Electrical Noise Causing Axis Jumps

More
13 Mar 2023 13:52 #266553 by jlennox
Hi Guys, thought'd i'd post my issue here so some of you gurus could possibly give me ideas to solve my new problem.

Machine is a Mazak VQC 20/40B retrofitted with a 7i92, 7i77 & 7i84. Machine has been working and ran its first parts perfectly fine a couple of days ago. Went to do a test on the weekend and the machine started jumping on each axis and had a complete runaway moment on the z axis. Performing some breif debugging shows this issue only occurs when the spindle is running and the issue disappears with spindle off.

The only difference from the first parts I ran to now was the position of the head in the z axis. Head was close to top of travel for successful parts and low in the z travel when this issue occured. I have attached photos of halscope showing what the x axis encoder velocity looks like without the spindle running and a capture of the x axis jumping (spike in encoder velocity, jogging at 319mm/min) with the spindle running. I get the same issue on all axes repeatably with the spindle running. Spindle off everything perfect. Another issue I now have is even when I run the spindle and do not jog the machine, I get random estops in gmoccapy for no reason (see attached picture) which has to be noise causing this.

I would also like to point out, to see if it is a possible contributor is, I have shielded twisted pairs (earthed at contol side) coming from the encoders on each axis to my custom breakout boards, which then go to the 7i77 through CAT5 cable (see attached pictures). Is it possible noise is being introduced into the encoder feedback through this unshielded cable despite being a short run?

Any ideas to test for what is causing this would be appreciated!

Please Log in or Create an account to join the conversation.

More
13 Mar 2023 15:26 #266558 by andypugh
As a first step get an input power filter for the VFD.
Look on eBay for the brand "Rasmi" and get one rated for the voltage / current / number of phases of your spindle VFD.
There are other manufacturers, but Rasmi is a good search term. They are not particularly expensive. For example www.ebay.com/itm/304208599566.

It might seem odd to be filtering the _input_ to the VFD but it does seem to work.
The following user(s) said Thank You: jlennox

Please Log in or Create an account to join the conversation.

More
13 Mar 2023 15:27 #266559 by andypugh
Also: Where is the VFD mounted? Maybe consider putting it outside the control box if it is inside.

Please Log in or Create an account to join the conversation.

More
13 Mar 2023 16:03 #266565 by tommylight
Are the encoder wires shielded and grounded properly?
Or did you use the existing resolver wires?

Please Log in or Create an account to join the conversation.

More
13 Mar 2023 22:18 #266611 by rmu
Are you sure noise isn't affecting the command voltage (I assume +-10V)? encoders should be relatively robust, i think it would be strange to count multiple ticks into wrong direction and back again. To me it looks more like axis jumps (because of noise on command) and PID tries to correct.
The following user(s) said Thank You: jlennox

Please Log in or Create an account to join the conversation.

More
14 Mar 2023 12:42 #266656 by jlennox
Thanks for the responses guys.

Andy, the spindle dive is mounted inside the main cabinet with the rest of the drives. Thanks for the link, i'll be sure to scope them out and I must say they are very reasonably priced. As the machine never had a filter on the spindle drive, I would like to solve the problem without adding additonal hardware but if all else fails, thats what i'll be putting in.

Tommy, I resued the original resolver wires as my encoders are only AB quadrature. My machine manual calls out shielded twisted pairs for the resolver cables hence why I reused them but admittedly I havent actually split apart one or the MR connectors to check the cables actually ar shielded and grounded. The shiled/case gound pin is certainly grouded as I check this with a multimeter.

RMU, I feel like you are more on the money here. The axis jumps at a repeatable frequency (consistent timing between jumps) while jogging with spindle on and yes would be strange for the encoder to jump multiple pulses in one direction.

As I get random estops when the spindle runs I suspect I am getting noise on the 10v signal lines as well as my inputs. Would the following scenarios be possible:
- Issue with the 24v power supply generating noise as a result of noise on the AC line due to spindle drive?
-  My 0-10v analog signals wires are just single core cables I twisted as pairs (see wires labeled ER(R)1, ER(R)2, etc in first post), despite the short run, could noise be introduced on them and am I better off swapping to shielded twisted cable? This wouldnt explain random estops as those would have to be from noise on inputs.
- Failing/deteriorated transistors in the spindle drive causing the noise? This wasnt an issue before but now suddenly is so wondering if the drive is failing somwhere.
 
 

Please Log in or Create an account to join the conversation.

More
14 Mar 2023 13:00 #266658 by tommylight
I am pretty sure on your picture those are UTP cables on the encoders, not the original ones.
Use STP of SFTP cable and wire the shield to ground only on the Mesa side, leave encoder side unconnected.
The following user(s) said Thank You: jlennox

Please Log in or Create an account to join the conversation.

More
14 Mar 2023 13:26 #266660 by rmu

As I get random estops when the spindle runs I suspect I am getting noise on the 10v signal lines as well as my inputs. Would the following scenarios be possible:
- Issue with the 24v power supply generating noise as a result of noise on the AC line due to spindle drive?
-  My 0-10v analog signals wires are just single core cables I twisted as pairs (see wires labeled ER(R)1, ER(R)2, etc in first post), despite the short run, could noise be introduced on them and am I better off swapping to shielded twisted cable? This wouldnt explain random estops as those would have to be from noise on inputs.
 

Why and how do you get random E-STOPs? I supposed the machine stopped because it thought an axis "jumped" outside of machine limits, but that seems not to be the case. If you have electrical issues with ESTOP triggering via noise, fix that first. E-STOP usually is a loop of "N"ormally "C"losed switches that open when actuated. This way broken or faulty ESTOP chain won't let you activate the machine. Hard to imagine you can influence that with induced noise?
Check for loose connections and if everything is grounded properly. Differential signals also need ground (usually).
If you can use an oscilloscope, check signals with spindle on/off, but be careful to connect probe gnd very near where you pick up the signal. 

Please Log in or Create an account to join the conversation.

More
14 Mar 2023 16:32 #266672 by tommylight

-  My 0-10v analog signals wires are just single core cables I twisted as pairs (see wires labeled ER(R)1, ER(R)2, etc in first post), despite the short run, could noise be introduced on them and am I better off swapping to shielded twisted cable?

 

Also must be shielded, but if this was the issue, the machine would jump a lot, so i do not think this is a problem for now.

As I get random estops when the spindle runs I suspect I am getting noise on the 10v signal lines as well as my inputs.  

This for sure means something is very wrong with grounding/shielding on the wiring.
Start from the beginning:
Does the shop have proper grounding?
Does the machine have grounding wired properly?
Does the shop have a FID or GFR of whatever the name of the thing that cuts the power to the shop if anything goes wrong with the grounding?
These are a must, there is no way around it.
And just in case, change the power supply for Mesa, a while back had a very noisy one.
The following user(s) said Thank You: jlennox

Please Log in or Create an account to join the conversation.

More
14 Mar 2023 19:49 #266690 by andypugh

Why and how do you get random E-STOPs? 

Generally through noise on a limit switch input. 

I generally debounce the limit switches, as nuisance tripping can be expensive, whereas a 10ms delay in tripping the limit switches probably doesn't matter that much. 
 
The following user(s) said Thank You: jlennox

Please Log in or Create an account to join the conversation.

Moderators: cncbasher
Time to create page: 0.118 seconds
Powered by Kunena Forum