Thanks for the video. That all looks great, it all seems to be exactly as intended.
I agree with your assessment that something is binding and the most likely culprit it seems to me is going to be the spool on the plastic of the arm. Is it possible to put a little bit of the lithium grease on the inside edge of the spool and see if that makes a difference?
Throwing a little bit on the gears while weāre at it isnāt going to hurt anything 
thanks for the idea we tried it and it didnāt appear to have any effect than you though.
Darn, I was optimistic. So something is preventing the spool from rotating freely and itās not clear what that is.
If you put the spool on one half of the arm does it rotate freely there?
yes. the spool definitely felt a little wedged in between the idle gear and the arm however. There is definitely tension on it.
I ran the arm half-closed, and it felt as though the tension that pushed it over the edge was when I pushed the metal shaft into its receiving hole. Almost as though it was happy to rotate if it could push the gear away from it, but once the tension of the two gears held tight against each other was established, it gives up.
After doing retract all if you press Extend All are you able to extend them OK? Like is the issue only in the retracting direction?
It seems so. Extend is supposed to respond to my pressure right? It extends very easily when I pull lightly on the cable and begin to guide it. It does not spool out on its own.
Yeah that sounds exactly right.
I think that nothing is wrong, but the threshold for when it detects that it has fully retracted is just too low. Iām going to make that a setting that you can change in next weekās firmware, and until then Iām going to make you a special version with a higher threshold (because it will take a little longer to link everything up right to make it a setting).
to be precise, when in extend mode, it looks for the belt sensor to say the belt
is moving, and drives the motor to extend it more. the belt has a little stretch
in it, so you can move it just enough to trigger the sensor without the spool
moving.
David Lang
That is definitely what weāre seeing. Thanks!
Thank you!
Give this version a try. This should increase the threshold at which it thinks that the belt is fully retracted.
We can go even higher if need be.
firmware.bin (1.6 MB)
Perfect thanks all the belts fully retract and fully extend thanks for all of the help and I am looking forward to using this machine. 
Excellent! Iām glad to hear it. Iām working on adding it as a setting to the configuration file so that next time someone has the same issue we can solve it right away.
Thanks for working through it with me so that we could find the solution 
absolutely!
Iāll try this version too as mine donāt retract and report back
Iāve been using a small amount of the silicone grease on the inside edge of the cassette anyway when adding the spools - canāt harm it !
Iām not ready to fully test my M4 that I assembled yesterday, but I did fire it up the first time just so I could get it lowered onto the lead screws and the belts retracted. I updated the firmware to v0.65 first. All of the belts would extend but only 2 would retract.
With the modified firmware 3 of the 4 worked, but I had to extend the fourth about 5ft before it could build up the speed to retract (tested twice).
Funny I had it powered off and just turned it back on, and now all 4 are retracting fine with your modified firmware here. For kicks I went back to v0.65 but once more only 2 would retract. So again switched to your modified firmware in this thread, and again all 4 are working perfectly fine. 
That is great to hear!
Those changes have actually been incorporated into 0.65, but now itās a setting. In 0.65 and later you can set Maslow_Retract_Current_Threshold: 1300 to be something higher in the .yaml configuration file and then it will work again. That modified version has it set to 1800.
Once we figured out that it was a working solution we integrated it to the main firmware.
Iām having this same problem. I just loaded the latest firmware from github, which claims to be v0.65, but the status display in the fluidnc tab shows the loaded firmware to be v0.64.
When I hit āretract allā, only one belt retracted. the others didnāt even twitch.
I went into the fluidnc tab, local files, config items, and set the /Maslow_Retract_Current_Threshold to 1800 (was 1300), hit SET, and went back to setup/retract all. no movement, and the following immediate response in the message window:
[MSG:INFO: Retracting all belts]
[MSG:INFO: 3 pulled tight with offset 0.011]
[MSG:INFO: 1 pulled tight with offset 0.000]
[MSG:INFO: 2 pulled tight with offset 0.000]
[MSG:INFO: 0 pulled tight with offset 0.000]
Do you think this may be a consequence of the lack of gear lash that we discussed yesterday?
I havenāt actually tried changing it in the settings like that. I think that it should work, but Iād trying changing it in the Maslow.yaml file and then restarting to be sure the change is really taking effect.