Remora - ethernet NVEM cnc board

More
26 Sep 2022 20:54 #252801 by scotta

Hey there,

my attempts to reproduce the NVEM-linuxcnc configuration ends up with an error message, saying there would be a positioning error for the motors.
Error message in the terminal says "emc/task/taskintf.cc 942 : error on joint 0" followed by the command number.

I've been looking through the HAL-files but couldn't find a thing.
The board gets detected and i can see the different I/O-pins. Also i am able to observe any inputsignals trough this pins.

My Linuxcnc is running on an Desktop in version 2.8.4. [Debian 10 (New installation)] all configurations for the board and L-cnc made like in Scott's video.

Any idea how to solve this ?  


 

Hi, did you have the board running with the example config included? It's known to work and the best place to start to ensure that all the hardware is talking correctly. I'd then move on to your own HAL and INI config.

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

More
26 Sep 2022 21:00 #252802 by scotta

@Scotta
I am using Remora for a bench prototype with the NVEM v2 board with 6 axis and 3 motors. I am using the 1.0.0 release.

I discovered the following strange behavior of the driver card:
0. Start LinuxCNC
1. Enable drivers
2. Jog a little bit
3. Power off card (disconnect 24V)
4. Power on card
5. Unset EStop
Now the commanded position and the position feedback from Remora differs (about 0.1 - 0.05mm).
This is not corrected, because the EStop is active.

Now, if I enable the drivers, the motor moves a little bit. LinuxCNC corrects the position error.

What could be the reasons for this? Why does the position feedback change after power on of the card?
Are maybe the PID values wrong?

I hope I described the problem to be understandable (

Not a strange behaviour, the firmware does not keep / store the DDS position accumulator values during a power off. The LinuxCNC component adds the difference between two position reads (incremental) to the last calculated position value. So effectively you are resetting the next position count back to zero.
The following user(s) said Thank You: hydroid7

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

More
27 Sep 2022 07:15 - 27 Sep 2022 11:15 #252831 by Capper
yes, i am using the NVEM-basic-configuration from youre github. As you mentioned, my plan was to take this as an example for further adjustments.

That error occures every time the software tryes to move the Motors (not only while jogging). After that error i have to reinitiate the machine.

By turning down the feedspeed to ~0.9mm/min, i could observe that the pos_cmd seems to count up until it reaches the FERROR-value from the config.

My setup constist out of the NVEM with one Stepperdriver (DM422 without feedback) attached to the x-axis. Endstops and other stuff are detached.
Instead of an raspberry i am running an Desktop-PC with the Linuxcnc-Debian10 (Buster) version you can find here in the forum.

could it be possible, that the communication between NVEM and linuxcnc has any problems? I had to set the static IP via "/etc/network/interfaces" instead of the "etc/dhcpcd.conf", like you've done in the video.

if there are any logs needed, just let me know.

Edit:
 

This board is about 4 Years old btw. There are clearly differences in the layout between yours and mine.

Don't think this really matters but who knows.

Also, i have to withdraw my statement, where I've said the inputs can be observed. Checked it and it's not the case.

So, after uploading the Basic-configuration.txt (successfully), with x,y,z Motors and  and several Inputpins (Input 3 & 4)
i am not able to get any feedback at all.
Attachments:
Last edit: 27 Sep 2022 11:15 by Capper. Reason: Update

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

More
27 Sep 2022 20:39 #252890 by scotta
Hmm, if you have Ethernet comms then it has to come down to the config not being loaded on the board. The only way to see what is going on is to connect the serial output. If you've got an RS232 to USB adaptor connect TX-RX RX-TX at 8N1 115200 baud.

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

More
28 Sep 2022 18:34 #252962 by hydroid7
I have the same board with the same setup as you described and it works ok.

Make sure that your PC is directly connected to the board.
Try to ping your board. For me the avg ping time is ~0.210ms.
If your ping time is more, there is an issue with the connection. Otherwise I'd look at the configuration.

For me, this error occurred when the axis scale was set too high. Try to reduce a little bit.

Did you try to repower the device after uploading a configuration?

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

More
28 Sep 2022 18:52 #252964 by hydroid7

@Scotta
I am using Remora for a bench prototype with the NVEM v2 board with 6 axis and 3 motors. I am using the 1.0.0 release.

I discovered the following strange behavior of the driver card:
0. Start LinuxCNC
1. Enable drivers
2. Jog a little bit
3. Power off card (disconnect 24V)
4. Power on card
5. Unset EStop
Now the commanded position and the position feedback from Remora differs (about 0.1 - 0.05mm).
This is not corrected, because the EStop is active.

Now, if I enable the drivers, the motor moves a little bit. LinuxCNC corrects the position error.

What could be the reasons for this? Why does the position feedback change after power on of the card?
Are maybe the PID values wrong?

I hope I described the problem to be understandable (

 

Thank you for the clarification. It would be great if this would not happen.
Could it be a solution to set the accumulator after connecting in the LinuxCNC component (without firmware update)?

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

More
29 Sep 2022 08:18 - 30 Sep 2022 03:34 #252999 by MX_Master
New NVEM v2.1 (imxrt1052) is arrived to me. My stlink clone was successfully converted to the DAPLink and works well with all of my STMs. But I can't read/write anything to the RT1052 (using MCUExpressoIDE, Platformio (vscode), mfgtools). Need help of a professionals :) 
Last edit: 30 Sep 2022 03:34 by MX_Master.

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

More
30 Sep 2022 11:51 #253089 by MX_Master
Just tried to convert ST-Link v2 clone to J-Link (using unofficial methods). No luck. Will try a Black Magic Probe firmware.

Of course, we can use a 50 (and more) bucks official debugger/flash tool to upload the firmware, but it's not reasonable for general user. Controller cost is about 100 bucks.

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

More
30 Sep 2022 22:03 #253125 by scotta

yes, i am using the NVEM-basic-configuration from youre github. As you mentioned, my plan was to take this as an example for further adjustments.

That error occures every time the software tryes to move the Motors (not only while jogging). After that error i have to reinitiate the machine.

By turning down the feedspeed to ~0.9mm/min, i could observe that the pos_cmd seems to count up until it reaches the FERROR-value from the config.

My setup constist out of the NVEM with one Stepperdriver (DM422 without feedback) attached to the x-axis. Endstops and other stuff are detached.
Instead of an raspberry i am running an Desktop-PC with the Linuxcnc-Debian10 (Buster) version you can find here in the forum.

could it be possible, that the communication between NVEM and linuxcnc has any problems? I had to set the static IP via "/etc/network/interfaces" instead of the "etc/dhcpcd.conf", like you've done in the video.

if there are any logs needed, just let me know.

Edit:


This board is about 4 Years old btw. There are clearly differences in the layout between yours and mine.

Don't think this really matters but who knows.

Also, i have to withdraw my statement, where I've said the inputs can be observed. Checked it and it's not the case.

So, after uploading the Basic-configuration.txt (successfully), with x,y,z Motors and  and several Inputpins (Input 3 & 4)
i am not able to get any feedback at all.

Hi, I've been thinking on your problem. Please do a full chip erase and re-flash the firmware and DO NOT upload a config file. This will ensure that the default 3 axis step generators are loaded. If this does not allow movement (feedback to LinuxCNC) then there is something else like Ethernet going wrong.

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

More
30 Sep 2022 22:08 #253127 by scotta

Just tried to convert ST-Link v2 clone to J-Link (using unofficial methods). No luck. Will try a Black Magic Probe firmware.

Of course, we can use a 50 (and more) bucks official debugger/flash tool to upload the firmware, but it's not reasonable for general user. Controller cost is about 100 bucks.

MX_Master, I agree and this will be the challenge for these boards. Having an affordable reprogramming method will be key. I did order a J-OB-V2 from Aliexpress as it apparently can work with an Cortex-M7. It had not shipped after a long time so I've just cancelled the order and ordered from a different seller.

The other approach I'm going to try over the coming days is the DAPlink. I've been researching this week and hopefully we can create a DAPlink with the same USB vendor and product ID as the DAPlink that is on the RT10502 evaluation boards. This would, hopefully, allow the use of the software tools available for those boards, which included a flash loading utility.

How did you determine which pins on the NVEM v2.1 were SWDIO and SWCLK? 

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

Time to create page: 1.104 seconds
Powered by Kunena Forum