the issue is that you are detecting edges and the relationship between them.
if you move back and forth slightly right at an edge, the system knows that you
are in the same place. Once you start dividing, you don’t know if the string of
pulses is the result of movement in one direction, of bouncing back and forth in
place.
This is why I’m saying I would have to plot out all the edge cases of movement
before I could convince myself that this was safe.
or putting it another way.
quadrature encoding isn’t counting pulses, it’s making transitions between
different nodes in a state machine. once you change the input like this, the
state machine needs to be analyzed carefully.
David Lang
Hmm, for a couple bucks you could add a Nano or one of those ARM blue pills (or maybe use up the stack of $4.30 MSP430s) as a prescaler and handle all the encoder issues.
Back when I thought that kind of stuff was fun I’d go one step further and try something like making a step/dir to servo controller so the motors/encoders would look like a stepper. Bit more work and you could move all the kinematics to it and make it look like a cartesian device and move to grbl/tinyg/linuxcnc(via machinekit?)/etc. Although, by that point, you’re back to reproducing Bar’s firmware. Maybe just stick with a prescaler and avoid the feature creep
That leads into another question, just how sensitive is the current encoder wiring to induced noise? Mine is cable tied to the router cord to keep it from getting tangled; by the time a vac (not wired adjacently 
) is added to the mix how many problems is that going to cause with the encoder signaling?
I ziptied mine to the router cord too and I haven’t seen issues yet 
There are 8148 transitions per rotation. At 63.6mm per rotation, that works out to .0078mm chain travel per transition. Even at the most oblique of angles, we can afford to miss one once in a while…
For that matter, the machine might well benefit from a reduction in encoder activity. @Bar, have you experimented with any other encoder resolutions?
Decoder IC’s are available that could offload some of the burden from the Mega, though the decoder routine currently used is very tight. Any hope to wring higher speeds from the Mega will need to look for find more efficiencies, though.
The decoder file documents the truth table pretty well.
What about a bigger sprocket? Seems like with less than 10 micron resolution there’s room to increase the distance per rev. You’d get the same effect as lower resolution encoders
There are 8148 transitions per rotation. At 63.6mm per rotation, that works out to .0078mm chain travel per transition. Even at the most oblique of angles, we can afford to miss one once in a while…
you run into trouble if when you miss them because it then throws of your known
position adn the errors accumulate.
Also the way these encoders work, if you miss a pulse from one channel, the
machine thinks you have backed up, so you are actually off by 2x
For that matter, the machine might well benefit from a reduction in encoder
activity. @Bar, have you experimented with any other encoder resolutions?
remember, this is only a 11 pulse/rev encoder to start with, you can’t really go
a lot lower than that.
Decoder IC¢s are available that could offload some of the burden from the
Mega, though the decoder routine currently used is very tight. Any hope to
wring higher speeds from the Mega will need to look for find more
efficiencies, though.
It’s far easier to move to faster chips/boards than to go down the path of
adding decoder ICs and then communicating the encoder positions to the cpu.
David Lang
the larger sprockets would also increase speed and decrease torque. I ordered a
pair of 25 tooth sprockets, but I haven’t built my machine yet (need to make
space for it)
David Lang
Every issue has hardware solutions and software solutions. Easier depends on who is implementing 
…
In this case we are talking about two different hardware solutions.
k
I found these 3 motors:
https://www.aliexpress.com/store/product/Wholesale-5840-3650-Dc-Brushless-Worm-Gear-Motor-12v-Dc-Motor-Brushless-Dc-Motor-And-Worm/1499567_1000004264047.html?spm=2114.12010612.0.0.5bdadf47FkcUOY
They are the closest thing I found up till now, what do you guys thing about them?
The first one is a brushless motor which won’t work with the Maslow control board but could be made work with an appropriate controller ( would be a nice “upgrade” for maslow!) The other two lack encoders as they are listed but encoders possibly could be added to them.
This one with encoder, the specs is not exactly the same, but I don’t know how accurate those numbers are.
A58SW-555B Worm Gear Motor with encoder disk
The 555-1280 version with 290 ratio seem promissing.
It does look interesting. I wonder how many pulses per rotation the encoder on the motor shaft sends. I couldn’t find the encoder diameter listed, but judging by the pictures I’d guess 16ppr on each channel. With an output ratio of 290:1 that’s 18560 pulses per rotation of the output shaft. That’s something to consider as discussed earlier. Switching out the encoder might be an option, though.
Hey i just found your thread.
Im relativ new to this and im not shure wich one tu burchase.
im quite exited to start bulding .
greedings from germany
-oberlixgmbh
I can tell everyone that I got to the point of documenting the nomenclature for worm geared motors with encoders as well as specs. After awhile you start to see similarities between manufactures.
Just to note the link above that was offering up a motor: I was not able to quickly get the specific info needed but I am pretty sure that one was a no go due to either low torque or max amps.
Although the motors are configurable (If purchasing large amount) we basically have to look for the leftovers of big orders that are being sold on AliExpress, eBay etc.
For those new to looking. You need to find worm geared with encoder (best to keep close to original for compatibility) so the following are important factors. Some can be worked around depending on skills but if you were widely skilled and had the time, just design the power from the bottom up.
Torque rated torque of 30kg cm / now simply for the chance to have something to build and still be involved in the project I broke down and picked up a couple very cheap, small worm geared motors with only 16 rated torque…but with my planned built it may function…we shall see.
They are 12 volt, 324:1 ratio, 11ppr encoder, and way less than max amps (forget right now). You must keep alot in mind…need torque high so high ratio, but keep in mind if you go over 291 you are going to be sending much more signal pulses. Pulses equal to ratio x 11 x 4.
It has not been established, from what I have read, the max pulses able to be handled by cpu. Now i do not know if it will help but when wiring any signal wire best practice is to not have near power wire and if you must cross it, do it at an 90 degree angle.
I think I was able to source a motor that almost hit the original specs but cost like 90 each from robot seller. If anyone interested I will dive into my logs and dig it up for you, otherwise, I am holding off due to funds and the fact that I am building with what a have at this point. Note…my sled is much lighter than the original maslow, plus, have planned to experiment with the sled idea that centers on the mill itself…as I had planned to do that from the beginning after researching plotters.
So… I am late to the party. I want to build one…
Any success stories of a functional Maslow with alternate motors? Or successful purchase of small quantity original style motors? Or does @bar have any motors stashed way that could be sold in small quantities?
Thanks guys. The Maslow looks so cool. I really want to make one!!!
Bar expects to have more kits available in about a month (sometime in
September), nobody has found a place to get the motors in small quantities, a
few people hae found other motors, but I don’t remember hearing of any completed
machines from them (other than the one guy who did stepper motors and worm gears
with a fork of the firmware)
David Lang
Good sum up of the last few months. I did not know Bar was going to sell kits again already. Does not really help me because I literally have everything but the exact motors.
As far as building from scratch. Save yourself some time and buy the maslow pcb in the store. It is pretty limited but I wanted to stay as close to original as possible just to make participation better on here. Now I have motors that have smaller torque, but because I am making a lighter sled. Limited time and space and the ever present coefficient of wife (w). Having a few routers with only one being appropriate with z axis makes the design a bit tedious. Barring any unforseen circumstances I should have a version built in a couple weeks but may change course again and do the go back to using the sled I had built before they had added the triangular function in gc. I switched to standard maslow because weak programmer and no time to make gc work with it…but now they added functionality.
Now I hope some day we source a motor equal to original maslow’s, but quite honestly, I can see just designing my own setup with all that has been learned here. It really is a great bunch of guys. Last I read Bar had no plans to put together more kits. But as stated above that may have changed. If you can get a kit…get a kit. Hopefully once he fulfilled the orders he was like “this thing I made is one of the hottest things in the hobby world right now … should not waste all that promotion and sell some kits”…after chilling from the fulfillment phase. I am so impressed with how these two have ran this project and the forum is filled with great people and ideas. I think it will become a highly accurate and precise machine as Bar never stops using the community to improve the thing.
See ya all later,
Thuber
PS The late to the shoe guy…read the forums about the motors…it can save you alot of time and false leads. Plus if your find one post the link as we all will if we do.