It is not that simple…
Force distribution for horizontal arrangement is completely different: accuracy and precision depend on alignment of railing, gantry etc. table and railing are bendy - how you will make cheap tubing to be straight and not bend on the length of 3m? The brilliance of Maslow design is in fact that only one factor contribute to error - length of chain! Z is local and we compromise on speed. This is not the case with XYZ horizontal cnc - no wonder that it cost 4 digit dollars for 4x8’. Another issue is precision - we achieved 0.5mm but going to 0.1mm is completely new ball game.
I think that Maslow is fantastic compromise between price - accuracy - space savings - DIY capabilities. Personnaly I have a lot of respect to Maslow inventors for elegant design. Make it more complicated and costly is easy - KISS theory!
My gantry design is just a variation on the topic - principles stays the same.
Cheers
Tomasz
Hello Tom,
I am fine with the 0.5 mm accuracy…if it is achieved on the full 4x8’ sheet. And I have respect for anyone that pours his heart into a technically challenging project like the Maslow…and then has the guts to put it out there for people to criticize.
Sincerely,
Aymeric
Something that hasn’t been considered with the gantry that you built is that you can eliminate the ring kit from your design with some minor coding tweaks. The ring kit (and all other triangulation kits) were developed because the math to figure out how much the sled rotated when the chain end points were fixed was complex and, iirc, not very accurate. But in your setup, the sled can’t rotate. As such, the math is becomes greatly simplified. All you need to be able to do is accurately determine where the end points are in reference to the router bit (i.e., 5 inches up and 5 inches left/right).
Wow! Thats something! This would be great simplification - no ring, no rollers, no concentring ring - I love it! I can connect chains directly to router holder where handles were attached (as long as hight is correct). I can lower the weight of “bricks” - less stress on motors… hm…maybe it is time to go back to Maslow accuracy - simplification proposal
Is Ground Control ready for that now?
… on the other topic… I made my workspace flat and true - it take ages but it is worth it. Picture taken after 2hrs - started from bottom left.
T
Seems reasonable… are they threaded? I don’t recall… but if so, maybe a 3D printed screw of something strong can be used with a chain attachment built-in. That way you know precisely where the end point is of the chain. If it is something you are willing to try, I’m sure I can make a revision to the firmware that you can try out (might have a few hard-coded values for simplicity sake… based upon how you ultimately attach the chains).
You’ve made my day!!!
It is 21:00 in Poland so this will be my task for next couple of days 
T
the stock rigid router had bolts holding the handles on. Maybe something like this would work. (I don’t recall the bolt size, so just guessed M8)
With the added bonus of defining the workspace, and perhaps adding a lip to hold material at the bottom? Love it!
Cost and ease of accurate construction.
Look at the prices of gantries and the effort needed to make the rails straight,
square, and non-flexing (and parallel to the bed). That is FAR harder and more
expensive to do than hanging the sled off of one beam on chains.
David Lang
Wow! Thats something! This would be great simplification - no ring, no rollers, no concentring ring - I love it! I can connect chains directly to router holder where handles were attached (as long as hight is correct). I can lower the weight of “bricks” - less stress on motors… hm…maybe it is time to go back to Maslow accuracy - simplification proposal
it’s not that simple, you still have the chains pivoting on something, so you
need either have a clean pivot point or you still have movement to account for.
Is Ground Control ready for that now?
Ground Control doesn’t care about things like this, it just sends the g-code to
the arduino.
David Lang
@dlang is correct about needing to allow the chains to pivot. If the last link is fixed (i.e., attached such that it’s immovable), then it will pivot around the next link and it’s that pivot point that you need to measure to. Look forward to seeing your connection idea.
What’s the likelyhood of adapting the Gantry and code to where one motor is for x and the other controls y.
I.E. one motor located at either the top left or right with the chain connecting to the other side with the Gantry connected to it (think of a garage door opener) and then the second motor controls the vertical motion, which is mounted on the Gantry itself and connected to the sled.
I would think this would simplify so much, except maybe adding a pulley to the vertical motor to reduce stress. If we wanted to get fancy we could mount mag sensors to the extreme limits and it could calibrate itself.
This makes me really want to build a Gantry frame to try it.
I’m just going to say…
Some may not be pleased with where this thread is going, and I don’t always have the tech savvy to contribute what others may.
What I have is my mistakes to share and my “silly” ideas that I see others of you pull off successfully.
I also have encouragement and we all have our place here.
With that said… I encourage everyone to keep up conversation and ideas like this. This is the imagination in raw form that will change everything for the better!
I appreciate that this forum of contribution doesn’t say “That can’t happen.” but instead asks
“How do we make that happen?”
Kudos. 
that would be a major change the the maslow firmware, but is one of the common
modes for the grbl firmware. We just need to make the fork of grbl support the
maslow kinematics as an option.
but you should look at at coreXY http://corexy.com/theory.html it is a bit
easier to do in practice than plain XY (if you have a horizontal gantry it lets
you avoid having any rails or chains below the workpiece, so they don’t have
sawdust/chips falling on them)
David Lang
Something crudely like this?
That’s a standard Cartesian CNC machine.
What about making a post of some sort for each chain with a threaded hole in the top to insert a M2? The screw would fit into the empty hole of the last link and provide a pivot point.
Wow, I’m seriously considering trying this for our Maslow. After all, we already have the rails for our panel saw option.
I really like this idea and I will try to make my own version. Thanks for all the info.
I was thinking about making a design so the cutting bed is adjustable. Maybe bolts to adjust it in and out to help get it flat/parallel?
WB
I’m a big fan of this design as well. I’ve been through the Community Garden / Github files and I have some questions and/or seek advice:
I’m in the US and haven’t been able to track down the “carpentry angles” you reference in your BOM. I looked pretty far and wide, even referencing the ‘DMX’ brand apparent in one of your photos: Can you please provide us with a little more specification or description of those elements?
I noticed that the Fusion & STL files have a slightly different design for the ‘Carriage Bracket’ than is shown in your photos. Looking at the STL, I see that the printed element has thread that might take an M8 screw… I have to say, I prefer the design from your photos that doesn’t rely upon a 3D printed “nut” to fasten into. I’m likely capable of making this change to the design myself, but I’m curious why you changed the STL from the “as-built” bracket.
Until you proposed this strategy for a frame, the conventional wisdom in this forum had been that additional accuracy could be gained by extending the distance between motors and, to some extent, the distance between the top corners of the work surface and the motors. Achieving 0.5mm accuracy across all points of a 4x8 sheet of plywood is good enough for me, but I wonder if you have any insight or can speculate on whether or not your design could be improved by increasing the distance between the motors and/or the distance of them to the top corners of the working surface.
My last question is to simply ask for additional detail on the strategies you used to allow adjustment of the elements, especially gantry’s rails. From what I can tell from you photos, you have bolts to adjust the rail in the Z axis. How many of those did you use? Where did you place them?
Did you implement any strategy to accommodate the Y axis?
I ended up quickly building a sketchup model that combined the ‘bracket’ Fusion file and the table drawings… simply to get a better idea of the assembled … assembly. I’m happy to share it, if it seems like it could be useful.
Let me also say that my Git skills, while lacking, might just be enough so that any progress that I make toward elaborating instruction on how to build a TomD frame could be merged into the Master, if you’re willing. Please let me know and I’ll assure that any changed I’ve made and will make are tracked correctly by Git.
Best wishes,
dugan
going to a coreXY drive would give you far better than 0.5mm accuracy, without
needing to go any wider than ~9 ft (6 inches off each side for skirting)
using motors on the top corners to move the sled will cause you to have to fight
with all the inaccuracies and still deal with the min/max force issues, needing
a wider/higher top beam to get better forces inthe corners.
David Lang