms being on the inside of the verticals gives 32"'s of space for the operator.
Updated LT master CAD so arms are correct width, lift cylinders are in position and curl cylinders are aligned with arms. Space for cab is 32" wide.
Working on QA. Moved QA mount plates to the edge. Need to move pin assembly ~4.6in out and finding specs for the new hole.
I don't see a file for a separate female plate, but it may be in an old file version. Referencing
network error confused my file versioning on upload, but I think I got the male QA plate modified for LT width.
Fri Oct 27, 2017
Finding links to cylinder info for LT. The old BOM doesn't list 36" Cyls, but I'm estimating the stroke from old diagrams and whats available from surplus center. http://opensourceecology.org/wiki/Bent_Loader_Arms#Hydraulics
Surplus center lists 36" Cyls with strokes as low as 30", which seems like a lot, especially for any angular stresses.
It looks like 30" is excessive if thats the stroke. In fact, if the cylinder could be raised up to 80deg the max extension is only ~47" and the cyl nearly hits the top shaft. Obviously, it can't go that far since the arms would be behind the vertical supports.
I do not see an easy way to lower the frame end of the arm cylinders without major frame mods, so moving the cylinder (arm) ends up a little higher into the arm corner will give it a better lift angle.
I've made a mess of the LT master CAD file while making reference points, so time to clean it up...
Thurs Oct 26, 2017
Started looking at the cylinder movement and placement by copying in the bucket and cyls from the MT and the geometry sketches.
Wed Oct 25, 2017
trying adjustments for LT arms.
The cab looks longer than necessary so bringing of the front frame in still leaves ~3' of space for the operator, but the 23" between the arms is a little cramped. I'm not sure how the controls were positioned in previous designs, but at the left and right sides for front ingress/egress of operator seems easiest. However, that will leave little clearance for the operator to get in and between controls and arms. Also having the controls mounted in a single front central mount might be easier for routing and operation.
Tues Oct 24, 2017
Checking existing measurements on LT arm width, clearances, and spaces around the cab.
Considering what may have to change to accommodate standard bobcat QA.
MC female plate width extent is 42". Extent of outer holding pin holes 35.89".
LT current arm outside width 57.5". Inside of arm wdith 49.5". Outside of cab frame ~40". Inside width of vertical frame arm supports 40.097"
current LT_quick_attach female plate is 58" wide.
It looks like the only easy way to create direct compatibility is to bring the LT arms in a lot to the same width as the MT. They will have to come in line with the cab frame, which the current arm design interferes with.
Otherwise, it might be a nice feature to keep the arms wider on the LT and build a universal adapter frame. I think that would require a different design of the male QA mount though. It would need to accept vertical pins. Maybe additional boxes on the sides would allow for both, but to reduce waste a universal design would be ideal. First thoughts, it may require an extra layer with horizontal bars with holes for multiple widths. It may also require 90 deg pin adapters. If the pin mounts on the arms are best left horizontal. Such a universal adapter frame would need to be large enough for 1" pins and add extra weight though.
Just from looking at the MT CAD that toothbar looks time-consuming to build. I think cutting a sheet in half with a tooth pattern such that you make two bars might be faster and almost as effective. Not sure how fast a human could cut that obviously for CNC that would be easy. Maybe cutting it out of thinner material then bolting both pieces to the bucket would work ok too.
Mon Oct 23, 2017
Created page LifeTrac_17.10 also looking for page on versioning, but search and google not turning it up yet. Lots of old pages on documentation standards, taxonomy, nomenclature, versioning, etc. Hard to decide where to agregate links to the new stuff.
I did some wiki editing and exploration tried to understand the categorization system again and created the subcategory Category: Nomenclature and Taxonomy for the naming convention pages. I'm not sure I understand all the formatting options yet or the best way to label pages. Most pages probably should not be categorized. Subpages are an option, but deep nesting would eventually be an issue.
Also attempting to edit the track unit CAD and constrain intermediate linking to the existing parts in a repeatable and flexible manner, but I suspect it will not work as needed.
I saw a recommendation on the freecad forum I try the 0.17 built-in nesting methods I was assuming would not be compatible. With the torch table not yet operating smoothly I'm making this a low priority until all tractor details are done, but I am curious to see if it is more improved recently than I thought.
Edited track unit adding second interlinked track unit.
I think I have an idea for interlinking the two different track links separately from previous reading on path arrays. Two copies of the sketch can be made with two different track unit compound path arrays. Then the path arrays might be adjustable to alignment.
Sun Oct 22, 2017
watched a few videos on YT including speed review of meeting again while reviewing current work and docs on micro tractor and more. Looking for subcomponents to examine for likely modifications I noticed many CAD files aren't linked on the wiki pages for the tractor, so I started cleaning some of that up and linked some pages.
Reviewed quick attach geometry and design.
Looking through concept doc for remaining changes. https://docs.google.com/presentation/d/1Ol6Mlq69kRfD8afiKQYVdXoNsmVhsEz_IFpmVYH5vHY/edit#slide=id.g2724b1ec29_3_0
Continued looking at current design and any potential issues or misalignments.
Current track link design is more accurate looking but does not contain the intermediate links, which technically makes the track unit space wrong. I'ver tried to think of a new lower labor method for making track units, but the previous/existing track method looks sturdy for such a small simple track. And without the torch table working any additional track work is probably easier to be done the old way. https://www.youtube.com/watch?v=77195alVAC4
edited track unit and uploaded then realized I now need to re-assemble the links so the intermediate links go on the inside as shown in video, which gives more space for ease of welding.
Sat Oct 21, 2017
Back from a longer trip than I expected. Catching up on tractor CAD. I see the Torch Table has some software motor control issues. I will shift more to finishing CAD details on the tracks since I was getting close on that anyway.
Updated tracks in master CAD to fit around idlers better. Not perfect, but the track appears to ride on the idlers now.
Sun Oct 15, 2017
Downloaded September or October? OSE version of Marlin. Reading up on it on their website. Inkscape Gcode tools extension also says there is an update availvable, but I'd hate to get into new version issues.
The green thing settings for plasma might be similar to gas torch, but I don't know enough details about Gcode Marlin toolchain yet.
More SVGNest testing. I thought I was getting close using inkscape to open DXF's and save to SVG. First, I thought I was selecting the bin and it ran, but it was the pad and parts were split into circles. Then I found Inkscape is to buggy to even copy and paste the 4x8 sheet bin to the parts file. I had to use import. The graphics look like they should work and I followed instructions to convert to paths and join nodes, which appeared mostly unnecessary converting from DXF. Maybe that is a problem, but they look ok unlike exporting to SVG directly using 0.16 with missing holes or 0.17 with polygons instead of circles and arcs.
Sat Oct 14, 2017
Looking for torch table software tool chain info. I see a spreadsheet that is dated older, but references a switch to arduino reprap style software. I expect it is similar to other D3D machines. The torch table software details may not be refined yet so I went back to the microtractor part details and updated the track parts so the position of the track and asociated parts can be refined.
I am now tracing the track path around the components in the master cad to refine its shape. Rotating the track unit compound relative to the sketch plane seemed to get the track pads aligned out correctly. This can be confusing, but the freecad documentation does explain this. Further forum searches and reading suggests changing the center or origin point of the part may make a difference. Making the sketch line segments more detailed and aligned around the sprocket just right should give a good track eventually.
updated track CS with reference parts from the master CAD. Close at least the position relative to the idlers is ok. Adjusting the position of the track unit compound relative to origin should make it rotate tangent to the sketch and fit the sprocket better. Now at least the distance needed to clear the arms can be checked.
I just remembered I need to reset the track link spacing to 0.8" for the 3/4" sprocket.
Fri Oct 13, 2017
Installed FreeCAD 0.17 and tested the SVG Export. FreeCAD 0.17 managed to export the SVG more completely except the curves and circles are now lines and polygons. I think I read there was setting to change, but I can't find it now.
Exported DXF's and copied parts together into a single file for track unit torch test. LibreCAD is buggy even at basic copy and paste. Actions repeatedly crashed unexpectedly or simply failed to work with consistency.
Opened DXF's in Inkscape attempted to setup G-code. I'm not exactly sure what settings should be for CNC torch table. More research needed.
It also appears it is not generating G-code because the lines need to be converted to paths.
Tues Oct 10, 2017
Made a video showing the inconsistencies in SVG output.
I also posted a query to the freecad forum.
Meeting clarified re-use of previous drive sprocket with 3/4" plate and possible design solutions for the track tension.
The box around the upright tubing method using 1" bolt is simple. Need to consider the position of bolt hole relative to frame tube holes. It'd be nice to design in an easy method to apply leverage to the box against the frame to add tension while tightening the bolt with an impact driver and so one person can do so safely.
Track units need 3/4"+ spacing between links. So the new track unit file needs re-adjusted.
DXF to G-Code in Inkscape
I also assume I need to brush up on LibreCAD for editing the DXF's.
Got a response on freecad forums suggesting 0.17 will work better then I realized I also failed to read all the way through the thread and there are more instructions on using Inkscape to fix the SVG's. Looks like a few more steps, but it should be worth it.
Mon Oct 9, 2017
After having worked with and looked at the SVG's generated by Freecad and a comment I see on jack.works I don't see how this will work directly from FreeCAD drawing SVG's. The SVG's are made of broken independent line segments and not continuous shapes and they need to be for the software to work. So evaluating the software that generates the SVG's in FreeCAD may lead to a solution long term. Other conversion routes like DXF to SVG using another app OpenSCAD or any other open source tools. Given this is a valuable open source project may be more interest can be found for recoding FreeCAD for compatibility.
SVGNest also appears to have a number of bugs and issues with types of lines. The question is whether further development can fix these easily or they can be worked around with minimal user effort https://qcad.org/rsforum/viewtopic.php?t=4334
There is recent ongoing work on a nester in FreeCAD, but it is alpha software and still missing needed capabilities.
I also found some other test SVG files on freecad forums when I uploaded to svgnest they appeared incorrect probably due to arcs and splines.
Thought I might be doing something wrong with the freecad SVG export since others on the forums suggest svgnest. So I followed instructions to use 2dview first then export to flattened SVG. This appears much better and would probably work except that it left out the holes in middle of the part. The 2dView shows the part correct, but not in the SVG. I'll try some more parts just to see how buggy it is.
I exported a 2DView of a 4'x8' sheet and the PTO motor mount which appears correct, so I created a layout of copies in a single file. The scale did appear ok. When trying to select the bin in svgnest however it still selects seperate segments of the bin and every other part instead of the whole rectangle for the bin. I assume this means the parts are not made of continuous lines such that is can identify them as a single component. I am saving the file as a plain SVG instead of the inkscape format in case this causes any issues.
Sun Oct 8, 2017
With reworking the track units from the basic sketches up I found some of the details were different for previous builds or files were just unfinished. I'm trying to make the CAD accurate enough that with metadata it is effective for more of the documentation.
There is a variety of different documentation and examples of track discussion so there are a few small things I am unsure of.
I see previously washers were welded on rods and drilled for cotter pins, but I think I recall a discussion in a meeting recently suggesting bolts worked better, but cost more depending on size and grade obviously. The current CAD looks like a rod with washers on both ends. What design is planned to use for this build?
Also since 1/2" plate is being used compared to 3/4" before the gap between the links has to be narrower to minimize the roller play. I am guessing this could make certain welding methods harder. I set it about ~0.6" in the current CAD. Obviously, this affects pin length as well. I'm assuming there should be at least an 1/8" of play in the pin.
I'm moving on to solving the nesting today so I can evaluate that and report a strategy for the next meeting. There appear to be many parts and assemblies made from 1/2" sheets. There must be DXF's for many of these from previous builds so I will search for those for testing. The first step is defining a clear process and making a how-to video.
Considering there are so many assemblies to break down and convert the best method may be sprints so work can be divided in parallel. Experimentation with all the parts in one nesting may help determine certain patterns. Obviously, multiple sheets are required, but given the mix of geometries, it is not safe to assume ideal part mixtures for packing order. Rounding off more corners on parts may increase packing given all the circular parts. Some small parts could be nested inside other waste parts, but that requires hand optimization of DXF/SVG and order of operations through gcode or however the torch table code chain works.
Looking at svgnest further it has options for part in part and concave search likely needed for parts like the sprocket. 3/16" spacing =~ 13 pixels assuming the software exports based on 72dpi. Inkscape appears to easily import and save between SVG/DXF. Batch capability would be great. I see some other extensions as well
hmm at the above link it says Inkscape uses 1/90 of an inch and mentions other potential bugs, but this is from 2009...
According to documentation, the flattened SVG file option should create a viable SVG using the raw CAM setting, but all I get is overly complex and garbled output.
It almost looks as if DXF-->SVG-->(1)SVG-->DXF will be necessary. This is way to complex and time consuming perhaps there is a way to create some python scripts or macros to fix this and batch process files. A more complex code project might be to adapt existing code libs to make svgnest into dxfnest.
I sent an email about all this. Also see User talk:AbeAnderson
I'm looking at modifying the isoview macro for SVG output. This seems excessive and adds more steps, but the FreeCAD SVG export is not creating useable files.
There are some existing tools to combine, merge, stack SVG's into one file.
This one works by merging and placing space between them.
I used the drawing WB to manually export SVG's of the track chain link ok, but it still requires editing out all the FreeCAD stuff then making 88 copies of the link in inkscape. I am assuming scaling of 1.0 will correlate the size correctly relative to 90ppi in Inkscape. Further reading of https://github.com/Jack000/SVGnest clarifies some functions. I'm thinking the easiest way to create a 4'x8' bin for the steel sheet is to draw and export it as an SVG as well. Then all the other SVG's at same scale can be copied in.
I'd like to think we can bypass that extraneous FreeCAD code and just write each part to a plain SVG, but the built-in class functions may make that more code work than ideal.
DXF files are dimensionless so as I recall you can set the scale in LibreCAD. So now I am thinking as long as the scales are the same for the orthographic projection in drawing bench (maybe 0.1) parts can be copied in to the 4'x8' bin file for upload.
Problems with scale, exporting everything at 0.1 did not help apparently. Also still have some inconsistencies or confusion in part diameters. I created a test file anyway. It doesn't seem to work I think because when I try to select the bin it only selects a side of the box. So something about the XML is off.
Fri Oct 6, 2017
Issues with existing components of the track are making it hard to edit so I am going back through the parts and constraining and editing them to more recent standards. For small simple parts like this, I think a fully editable assembly would be nice if I can figure out an ideal working chain.
added notch to track link. Also noticing the existing track unit example files are drawn with parts of various incorrect measurements like 3/4" thick rollers. Not sure what the actual specs are that have been used in the past other than logically basing it all around 1/2" plate. Probably better not to have so much slop in the rollers, but spacing the track links close to 1/2" may make them hard to weld to the track pad.
re-assembling track unit with fully editable parts constrained symmetrically for ease of future editing. spacing on the links is ~0.6"
next steps are the 3/4" bolts needed to replace old pin design and 1/2" thick rollers. Unsure if roller specs need to be adjusted +/- 0.05" for bolts. Sounds safe to assume torch kerf will be enough.
Tues Oct 4, 2017
Researching track design further and attempting to notch the tracks without redrawing entirely, but some glitches or misalignment in the original design seem to be making that difficult. I see the pin size is also likely incorrect since the roller ID is 0.75", which seems more reasonable if as large high-grade bolts are rather expensive. I see information that the bolt size for 1" is excessive. It would be ideal to reduce sizes as much as possible to reduce waste in nesting. For nesting, there may be scraps that could have smaller parts cut out first, but this would require hand optimization of the nesting since the software is unaware what is scrap.
Mon Oct 2, 2017
Created PTO motor mount plate sketch absed of PTO motor polanar to frame. sketch still needs work.