Are the motor wires individual strands or in a sheathed cable? I normally avoid wiring anything as individual strands if I can help it because of the variability. If I’m using 2 wires then I’ll twist them with a drill. If I’m using 3 or more then I’ll braid them together. It doesn’t do anything for field cancellation or anything, it’s just a way to keep things tightly together to avoid those inductive loops opening up.
That’s definitely a case where using a shielded cable for the motors would help, as well. A lot of EMC/EMI is symmetric, where things that reduce noise coupling into a cable will also reduce the noise escaping from a cable.
With my MPCNC I wired everything using shielded multi-core cable. Much like the proverbial tiger repellant, it clearly worked because I’ve had no issues running longer jobs etc. Then again, many other people won’t have issues even with individual wires that are ‘poorly’ routed from an EMC standpoint, it’s all just down to application.
The problem with a lot of pre-emptive EMC measures is that they seem like snake oil until you get into an EMC test chamber because things either work or they don’t, and if they work then there’s no easy way to visualize safety margin. It’s much like building a bridge. You have a bare minimum that’s needed for functionality but then you overbuild to provide a safety margin for things changing over time, real world scenarios not being quite what’s expected. It’s very easy to make something that functions fine in a quiet lab but then fails in the real world for some proportion of people. Back in the original GSM days, it used to be common to see things fail or bit error rates to go through the roof if a cellphone was nearby, due to the modulation used. Now it could just easily be something like whether it’s placed close to a WiFi router, whether there’s mains power quality issues being conducted through the power supply, all sorts of stuff like that. One story I remember was the original arc-fault detectors that were being installed in switchboards could be tripped by keying a CB radio nearby. Rough situation because they’re essentially intended to detect broadband noise, but that was a clear scenario where the stated function was completely ruined by a lack of ‘noise’ immunity.
I just went through the mk4 build manual. If the motor cables match their photos then that’s about as good as I expect you can do without moving them into a shielded cable.
Yeah, that adjacency is probably ok for short lengths, but I’d certainly be suspicious if any of the cables were routed alongside one another for a long time.
If it were me, I’d be disconnecting the LCD and running the cable directly away from the printer to the left in free space. If that solves the issue, I’d be reasonably comfortable that it’s a noise related issue caused by the motor wiring.
I tugged at the cable today and pressed on the connectors again, had a 1:20 print job run without the screen doing weird stuff. Prusa said to:
Regarding the LCD issue, could you try to:
factory reset of the printer (hard reset)
reconnect the cables again (mainly those on the LCD side itself)
try to clean the connectors with canned air
doublecheck the LCD cable for any kind of damage
I can rule out the last one, the others might be a good idea.
Regarding of getting the cables out: I did see a video of someone who did that and said it fixed it. But again: Could be coincidence. I will definitely try the clean and reconnect if it happens again.
Yeah, that’s the thing with intermittent faults. It’s a case of try something, wait and see if it does it again, if it does then try the next thing. I usually try to brainstorm all the things that might fix it and then try to list them in terms of what’s easiest to try and most likely to help. Having a clear idea of what to try next really helps with keeping the motivation up to work through things methodically and can end up with a list of things that didn’t help which can be used to inform future decision.
No harm in trying to reconnect the cables and making sure they’re clearly seated, for sure.
All this interference chatter got me all misty-eyed… You’ll probably appreciate these anecdotes, @jono035.
Both are from a large university medical center.
The first was a network connection that was flakey and inconsistent. We could never find a problem, it was always fine whenever we tested the jack, and the port in the closets. Until I was on the floor, under the desk when the phone rang, and my boss (we weren’t tech support, but we were close, and willing to help) noticed that the network dropped at the same time. We exchanged meaningful glances and opened up the outlet. Yep. someone decided that since the network only used two of the four pairs in the CAT5 cable, they’d just use the other pairs for phone. That triggered a cascade of techies checking jacks for about a week.
The second was when we were trying to map the backbone of the main building (a city block footprint). Well, the 9 story building had thicknet running from the ninth floor of one corner, across the ground floor (you really can’t have basements in New Orleans), up the the other ninth floor corner, then back down to the seventh floor and into the middle of the building for the PDP-11. Now, aside from distance issues, and the washing machine-sized terminal “servers” stacked up in closets this mess had, they had a surplus of thicknet, and rather than trim off the excess and terminate it, they just coiled it up and put it on top on an electrical box (thankfully, not a transformer). Combined with things like an affiliated group being isolated from the network (and the internet) because when they had assigned IP addresses internally, they neither acquired their own IP range, nor did they choose a non-routed range (10.x.x.x, 192.168.x.x, etc). Back when IT was purely a cost center, and got absolutely NO respect.
OK, got that out of my system. I’m good for another 5-6 years now…
It could be something they are dealing with, but the preassembled ones are tested before shipment. Or, the users there contact support more privately, since they may be more likely to just ship it back for an exchange.
I remember that being super common for a while there. I thought that there was even a de facto ‘standard’ pinout for the 8P8C modular sockets that incorporates 4-wire ethernet and phone together but didn’t find it with a quick googling. That’s a particularly hilarious failure mode, though.
Yeah, that definitely made my eyeball twitch. Ugh.
BTDT. I have had an unused pair used in ethernet. Worked fine for 10BT and started giving hiccups for 100BT, typically throwing the link down to 10Mbit. (Switch to switch link, because the same penny pinching bean counter would only pay for the one cable run to the copy room/print farm, and then needed a line for the fax…)
Used to have trouble with the coax cables, one wiring bozo decided to string it over top of the flourescent light fixture ballasts. It was actually OK unless you turned the lights off/on when anyone was trying to get a file from the Netware server. To this day, Novell gives me a case of the shudders.
The office space we had built out in 2004 was still doing the ‘ethernet + phone’ on one cable thing. All the switches were still only 100Mbit, so it didn’t matter and worked pretty good.
It may work with some systems, but whatever PBX they were using was pushing some serious power over the wire, and generating enough noise to freak out the network. Granted, we were in the process of upgrading to 100MB (back when ATM backbone was the shizzle), so that was part of it.
Yeah, somewhat unsurprising, unfortunately. The suggestions from Prusa are pretty solidly along the lines of “well, we don’t know what it is, but it ‘could’ be one of these”.
I’d definitely be taking the LCD off and placing it somewhere such that the ribbon cable gets away from the rest of the electronics/cabling as directly as possible. No idea how practical that is, but I think that’s probably a decent shot at fixing it if it’s a motor noise coupling issue.
I was thinking more about the coupling mechanisms and I think that there’s probably not a lot of high frequency component to the motor cables because the stepper coil inductance is so high and the switching frequency of the drivers isn’t low. They also seem like they may be just running open loop or at least with only average current feedback, which would make sense for being able to drive the switching frequency higher easily. The motor cables probably have a lot of dV/dt on them, given that the stepper inductance is basically acting as the inductor in a buck converter. The motor cables are probably switching between +/- 12V or 24V pretty quickly and likely completely unsnubbed, so they’ll be ringing like a bell. I’ve been meaning to drag my scope down to take a look at the ones on my MPCNC to see what’s going on.
That makes it more likely that there’s higher frequency components causing an issue and primarily E-field components, which means screening may work better than I originally expected. If that is the case, then simply putting a copper or aluminium plate between the two and connecting it to chassis may work. Or slipping the cables through an aluminium or copper pipe for as much of the straight runs as possible and then connecting those to the chassis.
I really still don’t get why not more people have it. There are some, also on video, but it does not seem to have been a wide-spread problem. Weird.
I also had a look at that braid, but I have two problems: a) I might fry the board because all those cables are basically touching the board in the tight buddy board box and b) it’s really expensive.
I am going to wait for the official response that prusa gives me. Otherwise I will try to print an enclosure for an aluminium/copper plate that I can “wrap” around either the motor cables or the ribbon cable.
How I’ve used the braid in the past is that I’ve heat-shrunk the wires, passed them through the former or used the former to pull them through, laid a stripped copper wire along the outside as a ‘drain’ wire, heat-shrunk the entire thing and then connected it to the nearest chassis/shield connection to the source of the noise using that drain wire. That way the entire thing is insulated correctly and shouldn’t cause issues.
It’s a pain in the ass to work with, though, which is why I just use braided multi-core cable for damn near everything
The motor is a big inductor obviously so high frequency through the motor coils is not going to happen.
If there is capacitance on the motor end bypassing the coils, then the high frequency from the driver can still drive a little bit of high frequency AC through the wires and you could get still get magnetic coupling to nearby wires.
What if you added an extra inductor (4 inductors) on the driver end of the wire? With the extra L on the driver end, any high frequency from the driver would be blocked before it gets on the wire, or at the very least it would be a very different frequency, and the motor wires should stop radiating.
Maybe, and this is a super long shot, the LC of the motor wires and motor capacitance is close to the driver frequency for some people and it resonates and goes nuts, whereas for most people the LC frequency is far enough away that it doesn’t produce a problem.
I am just making stuff up. Where am I probably the most wrong?
Also, is it nearly equivalent to put a ferrite on each of the four motor wires individually, rather than wiring an inductor in series with each one?
