More random complaining about things that already happened and I can't fix: cnc mill edition

I have a marginal workshop and a lot of birds called flickers, which serve the same niche as woodpeckers elsewhere, so the flickers punch holes in the workshop walls, and then squirrels move in and eat the wiring and circuit breakers start tripping.

I’ve no idea what they managed to eat the other day, but: I have the CNC mill electronics (stepper drivers) plugged into one outlet on one circuit, the CNC controller (computer) plugged into another on the same circuit, and the mill spindle plugged into an entirely different circuit to help cope with its noise. (Different phase, even.)
I was touching the case of the PC that runs the software, and reached over and touched the heatsink that the stepper drivers are attached to because they were suddenly acting very weird and moving randomly, and got a 110V shock from hand to hand.
I recommend against this. I felt kinda crappy for the next couple of days.

Anyway, in trying to autopsy the results, it appears that the PC itself still runs, and that the eight outputs (this is a parallel port based, linuxcnc install) all still have the right waveforms, when attached to my scope: about 20khz square waves with about 2% duty cycle, at 5V peak to peak, which is consistent with the settings.
The Gecko stepper drivers, on the other hand, are holding the steppers solidly (which made me think they were working) but aren’t accepting input. If I apply a 5V PWM signal to them, they don’t do anything (nor do they react to the output of the parallel port.) So I’m afraid I’ve cooked them. (Also interesting: they have a 36V input, but if I look at their outputs into the stepper motors, there are nonsense values with a DC measurement, and if I switch to AC, it’s showing about 90VAC. They are choppers, so maybe that’s not broken, but… they don’t have inductors or very big caps, so they shouldn’t be able to boost, which makes me think this is something involving some of the FETs blown and the inductance of the steppers themselves, like only the high side of the FETs are working.)

So it looks like I need a whole new drive system.

I could get a Gecko 540, which has all four drivers I need in one big package (and I believe with optoisolation on the inputs.) I could also get some of those neat newer closed-loop feedback hybrid stepper motors. Anyone have any experience with them?

This system isn’t huge: each stepper is about 6A per phase. So any other suggestions would be welcome.

1 Like

Have you read Mark Rehorst’s post on using integrated servomotors? From a quick glance at aliexpress the ones he used don’t look very much different in price from closed-loop steppers. Makes sense; both have integrated electronics on the drive; I don’t think there’s that much difference in manufacturing cost.

I’ve been mostly looking at closed loop steppers, but I’d be happy with servos. I don’t even mind paying extra.

The one kinda questionable thing I’d like to keep is that mine’s set up with dual shaft steppers and wheels on the shafts, so I can run it manually if the stepper drivers are disconnected (separate earlier thread) and I’ve found a whole lot of nothing as regards how to maintain this functionality with closed-loop drives.

I really need to move to abandoning the Sherline mill to being purely manual for small work, and finding something a little larger with ball screws that have WAY less backlash and devoting that to pure CNC use.
But the problem there is that new mill plus ballscrews plus servos plus drive costs considerably more than a 20 year old Mazak or Haas dedicated VMC off craigslist, and they are FAR more capable.



… I would start by getting the AC power & ground operating rationally and correctly. Sounds like your AC safety grounds are not properly connected somewhere?


I haven’t seen anything that brings the shaft out through the encoder. You probably need more projects anyway, so designing your own closed loop control board using a hall effect sensor in a thin package that lets you still put the manual wheels on the shafts is the clear solution here! :tada:

Sorry. Not sorry. Sorry. :man_shrugging:

There must be more local old industrial iron around you than around me.

Since you are already used to backlash compensation from driving lead screws, you could probably look for someone trying to upgrade from a moderate-size bench mill (X3 class?) and put inexpensive ball screws in and just compensate for the relatively trivial amount of backlash in them. Even if it’s not cheaper than old mazak or haas, it might be easier to fit in the shop… :relaxed:

1 Like

I rewired the affected outlets and now everything works. For the moment I got a set of cheap TB6600’s to get everything going.
But the squirrels got into the walls somewhere else that I didn’t realize: I electrocuted one this afternoon, turning on the power, so I have more rewiring to do.
Now I’m looking at 2-3Nm closed loop servos.


Squirrel proofing your CNC apparently needs to be a thing now :rofl:

Maybe some heavy duty cable wrap, giving the :chipmunk: something to chew on besides the wiring? :thinking:

1 Like

1 Like

BTW I realized tonight that I can’t build a setup with the options of manual control and closed-loop stepper/servo, because the driver is tightly tied to the motor and there’s no way to disconnect it the way I do with my separate amp/stepper currently. So if I try to operate manually, the stepper driver is taking all the generated voltage from the stepper.

Now I’m thinking about building a cnc mill.


Mark Rehorst recently pointed to Returned Energy Dump Circuit | Geckodrive | Geckodrive which is a current-switched crowbar snubber; it has just a diode drop in normal operation, but when back-EMF reverses current flow it senses and dumps to ground through a low-value resistor. An electrolytic capacitor acts as a small RC filter for the initial transient.

I don’t like their schematic because i think it would read more clearly if they inverted R1 and D1 but I don’t care enough to redraw it… :relaxed: