I have a late-batch Maslow4 kit (ordered end of April) and have been slow to collect my notes - so this single post covers assembly feedback, summary of first few cuts and other thoughts. For context: My goal is custom furniture construction, aiming for under ± 0.05 in. error on cuts across 8 ft. distances (about ± 1.3 mm on 2438 mm) in 3/4 in. (~18 mm) plywood. I’d looked at past Maslow iterations, but never took the plunge (router pun intended 
) until seeing the Maslow4 proposal.
Overall reactions
This is my first experience with a hardware Kickstarter product - so here’s my first impression: Wow! I got from the boxed kit to first cut with ~10 hours of work and only a single issue (one of the kit’s linear rods was too long - already well reported in the forums by the time I was assembling). Pushing harder on the machine (working towards my error targets across more of the work area) has been challenging, but I’m making progress and remain impressed.
I expect that the core team (Bar, Anna, Roman) already feel great about what they’ve achieved and ongoing work, but I still want to add my congratulations and thanks. Your visible efforts (shipping working kits, relentless forum posting, publishing 64 weekly updates and 23 firmware releases, shipping improved parts, etc.) are impressive on their own - and personal experience with product development suggests that your additional “invisible” efforts have dwarfed that. So, again, congratulations and thank you - if I ever meet any of you in person, expect free food and beverages!
I’m also really impressed with the community; the forums are packed with thoughtful and helpful discussions. Most issues I encountered were addressed by existing threads with thorough descriptions, diagnostics/tools and work-arounds/fixes. The variety of community expertise is a big help - and it’s cool to see the mix of Maslow veterans and rapidly-ramped new folks who help steer. So: A big thanks to the community, too!
Having said all that good stuff, back to the challenges. The main issue is that I have not yet achieved my target tolerances on large parts (7 x 1.25 ft.). I’m fairly confident I’ll get there because (1) I’ve gotten comfortable margins on small parts - under 0.04 in. error across 7 in. (~1 mm across 178 mm), (2) searching for and following ideas from existing forum threads has taken me a long way - initial errors of 0.5 in. on large parts (10x my target) are now under 0.2 in. (4x my target), and (3) there are still many ideas from the forums that I haven’t tried. However, progress has been a slow iteration between reading the forums and experimenting.
A related challenge is that some progress has been more “magical” than I’d like. In other words, the root cause of succcesful changes is not always clear; this leaves uncertainty about making real progress versus getting lucky. My primary example here is calibration (feels more like an art than a reliable “one click” process). This is not intended as the start of a troubleshooting thread (I really do still have plenty of things to try) but there are more notes at the end if anyone’s curious.
So, in all: I’m optimistic that the time investment for fine-tuning will be worth it.
Assembly/setup experience
I ordered my kit at the end of April, received it in mid May (no issues) and built the wooden frame/assembled the kit in mid June. Total time was ~10 hours (3 hours for frame, 7 hours for kit - identifying/resolving the over-length linear rod issue probably ate 1-2 hours of kit time).
For the most part, I thought the Basic Wooden Frame Directions, Maslow4 Assembly Guide and User Guide were clear and complete. With the benefit of hindsight, I’d suggest the following adjustments:
- Wooden frame directions: Provide more context for the last step, specifically the option to “use 12’ 2x4s to get even better performance”. I blindly followed that suggestion for a 12 x 8 ft. frame - because, if you’ve got the space, why not get better performance? While I think that ended-up being reasonable for me, I didn’t understand the full implications for undistorted work area. Based on forum threads (e.g. Maslow 4 frame size checker, Beyond the green) and @dlang 's maslow4-frame tool (GitHub), I now believe that (relative to a 10 x 8 ft. frame) I gained total undistorted work area (good) at the expense of some mildly-distorted mid-width work area (potentially bad - it depends on part tolerances and layout within the stock). Given the subtleties of Maslow4’s kinematic constraints and implications for frame design, it would be helpful if the suggestion for a higher performance frame included more background - I think this could be as simple as a one-sentence caveat with pointers to relevant forum threads.
- Assembly guide: Address over-length linear rods in the Putting it all together section. I know that linear rod quality is already part of planned hardware improvements but, similar to other forum sentiments, I feel strongly that the assembly guide needs to (1) quantify the expected “small gap between the base of the upright and the sled” and (2) document remedies for persistent large gaps (or at least point to relevant forum threads, e.g. Maslow 4 Linear Rod length, Linear rod support - bent after tightening). Hopefully point (2) becomes irrelevant once the bad rods are out of circulation, but I think point (1) will always be a valuable hedge against potential future issues.
- Assembly guide: Clarify expectations and approach for the tight fit between the spool/belt and arm in the Assembling The Arms section. The current language acknowledges that “the fit will be quite snug”; I had no trouble with the gear fit (motor, idler, spool) but found it impossible to squeeze the last turn of belt into the space between the spool and the idler gear’s collar. I gave up and unwound ~2 ft. of belt past the rollers before seating the spool/belt into the arm. Having since poked inside a fully-assembled/retracted arm, I have to say that the final fit seems … unhealthy - I had the same reaction as @kyleschoen: “This can’t be normal, right?”. Assuming that the wedged fit really is okay, some reassurance in the guide would avoid lots of second-guessing and self-doubt.
- User guide: Remove the (doubled-up) discussion of defunct parameters (
Maslow_calibration_size_X,Maslow_calibration_size_Y) from the About the maslow.yaml File section. I may have missed something, but I believe the calibration grid is only computed inMaslow_::generate_calibration_grid()and uses onlycalibrationGridSize(value 3, 5, 7 or 9).
Cut experience
Unless otherwise noted, all of my cuts have been made with:
- Frame: Wood, 12 x 8 ft. following the Maslow-recommended design - detailed measurements (across outsides of anchor bolts, to nearest 0.25 in.):
Measurement Value (in.) Value (mm) Bottom edge 141.0 3581 Top edge 140.75 3575 Left edge 95.75 2432 Right edge 95.5 2426 Bottom-left to top-right diagonal 170.0 4318 Bottom-right to top-left diagonal 170.0 4318 - Orientation: Horizontal (sled parallel to ground)
- Router: DWP611 running at speed 2 (~18k RPM)
- Bit: 0.25 in. diameter carbide upcut (2.5 in. total length, 1 in. flute length, 2 flutes - Diablo DR75102)
- Cut depth per pass: 2-3 mm (haven’t seen any performance difference across this range)
- Feed rates: 500-800 mm/min for XY, 100 mm/min for Z (haven’t seen any performance difference across this range - staying slow until I hit the error targets for test cuts)
- Stock (work material): 12 mm Sandeply (8 x 4 ft.)
- Calibration: Fitness 0.51-0.55 using 9 x 9 grid with 2000 x 1000 mm area
Initial calibration/cuts were in mid June with firmware v0.77. While these were only “toy” cuts (single triangle at the center of the frame/stock, ~10 in. on a side) I was still very impressed that everything worked on the first try! Unfortunately my beginner’s luck didn’t last - see below.
My next calibration/cuts were in late August with firmware v0.83. There were no deliberate changes to the frame, but I decided to recalibrate because it sat in an uninsulated garage for ~2 humid summer months. Working-around several failures (error message Unable to move safely, stopping calibration), I got a healthy calibration run after manually setting maslow_tlX, maslow_tlY, maslow_trX, etc. in maslow.yaml (prompted by reading forum thread Issues while Calibration today - F70.2 and using @dlang 's Onshape calibration calculator). During resets between calibration attempts, I had issues with incomplete retraction for the BR belt. Bumping Maslow_Retract_Current_Threshold to 1600 provided reliable retraction; I also set Maslow_Calibration_Current_Threshold to 1600. I never figured out why this calibration experience was different from the immediately successful run in June with v0.77, but I also didn’t spend much time investigating.
The August test cuts were 6 faces of a small box-jointed cube (7 in. edges) with 2 faces cut from the center of the stock and the other 4 cut at each of the corners (at least 6 in. clearance to each edge). I was very pleased with the results! I designed the box with 1 mm clearance around each tooth, but it looks like I could could have been less generous and it would have still fit well. When I measured for errors, the worst I was able to find was 0.8 mm.
Since late September, I’ve been working to reproduce those small-part successes at a larger scale (e.g. parts on the order of 7 x 1.25 ft.) with firmware v0.83 and v0.84. The larger parts will use more expensive stock so, for now, I’m doing 1 mm depth dry-runs directly on the spoiler board. I’m still dialing-in to my target tolerance with a best-yet outcome of 0.18 in. error. The worst issues were with cuts near the TL corner; the combination of relatively large TL anchor-to-Maslow vertical angle, loose BR belt (presumably due to bad calibration) and sled friction resulted in large (over 0.25 in.) “waves” along cuts. The largest improvements have come from (1) remeasuring the frame and re-running calibration and (2) raising the TL anchor point by 1.5 in. At this point waves are no longer an issue, only reproducible dimensions. I’m still digesting relevant material from multiple forum threads (particularly Loose belts during calibration, Belts become loose on any axis movement, … bad results with wobbly lines … and Calibration Experience Report) as well as my own observations (Arm diagnostics via telemetry).
…We’re just getting started.