PDA

View Full Version : M542 driver - motor vibration at low speeds



birchy
18-11-2013, 01:24 AM
I'm currently testing the first (Y) axis on my Warco WM16 mill conversion. I'm using standard M542 drives with THESE (http://www.zappautomation.co.uk/en/nema-23-stepper-motors/380-sy60sth88-3008bf-nema-24-stepper-motor.html) 4.2A, 3nm stepper motors and have noticed significant vibration at very low speeds. Is this a problem caused by lack of cable screening or is it a setup issue? At certain speeds the whole machine and stand will vibrate. At higher speeds, there are no issues. I've tried running the motor unmounted but it still vibrates at very low speeds. The motor is bolted to a solid mounting. Should I use some sort of rubber mounting to reduce the vibration or is it something I will have to live with?

EddyCurrent
18-11-2013, 07:47 AM
What's the switches set to ? and also Mach3 settings

Jonathan
18-11-2013, 12:05 PM
Answer Eddy's question first, but what your describing is resonance. When the machine is excited at particular frequencies, the vibrations will build up (like a tuning fork), so ideally you want to avoid these speeds or damp out the vibrations. One way to avoid them is to increase the micro-stepping resolution - this increases the frequency and can reduce the range of speeds at which resonance is a problem. One bonus of driving the ballscrews with timings belts is they have a noticeable damping effect.

If the motors aren't actually stalling you might be fine. A video may help...

EddyCurrent
18-11-2013, 02:41 PM
Yes, I was thinking they might be cogging badly at low speed due to low micro-stepping resolution, or starved of current.

birchy
18-11-2013, 06:18 PM
I'm using LinuxCNC set as per the picture below. Dip switches on drive are set to 800 steps (all switches 'off' except for 5, 7 and 8). I've also tried running at 1600 steps (and changed "Driver Microstepping" to 8) which did improve the resonance slightly. At 3200 steps, I had some error message related to the RTAI (realtime kernel). I didn't look up the specifics of the error as I just assumed it was related to the step rate being too high.

10817

EddyCurrent
18-11-2013, 07:08 PM
Sorry for assuming you used Mach3, from my limited knowledge the settings appear to be okay, you've probably been at this longer than me anyway. I would be into trying a different motor and swapping the M542 just to eliminate those from the equation. If you think it's maybe a screening problem why not run a temporary cable between the motor and M542 that is well away from everything else. There was another thread a few weeks ago about rubber mountings and I suggested using heatproof silicone sheet like this SILICONE RUBBER SHEET 200MMSQ, 1,1.5,2,3,4,5,6,8,AND 10MMTHK | eBay (http://www.ebay.co.uk/itm/SILICONE-RUBBER-SHEET-200MMSQ-1-1-5-2-3-4-5-6-8-AND-10MMTHK-/330788411590?pt=LH_DefaultDomain_3&var=&hash=item4d048698c6), but the opinion was to bolt the motors straight onto metal.

birchy
18-11-2013, 08:17 PM
you've probably been at this longer than me anyway.Linux - probably...CNC - almost definitely not...

I'm beginning to think that the resonance is a feature of the cheap setup rather than a fault? I'd be interested to see what Jonathon and John S have to say about this, as I know they use these drives and motors on a regular basis. The motor is bolted to 200Kg+ of metal so I'm surprised there's any noise. I'm thinking I might need to isolate the motor bracket from the main machine using some rubber sheet. Having said that, the issue is with the motor setup rather than something mechanical. Pics for reference:

10818 . 10819 . 10820

birchy
18-11-2013, 08:27 PM
Is my BOB too close to the mains, I wonder? Having said that, I ONLY get resonating within a fairly tight range of low end speeds. What problems are usually caused by not screening cables? My panel is just straight tri-rated panel wires. I've tried to keep the mains and signal cables apart as much as is practicable. Cables connect to the motors via metal cored industrial 20mm Adaptaflex.

Clive S
18-11-2013, 10:14 PM
As a matter of interest are you using ball screws or the original screws? Big credit to the neatness of the panel. ..Clive

EddyCurrent
18-11-2013, 10:25 PM
Is my BOB too close to the mains, I wonder? Having said that, I ONLY get resonating within a fairly tight range of low end speeds. What problems are usually caused by not screening cables? My panel is just straight tri-rated panel wires. I've tried to keep the mains and signal cables apart as much as is practicable. Cables connect to the motors via metal cored industrial 20mm Adaptaflex.

I don't think your BOB is too close to the mains, I used CAT5 4core (2 x twisted pair) networking cable for the connections between the BOB and Stepper Drivers as this eliminates a lot of interference.
The problem you describe does not sound like a wiring problem and is likely to be resonance. I noticed the gap between the motor bracket and the machine was very small in places and I wondered if they were vibrating against each other like a tuning fork ?

birchy
18-11-2013, 10:25 PM
Ballscrews. The resonance is coming from the motor. I think I may have gone OTT with the drive cooling fans though? Here's me drawers...

10827 . 10828 . 10829 . 10830 . 10831

John S
18-11-2013, 10:27 PM
Birchy,
Sorry can't help as never had the problem.
Don't run Linux in case that's the problem. having said that fitted rack loads of the 542's and similar motors.
Are these zapp motors ? Reason i ask is I had some [ not sure where they came from ] where the wiring colours wasn't what it seemed. Having said this bad wiring will be with you across the range, not just in a band.

So far all my conversions have been done with straight 4 core wiring to the steppers, never had an interferance problem to warrant screened cable, yes I know it makes sense but just never had a problem.

Done about five WM16's to date plus a few WM14's

birchy
18-11-2013, 10:29 PM
I don't think your BOB is too close to the mains, I used CAT5 4core (2 x twisted pair) networking cable for the connections between the BOB and Stepper Drivers as this eliminates a lot of interference.
The problem you describe does not sound like a wiring problem and is likely to be resonance. I noticed the gap between the motor bracket and the machine was very small in places and I wondered if they were vibrating against each other like a tuning fork ?
I didn't think it was a wiring issue either, but wasn't sure if the steppers and drives are prone to doing weird things with mains cables in the vicinity, etc. Yes, the clearance are a bit tight in places but the motor is still resonating when I take it off the mounting and sit it on a rolled up rag. The noise is obviously less but the vibration can be clearly felt at certain low speeds.

birchy
18-11-2013, 10:35 PM
Birchy,
Sorry can't help as never had the problem.
....
Are these zapp motors ?
Bugger my luck. Yes, they're Zapp motors. Bought them about 2 or 3 years ago and dragged my arse on the CNC conversion as I was doing manual milling. I'll take a look at the motor wire colours and report back...

birchy
18-11-2013, 10:37 PM
More pictures and a video showing g1 codes at feed rates from 30 upto 2000...

10823 . 10824 . 10825 . 10826



http://youtu.be/OiyVxTNtaSM

JAZZCNC
18-11-2013, 10:55 PM
How have you wired them Parallel or Series.?

If wires are Ok Can you swap with another motor.? If it's so bad it's rocking the machine then I very much doubt it's a resonance problem has often resonance happens at Mid band not low down and just makes motors run slow and sound terrible. It won't rock the machine. Same goes for having wrong micro stepping set, it will cause poor performance and horrible sounding motors but won't make the thing jump around.!

Doubt it's Phase or miss-wire issue either because like John says it would stay with you across the range.?

This may sound obvious but you have checked the machine mechanically first before blaming motors.?

EddyCurrent
18-11-2013, 10:56 PM
Can't see your wiring for certain but looks like it should be

Blue/White - Red A+
Red/White - Blue A-
Green/White - Black B+
Black/White - Green B-

birchy
18-11-2013, 11:04 PM
As I said before, the motor is vibrating when unmounted and has no load, so this is almost definitely a config issue. The machine doesn't actually rock about, however at certain speeds the vibration will catch it "in a sweet spot" and the whole thing sounds like I'm shaking a biscuit tin full of spanners. I'll have to re-check my wiring and try swapping to another drive, then another motor. Will also ask this question on the LinuxCNC forum as it may be a software setup issue...

And before you suggest using Mach, I'd like to point out that I've not used Windows since 2008. Yes, I'm a geek.

EddyCurrent
18-11-2013, 11:15 PM
I worked on a Unix system early 80's and tried Linux a few times since but always went back to DOS then Windows however I recently got a Raspberry Pi so it's back to Linux again.

JAZZCNC
18-11-2013, 11:22 PM
And before you suggest using Mach, I'd like to point out that I've not used Windows since 2008. Yes, I'm a geek.

Wouldn't have anyway has the control won't be doing this. This most defiantly won't be a software Config issue if thats what your meaning by Config issue.?
If you was getting missed steps or positional loss then yes.
Only Config setup thing it could possible be is Wrong micro stepping, after that it's something externally affecting motor/drive. Resonance is the most obvious and if your saying at certain speeds sounds like can of marbles then will be Mid band resonance in which case I'd say you have dodgy drive or motor has those drives are not known for having Mid band problems.
I don't think it will be noise either because again wouldn't do this and mostly would just have missed steps or some other strange happening.

Personally given the fact the motors have been stood 2yrs then I'd start there first and not the drive.

birchy
19-11-2013, 01:51 AM
OK...so I've tried the motor on a different drive - no joy. Then tried a different motor on same drive and still have the problem. I only have 3 motors, all bought at the same time. Will try the 3rd motor tonight but am not holding much hope...

I've played around with various settings (acceleration, max velocity, etc) but had no joy there, either. I then tried changing the gear ratio in software to 6:1 and had no problems as the motor runs faster even at low feed rates. The question is - should the motor be smooth throughout the range or will there be inherent "bad" spots due to the Chinese drives and motors? I'm convinced this is a stepping issue that causes resonance but only at low speeds. If I increase the step rate to 3200, the drive sounds much "cleaner" but I get a "realtime" error in LinuxCNC. What step rate is "normal" for this hardware?

Jonathan
19-11-2013, 10:57 AM
Slides 22 onwards in this document (http://www.mech.utah.edu/~me7960/lectures/Topic14-DesignOfMachineStructures.pdf) give a good explanation of how resonance affects machine frames. If you read that then what is happening here should be obvious. The other lecture notes in that series are well worth reading too.


The question is - should the motor be smooth throughout the range or will there be inherent "bad" spots due to the Chinese drives and motors?

You can't blame this on the Chinese (or in John S's case linux), since the Chinese also sell drivers with resonance damping functions (e.g. AM882). That doesn't mean you need to change drivers though. I'd try testing cutting something at the speeds where it sounds bad and see if it affects the finish noticeably. Also, thoroughly test the motor tuning, e.g. by running random G-code (http://www.mycncuk.com/forums/gcode/5843-random-g-code-testing-motor-tuning.html). If it doesn't stall, goes fast enough and gets an acceptable finish then you can safely ignore the 'problem'.

At what step frequency do you start getting the real time error? Unless you're very unlucky with the computer it should be fine around 25kHz - quite likely more.

John S
19-11-2013, 12:49 PM
You can't blame this on the Chinese (or in John S's case linux),

Hey golden bollocks where did I say it was Linux ?

birchy
19-11-2013, 05:50 PM
Jonathon, I understand the basics of machine structure and resonance and that I may need to add dampers at some point, however the motor itself is noticeably vibrating at very low speeds when it isn't even bolted to anything. The machine castings and framework are obviously amplifying the vibration but I think that will always be the case until the motor is made stable.

Regarding the drives and motors, it wasn't a case of "I blame the Chinese", the question I was asking is could it be a problem due to the low cost setup? At the end of the day, you get what you pay for, etc, etc. I bought the motors, drives, BOB and PSU as a 3-axis kit from Zapp a few years ago. If I was buying again, I'd probably buy the parts individually via eBay or AliExpress but go for better quality. Having said that, there seems to be plenty of others using the same setup and not having any issues.

Jonathon - I believe you use (or have used) LinuxCNC? Have you used it with these motors and drives? If so, maybe you could share a known working .ini file? The LinuxCNC latency/jitter test on my PC settles at 15600ns (with CPU at 100% load), so I have this set at 16000 in stepconf. I'm guessing this may be causing the RTAPI error but don't yet know enough about the system to understand what's going on.

Jonathan
19-11-2013, 06:02 PM
Jonathon, I understand the basics of machine structure and resonance and that I may need to add dampers at some point, however the motor itself is noticeably vibrating at very low speeds when it isn't even bolted to anything.

The motor is entitled to resonate when not connected to anything. The motor is after all a mechanical system, which will have an associated resonant frequency.


Regarding the drives and motors, it wasn't a case of "I blame the Chinese"

I wasn't being entirely serious there!


the question I was asking is could it be a problem due to the low cost setup? At the end of the day, you get what you pay for, etc, etc. I bought the motors, drives, BOB and PSU as a 3-axis kit from Zapp a few years ago. If I was buying again, I'd probably buy the parts individually via eBay or AliExpress but go for better quality. Having said that, there seems to be plenty of others using the same setup and not having any issues.

Better drivers can help suppress the resonant effects induced by stepper motors, as I mentioned earlier. However that doesn't necessarily mean you need better drivers - if the system works, even with the motor vibrating, you might as well just use it.


Jonathon - I believe you use (or have used) LinuxCNC? Have you used it with these motors and drives? If so, maybe you could share a known working .ini file? The LinuxCNC latency/jitter test on my PC settles at 15600ns (with CPU at 100% load), so I have this set at 16000 in stepconf. I'm guessing this may be causing the RTAPI error but don't yet know enough about the system to understand what's going on.

Yes I use LinuxCNC. The motors on my CNC router are the same and the drivers are PM752 (like you, I bought them from Zapp a long time ago). I'm currently not at home, so can't check what settings I use, however that shouldn't be necessary. The value you've set to 16000ns in stepconf is the one which determines the kernel frequency and thus determines if you get the real time error. I'd find out what the minimum you can set this number to is without the real time error occurring, then add a bit just to be sure.

birchy
19-11-2013, 07:06 PM
Thanks Jonathon. I would of gone for the 70V setup if I was buying again, but this 42V setup *should* be OK. I'm going to re-run the latency test as I have a wireless USB keyboard connected now (which I didn't have previously), so that may be causing issues. Attached are screenshots of my stepconf. Do these numbers look OK to you? I'm not sure about the timing settings of 5000 as I can't find any actual values in the drive manual. There are a few "542" type drives on this page, but they all have drastically different timings: LinuxCNC Documentation Wiki: Stepper Drive Timing (http://wiki.linuxcnc.org/cgi-bin/wiki.pl?Stepper_Drive_Timing). Again, I don't have enough info to get these set correctly... :disturbed:

Out of interest, what step rate do you guys normally set your 542/752 drives at? 800, 1600 or some higher value? Don't forget I'm a newbie, so am just wondering what range of settings are considered "normal".

10839 . 10838 . 10840 . 10841

Jonathan
19-11-2013, 07:53 PM
Do these numbers look OK to you? I'm not sure about the timing settings of 5000 as I can't find any actual values in the drive manual.

Yes, 5000 works for me. You should be able to use much higher acceleration than 50mm/s^2. Around 500mm/s^2 is probably a good starting point.

Generally 'guys' normally don't use any finer than 1/8th microstepping, so 1600 step/rev. Search 'microstepping' on the forum and you'll find plenty of explanation...

routercnc
19-11-2013, 09:08 PM
Hi Birchy,

Not tried it myself as don't have the noise problem you are experiencing but others have tried adding inertia to the stepper via flywheels etc. and achieved a much smoother low speed running. If you have a shaft from the rear of the stepper you can add one there, or otherwise bolt it to the front face of the pulley (tap holes in pulley face).

Solving Stepper Motor Resonance Problem on CNC Lathe - YouTube (http://www.youtube.com/watch?v=mqCL7g4gwAU)

I would suggest adding the lowest inertia possible to just remove the problem, otherwise acceleration of the stepper will suffer too much. The one in the video looks a bit big - you could do some calcs, or just try something to at least see if that would help before refining the flywheel design iteratively.

birchy
19-11-2013, 09:40 PM
What about maximum velocity, Jonathon? Is 50 too high? Too low? Or somewhere near?

Interesting use of the counterweight there. I was thinking that I need to add a wheel of some sort so I can manually turn the motor if one of the E-Stop limit switches is activated. Two birds, one stone...

EddyCurrent
19-11-2013, 09:46 PM
This is probably nothing to do with it but I'll mention it anyway.
When I first powered up my system, 4 x AM882 and CNC4YOU Nema23 4Nm motors, it sounded like a load of whales singing from one particular driver and the motor was noisy when running. I thought it was goosed so I tried another motor on it with the same result. There is a utility with the AM882's called Protune and I had been playing with this connected to one of the drivers via a special cable (connections available on Leadshine website to make one). I therefore connected the noisy driver to Protune and ran the Auto Current Tuning section, immediately no more whale song and motor nice and quite when running. Without looking at the manual for your drivers I don't know if this utility is relevant but it shows that something on the current side was having a huge effect on motor noise.

birchy
19-11-2013, 09:46 PM
Just typed "stepper resonance" into Youtube and came up with this. Interesting that they have an almost identical machine to mine (looks like the Amadeal version) and to quote "I was so FED UP of stepper motor resonance". So he fitted Servo motors instead. Seems a bit OTT though.


http://www.youtube.com/watch?v=y7gOPmVlh1c

JAZZCNC
19-11-2013, 10:25 PM
This is probably nothing to do with it but I'll mention it anyway.
When I first powered up my system, 4 x AM882 and CNC4YOU Nema23 4Nm motors, it sounded like a load of whales singing from one particular driver and the motor was noisy when running. I thought it was goosed so I tried another motor on it with the same result. There is a utility with the AM882's called Protune and I had been playing with this connected to one of the drivers via a special cable (connections available on Leadshine website to make one). I therefore connected the noisy driver to Protune and ran the Auto Current Tuning section, immediately no more whale song and motor nice and quite when running. Without looking at the manual for your drivers I don't know if this utility is relevant but it shows that something on the current side was having a huge effect on motor noise.

Nice feeling isn't it Eddy knowing you can deal with almost anything the machine throws at your drives/motors. This alone justifies the extra cost IMO not to mention the far smoother motor operation than analogue drives.

Birchy if you do end up buying new drives then the AM882 is the drive if your having Resonance problems because you can adjust the Resonance range the drive covers with Pro tune. Your Analogue drives are stuck to set range which is narrow and if indeed it is resonance then your drives Resonance algorithms are not looking in that range or even working at all.?

The Dampers you see are usually only affective on cheap drives that don't have any resonance handling capability and I'm sure the M542 does have some basic capability and although it may only be basic it's better than none.?

Regards Micro stepping then above 2000 is generally pointless has the motors can't resolve much more than that and your only stressing the machine and increasing chance of dropped pulses/steps etc.
Higher MS does help with smoother motors and I generally prefer 2000 if the PC/Parallel port can handle it but anywhere above 800 generally works ok.

birchy
21-11-2013, 07:45 PM
So now I'm more edumacated on steppers and drives, it's obvious that there will be resonance somewhere in the speed range unless using tunable digital drives. At 3x the cost of an analogue drive, I'm not so convinced that the digitals are worth the extra cost at the moment. No doubt they will become more reasonably priced when more people adopt them.

Since my last post, I have refitted the Y-axis motor and pulleys, belts and X table (with a 25Kg lump on top) so that I'm testing the motor under normal conditions. I then created a g-code file to test all feed rates from 0 to maximum velocity, increasing in 5% steps. This helped me isolate the speeds/frequencies at which resonance was most noticeable. Starting at 2 microsteps (1/2 stepping), I ran the test file for each setting up to 64 microsteps. The resonance band was tightest and least noticeable at 8 microsteps, so that appears to be the optimal setting for my machine. I also gave the machine a good hammering with randomly generated G0 codes (thanks Jonathon) at full speed and acceleration of 2000. The machine danced around a little due to the high inertia changes but the main thing is that it didn't lose any positional steps. I've now put the acceleration to a more sensible 250 to prolong machine life.

Now I need to stop farting about and get on with finishing the Y-axis before contemplating the X and Z...

John S
21-11-2013, 11:12 PM
Back in the early days when we were having loads and loads of problems with Gecko 210 drivers there was a problem using 10 microsteps.
Marris explained it at the time and it made sense but can't remember what it was.

When we moved onto the Chinese drives we chose 8 microsteps as the defacto default and have stuck with this over the years.

Nice to see some tests proove what we discovered then.

EddyCurrent
22-11-2013, 07:16 PM
There's this light reading which I found educational

Jones on Stepping Motors (http://homepage.cs.uiowa.edu/~jones/step/)