Originally shared by James Rivera Ok,

I’d have to look it up to find the photos. M controller used to be hanging in mid air by the wires, dangling right next to the SSPC and power supply. Your set up is pretty clean in comparison.
Think it’s mention above, NC all and off.
Looks like you still have some set to homing. Z max, A min, Y min… no switches = all off.

@Brandon_Satterfield I set XSN, YSN, and ZSX to 0 to disable homing switches after I posted all my settings. Are you saying I still need to alter something else?

@SirGeekALot it looks like your controller is hanging on the wall away from the cnc in between the 2x4"s? Hard for me to tell in the photo. How long are the wires?
That said you have nothing connected to the end-stop connections on the tiny G right?
You can also try grounding the end-stop signals at the connector just to insure nothing funky. Probably not since they are turned off.

I doubt your problem is radiated as it seems your controller is pretty far away. So don’t think a shield on the controller would help.

My guess is your problem is conducted noise through the power supply grounds.
or
Flaky board
or
noise from mains
or
noise from PC connection

When it stops:
Is it always at the same place on the same job?
What state are the tinyG leds in?
Does the stop correlate with any action like a spindle change in speed, change in direction or increased load?
Does the stop correlate with anything external to the cnc turning on in the shop?

Grounds
How are your power supply’s grounded did you use a single point ground schema as I suggested in my long post?

Hard to tell but I do not see the grounds on the 24V and 48V supply grounded to each other and to the safety ground.
If not find a single metal place to bring them all to with a good size wire.

These kinds of problems are hard to find. I found it best to “try stuff” until you get a symptom that makes sense and points to the problem.

@donkjr I’m not 100% sure on the TinyG LEDs, but I think they were in an error state. The failure is never in the same spot (or I would suspect the gcode was messed up). The failure always manifests itself as movement halted with the spindle still running, and the system is always unresponsive. I have both power supplies on the same power cord, and that is the only ground. I’ll attach a better picture of just the electronics in just a second.

Also, the main house power circuit breaker panel is just to the left of my K40 laser cutter, which I have used a wire to ground directly from the K40 ground post to the same ground my heat pump (on the other side of that wall…hmmm) uses which is the pipe on the power panel.

Is there any place your wiring is crushed, bent sharply, nicked, or any connections on your board / motors that are loose? How tight is that zip tie on your gray motor wires?

If nothing there, then let’s assume your board is fine, your wiring is fine, and go to the next possible culprit; your computer and connections. I have one of those see-through aesthetically pleasing USB cables and it’s garbage. I’d try swapping that out with another one.

Chilipeppr is a massive resource hog. What browser are you using to run it? What kind of computer / OS are you running?

You also have your spindle power wires running together with your motor driver wires. There’s all kinds of potential crosstalk happening there. I’d recommend separating out all 3 kinds of wiring: computer to controller (move the USB cable up away from the stepper motor wiring), motor driver wiring, and spindle wiring should be separated.

I’m sure someone may point out that it’s not the issue or due to some [physics] to not worry, but at this point we’re debugging in the dark and I’m eliminating zebras (Occam’s razor).

Is there anything you’re doing on your computer when the CNC is running? I’ve had both Chillipeppr and Cura fail when I’ve tried to watch a video while it was running and yet it didn’t throw errors, it just freaked out but looked normal. Does your screensaver come on automatically?

@SirGeekALot Try these.

  1. Run a separate wire from the ground of the 24V supply to the ground of the 48V supply then run a test.

  2. If that does not fix it connect a separate wire from the safety ground on the 48v supply to the ground where you connected the 24v ground while leaving #1 in place.

  3. If that does not work try flipping your 48v supply horizontally and run the input and output from the spindle driver down the right hand side and away from all other cables. Leave 1-2 in place

  4. If the spindle drive goes through a drag chain with other cables remove is if even just to do a test. Leave 1-3 in place

Note: my electronics are mounted below the cnc and close to the motors, sensors and gantry.
The spindle motor driver is on the gantry so the only 48 line going through the drag chain is the supply. The motor drivers output is a short wire from the motor control to the motor and does not go near any other wire bundles.

Google Photos

Google Photos

@Jace_Richardson I’m running Windows 7 on an Intel Q6600 quad core, and I do not run a screen saver; it just turns off the monitor after something like an hour or so. It is not a new PC, but it is still pretty beefy IMO.

I’m running CNC.js (not Chilipeppr; I agree with you about it being a resource hog) and I’m not running anything much in the background (aside: I have chrome up in the background, but not running anything intensive or with ads, and I have an extension that save the state of unfocused tabs, stops their processes and unloads the memory they use, so chrome is not the issue).

I also switched to a better USB cable a long time ago when I first starting having these problems. This one even has a ferrite core to help suppress EMF.

There should not be any nicked/bent/crushed wires, but I suppose I can double check. I will see if I can move the USB cable farther away from the other cables.

@donkjr 1) perhaps it was not clear in my photo, but the 24v and 48v power supplies use a single power cord wired in parallel, so they already have their grounds tied together.

  1. it sounds like you’re saying there is a separate “safety” ground on the PS? I’ll look for it.

  2. Is this to put the PS magnetic fields at 90 degree angles to each other?

  3. Yes, I do have pretty much ALL of the wiring going through the drag chain.

Is the reason for mounting the spindle driver on the gantry so that the raw/constant 48v power instead of PWM is running next to the signal cables? That seems like good idea (a forehead slapper, actually, now that I think about it).

Also, as I mentioned above, the external unit for my house’s heat pump system is almost exactly on just the other side of the wall where the electronics are mounted. I guess it turning on and off could be a huge source of EMF spikes. Do you think that could be the culprit?

@SirGeekALot

  1. The power cord safety ground is not the ground I am talking about. Although continuity-wise they may be connected they also may create a long return path for high currents. I am literally suggesting connecting a single wire between both ground lugs irrespective of how the safety ground and AC is connected.
    This sounds nuts as a meter will tell you they are connected. This kind of problem is not a static one its all about insuring high speed currents flowing in inductors are following a short and predictable path from source to destination and back leaving as small a transient voltage drop as possible.

  2. I am trying to get all the source supply grounds to exactly the same point (transient potential) and not let those currents flow through other paths. You want to make sure 24v and 48 v current flows directly back to there sources (power supply) and are no way shifting each others ground. If they are connected to the same physical point their grounds have to be the same potential.

  3. This is trying to get the high spindle motor currents away from the other wiring by running it down the other side. If you move the spindle driver to the gantry you don’t need to do this.

Why put the driver on the gantry? The spindle draws significant current on start up and during load transitions and can be the source of conducted and radiated noise. These transitions have very fast slew rates driving a lot of motor inductance. The long wire going through the drag chain is a good antenna and can easily shift grounds. On the other hand if the 48v supply has good ac response you should see few transitions in the motor driver supply leads. Having the wire from the driver to the motor short as possible is good hygiene from a multitude of perspectives.

I doubt your pump is causing EMF problems but it could be injecting noise into the AC that is making it back through the 24V supply. We should look at this as a last resort.

My bet is that your wiring is to long and you have radiated and conductive noise from the grounding design and wire routing.

I still do not understand what the LED on the Tiny G are telling us when it fails. What lights are lit what color?

@donkjr it would be pretty easy for me to relocate the spindle driver by attaching it to the plate on top of the z axis extrusion (see photo). Acceptable, or too close to the spindle and/or stepper motor?

Relocated spindle speed controller to gantry. Running a test now (the job I’ve been trying to finish).

Here is what the TinyG lights look like while it is running ok (so far).

@SirGeekALot I think you would be ok with that gantry location noisy motors next to noisy motors are usually ok.

@SirGeekALot Does this mean its working now???
If it fails again please note what the CP serial console reports. There should be a status line that has an error code. Perhaps a photo.