Cumark Closed-Loop VFD with EtherCAT (non-CiA402) Configuration

Hello EtherCAT gurus,

After generous help from OncedLoved and AI we have reached the end of our troubleshooting patience trying to resolve the strange behavior of my new closed loop VDF (Cumark 15 KW) connected to LinuxCNC via EtherCAT (non- CiA402).

Current Condition:

• The VFD is properly set up for the spindle (motor parameters and encoder) and it runs perfectly using the keypad.
• The drive successfully enters OP mode when LinuXCNC starts up and the PDOs are displayed in halshow (see attached screenshot and xml)
• In halshow, when I set Ref 1 (motor speed) to 500 and ControlWord to 2178 (decimal for hex 0x0882 = Start Command), the spindle instantly runs to 500rpm as expected.
• After a few seconds the motor speed changes to 2178 rpm! As if the motor reference value is the decimal start command value. This can happen from 10-90 seconds after the spindle is turned on.
• Then I set ControlWord to 2177 (decimal for hex 0x0881 = Stop Command), and the spindle unreliably either:
  • Turns off after a delay of 30 to 120 seconds, or
  • the motor speed changes to 2177 rpm after a similar delay range, and then turn off after and additional delay range.
• As a result, the spindle eventually turns off but not instantly.
• Actual values Act1 and Act2 are either displaying the wrong value or there is also a delay.
• This strange behaviour happens when controlling the VFD from TwinCAT as well. So I am guessing that something in the drive is not setup correctly.

Questions:

1. What is causing this inconsistent delay behaviour?
2. Anyone experienced in custom EtherCAT configurations (non-CiA402)?
3. How do I convert decimal values transmitted from the VFD  to legible Drive Status Word: (vfd manual page 59 section 6.05)

Attached is:

1. ethercat-conf.xml (just testing the VFD - slave 0)
2. generic ini and hal files for testing
3. spreadsheet of VFD ethercat relating settings and PDO mapping
4. Link to vdf and ethercat card pdf manuals, and xml from Cumark.
5. screenshot of the lcec hal pins when spindle off and on
6. Pictures of the machine frame build and spindle for your interest

It would be great to get this running and I am sure others would find this useful otherwise I would switch for Modbus or analog (not ideal).

Your expertise in this nuanced topic would be greatly appreciated!


Background about the Build in case you are interested:
CNC milling machining for my business in designing and building scientific instrumentation. The frame is made out mineral casting by Muller (Germany). The travel is 600mm X x 500mm Y x 350mm Z. It has 23 23-bit absolute encoders, AC Servos driven by EL8 Leadshine drives. It has a DECI 24K rpm, 15 kw continuous, 18 kw peak HSK 63f liquid cooled direct drive spindle controlled by a CUMARK closed loop VFD hopefully via ethercat.

One need a vfd expert rather than an EtherCat expert I would say.
You say you get the same results using TwinCAT, that's good since it rules out linuxcnc being the problem.
Unless you get some good tips here you should probably go to the seller/manufacturer and ask them to provide details to make the vfd work.
But the behavior of the vfd doesn't seem correct.

Anyway, I looked at the config and manual and it may perhaps be that the start command should only be given a short time, really only one cycle. So give start 882 and immediately after 880. Same for stop: 881 followed by 880. Just a possibility, you must test to verify. If that turns out to be the case you can use the "oneshot" hal component.

Not sure what you mean regarding 06.05. 06.05 is the controlword you send to the vfd, right?


 

Thank you very much, Hakan.

You are right to suspect it's an issue with the drive, and I am trying to get as much information as possible from the supplier. It's been a little tricky because the salesperson has to relay all the info to the VFD engineers in Chinese. The lesson for next time is to make sure the drive is CiA402 compatible.

It's hard to judge from the test results whether the 880 after start and stop makes a difference. Either way, I continue to experience inconsistent and prolonged delays. 
Regarding the 6.05 section, I was trying to decode the decimal values of the control and status words used in Halrun and TwinCAT. I have now updated the PDO spreadsheet to include the decoding based on the bit tables in the manuals. 

Any ideas what I should try next, if I don't have any luck with the VFD manufacturer?