Advanced Search

Hi everyone,
I seek help in commissioning a Motor with a Beckhoff Ax5000 series drive within TwinCAT to then transfer that tested setup to linuxcnc-ethercat. I cannot for the life of me get that motor to run over.
Setup:
Motor: HarmonicDrive LynxDrive-20C-100-MEE-AO (reduction 100, MEE Encoder (Heidenhain EQN-1125 Endat 2.2/2.1), motor values are known, commutation unknown)
Driver: Beckhoff AX5206-0000-0203 Fw 1.06
Goal: Be bale to move the drive from NC in Twincat to check for functionality of the setup
What did I try:
- Scan Bus in TwinCat to find devices
- Found the Driver
- Setup of power management, safety (off), set mode for channel 1 to position control
- Setup Channel 1: X11 for encoder connector, selected Heidenhain EQN1125 from list, manually setup the motor with relevant values (pdf: harmonicdrive.de/fileadmin/user_upload/E...8856_08_2018_V04.pdf, page 17)
    When I try to scan for the encoder it identifies as Endat 2.2 without a nameplate. However, when selecting an Endat2.2 from the list of encoders, it does not work. (0xF148: Build drive system: unknown feedback system)
    Question: I assumed that this table takes the values for the motor and not for the output shaft (which is behind the harmonic drive's reduction), correct?
- Setup of scaling
     Question:  Here I am not sure. To me it makes the most sense to set this up for the motor shaft (as this is where the encoder is). I.e. 360°/motor revolution (see screenshot) - correct?
 
Next would be to setup the commutation setup (mechanical and electrical offset).
This is where I struggle.  The manual (www.beckhoff.com/en-en/download/61857533) states two different chapters regarding commutation:

1) Chapter 9.1.4 Third-party Motors
Step one asks to ensure motor rotation: The output shaft is connected to the HD. Given a reduction of 100 it is really difficult to backdrive the thing. Question: Am I correct in assuming that the HD reverses the direction of rotation? that is: if the servo motor moves CW, the output moves CCW?
Next step is the electrical commutation offset determination.
I tried to follow page 168ff using P-0-0166:
I get values after starting the command, with equal directions being "No". Changing U and V accordingly changes direction of the motor, but has no effect. I cannot get a constant value. I keep chasing my tail around.
Note: At this point P-0-0150 Commutation Node is set to 3; Adjustable commutation offset (mechanical) is 0 deg

I then followed chapter
Manual 9.6 list commutation methods (p 201ff)
Accordinglx, P-0-0160 is executed first
Regardless of the actual value of P-0-00578 (Electrical commutation offset) I get a value for the mechanical commutation offset which I enter in P-0-0150 above.
Doing so, an flipping the polarity of UV, I then get the following result for P-0-0166: Equal Directions: yes; Offset: 357.55 deg
 
I do not understand what way is the correct way. Accordingly, I am not sure if the offset is correct.
Now I undertand I should be able to move th eaxis using NC-A: Online (image 3)
 
I can enable the drive and it does start to make light noises and one can feel light vibration on the output shaft.
However, when I send a command using F2/F3 I get an increase in following error and nothing moves. (Image 4).
 
Parameters of the axis enc are as indicated by image 5.
 
I assume something in scaling is borked and thus the PID cant work properly. But I cannot find anything that is missing.

- I have checked (to the best of my ability) the polarity and pins for both the power and the encoder connector. At this point I assume that these are correct.
- I have manually pulsed the motor with a low dc voltage, It then performs the expected jump. The encoder provides a new value afterwards as well.
I am open for any and all advice. It must be sooo close.
Thank you!
 

I think your motor encoder feedback needs to go into your motor driver while the glass scale quadrature signals into linux CNC PID module.

Looks OK

You can also simply delete the num_xxxx=n stanzas in the hal file
and all hardware in a particular configuration will be enabled.

Hello PCW,
I managed to get rid of the error messages by changing these values in the hal file:
num-pwmgens = 2 (from 1) and
num-stepgens = 4 (from 5)

Now I also can see pwmgen.01 in the halcmd show all *pwmgen*:

roland@linuxCNC:~$ halcmd show all *pwmgen*
Loaded HAL Components:
ID Type Name PID State

Component Pins:
Owner Type Dir Value Name
36 bit IN FALSE hm2_7i96s.0.pwmgen.00.enable <== spindle-enable
36 float IN 0 hm2_7i96s.0.pwmgen.00.value <== spindle-vel-cmd-rpm
36 bit IN FALSE hm2_7i96s.0.pwmgen.01.enable
36 float IN 0 hm2_7i96s.0.pwmgen.01.value

Pin Aliases:
Alias Original Name

Signals:
Type Value Name (linked to)

Parameters:
Owner Type Dir Value Name
36 bit RW FALSE hm2_7i96s.0.pwmgen.00.offset-mode
36 s32 RW 1 hm2_7i96s.0.pwmgen.00.output-type
36 float RW 18000 hm2_7i96s.0.pwmgen.00.scale
36 bit RW FALSE hm2_7i96s.0.pwmgen.01.offset-mode
36 bit RW TRUE hm2_7i96s.0.pwmgen.01.out0.invert_output
36 bit RW FALSE hm2_7i96s.0.pwmgen.01.out0.is_opendrain
36 s32 RW 1 hm2_7i96s.0.pwmgen.01.output-type
36 float RW 5000 hm2_7i96s.0.pwmgen.01.scale
36 u32 RW 0x005B8D80 hm2_7i96s.0.pwmgen.pdm_frequency
36 u32 RW 0x00004E20 hm2_7i96s.0.pwmgen.pwm_frequency

Parameter Aliases:
Alias Original Name
hm2_7i96s.0.pwmgen.01.out0.invert_output hm2_7i96s.0.gpio.025.invert_output
hm2_7i96s.0.pwmgen.01.out0.is_opendrain hm2_7i96s.0.gpio.025.is_opendrain


is this correct ?
Tahnks a lot and good night

Alright, I'm getting really close now (I think), but I'm to a point where I'm not 100% sure what the deal is. Unfortunately, the pin I needed to use for the QEI encoder module happens to be a bad pin on my SKR 1.4 Turbo apparently. I can't get it to pull high. I may have messed it up years ago trying to get neopixel LEDs working, I don't know. I set up a new pin as a digital input on the Y limit connector (Pin 1.28) and simply mapped the index pulse to that pin in my hal file. I'm seeing what appears to be a solid pulse in halscope. I'm also seeing it trigger on the phaseZ and index-enable pins. However, every time trying to do a threading operation, my program hangs at what appears to be waiting on the index pulse. I know the documentation states that the pins are pre-assigned and can't be changed, but I figured the index was handled differently since it included the remora.input.N is mapped in hal, but maybe something on the backend with the way the QEI encoder is handled. I apologize in advance if my HAL and INI look messy but everything else seems to work. I had to set my servo thread to 2000000 to get around the unexpected real-time delay error, but my latency in histogram --nobase seems great with isolcpus 1,2,3 on the Pi 4. I did try running the program with the servo period set to 1000000 in my INI just to see if that had something to do with a timing issue when registering the index-enable true when Linuxcnc expects to see it, but, it didn't work. Just about everything else seems to be working. Anyways here's my HAL again, my INI, and my config.txt

Did you enable 2 PWMGens in the hal file?

Success.  I went with LMDE6 and the install went much smoother.  Upgraded to 2.10 and can run a qtplasmac sim. 

Hello PCW,
I flashed my 7i96s with 7i96s_2pwmd.bin and put this line to my custom.hal:

# invert pwm output
setp hm2_7i96s.0.pwmgen.01.out0.invert_output true.

When I start linuxcnc i get the error message
"parameter hm2_7i96s.0.pwmgen.01.out0.invert_output not found"

The hmid-output looks correct:
TB2 2,3 25 IOPort PWM 1 PWM (Out)
TB2 4.5 26 IOPort None

Can you post a halscope plot of: 

1 Joint velocity
2. Joint following error

it's a static error. But when I put “I” > 1 I get a joint tracking error.

Put 1 resistor on each cable, much safer.
www.mouser.de/ProductDetail/Vishay-Dale/...rmWJGO8s8TXMfjOoM%3D

Thanks Ben, I tried your suggestion but the behaviour is the same. Works in Velocity mode but not in Step mode.
I spent the morning looking at signals with halmeter but I cannot figure this out.

Thanks for taking your time to look into this.

Never mind, I found the issue. It was my fault I mis-wired one of the pins on the bob.Great learning experience with the LinuxCNC software digging through the particulars! You guys pointed me in the right direction, much appreciated!

Hi PCW:

Thanks for your reply.
When you mentioned: "So if you don't have dual homing switches you need to setup immediate homing"...can you tell me more about that please?

Thanks for your help and support! I set up my parallel port in the step config just as your image above. I have my motors and bob set up on my desk as a test set up to work out the set up without injury to my machine. In a real world application Id need to be able to jog my Y axis (both motors driven at same time) tgether at once. Would I need to set up something in one of the hal or ini files?

Thanks again for your help.

Thanks.

So I could open the Plasma source, cut the cable behind the THC-plug and solder 2 x 1MegaOhm in series of the negative cable and put a shrink tubing over the 2 resistors ?

Is there an alternative to the 1MegaOhm-resistor ? I ask because in the mesanet-store it shows me over 200$ shipping-cost for the 3 resistors and there is an error when the website calculates shipping costs to germany.