gokhan
Junior Member
Posts: 89
|
Post by gokhan on Dec 12, 2016 18:57:20 GMT
I was experimenting with PID settings my platform. I changed P from 20 to 15 and saved. I change one setting at a time in order to see the result clearly. After a while platform started moving violently and i had to emergency stop it. I found out that all PID settings were changed. P became 39(should be 15), I became 40(should be 0) and D became 164(should be 0). Sensor directions was also changed. As a result some actuator turned 360 degrees. Fortunately nothing was broken. I had to reset the values few times because those values were coming back again(most likely i couldn't save them properly). Finally i succeeded to set them back to the values intended. After a complete hard reboot, now it seems like working normally. I want to find out what went wrong, what i did wrong to avoid repeating this disaster. Any ideas? I recorded ta video of the settings as i discover them. youtu.be/Kzzx0tZbcWg
|
|
|
Post by tronicgr on Dec 12, 2016 23:26:49 GMT
I'm working on a new firmware that will fix this. It's serial garbage most likely. Can you try with another USB cable?
|
|
gokhan
Junior Member
Posts: 89
|
Post by gokhan on Dec 13, 2016 7:07:46 GMT
I'm working on a new firmware that will fix this. It's serial garbage most likely. Can you try with another USB cable? I was using a 5m usb cable when this happened. Do you think this may be it? If this is the reason I will be relieved. Because i can use a shorter cable and locate the control pc away from the inverters. I already have problem with 5m cable on yoke connections so now i am using 3m cable with the yoke.
|
|
|
Post by tronicgr on Dec 13, 2016 11:36:11 GMT
I'm working on a new firmware that will fix this. It's serial garbage most likely. Can you try with another USB cable? I was using a 5m usb cable when this happened. Do you think this may be it? If this is the reason I will be relieved. Because i can use a shorter cable and locate the control pc away from the inverters. I already have problem with 5m cable on yoke connections so now i am using 3m cable with the yoke. 5m usb cable it too long for many usb devices, especially for usb 2.0. I never tried longer than 1.8m long on the amc1280usb and usually have a 30cm for my tests. The longer the cable the more voltage drop you get on the USB data lines that could lead to garbage in the data. Try shorter usb cable, possibly a better quality too (thicker with shielding) in case the motors frequency affects it as well. Thanks Thanos
|
|
gokhan
Junior Member
Posts: 89
|
Post by gokhan on Dec 13, 2016 14:23:31 GMT
I was using a 5m usb cable when this happened. Do you think this may be it? If this is the reason I will be relieved. Because i can use a shorter cable and locate the control pc away from the inverters. I already have problem with 5m cable on yoke connections so now i am using 3m cable with the yoke. 5m usb cable it too long for many usb devices, especially for usb 2.0. I never tried longer than 1.8m long on the amc1280usb and usually have a 30cm for my tests. The longer the cable the more voltage drop you get on the USB data lines that could lead to garbage in the data. Try shorter usb cable, possibly a better quality too (thicker with shielding) in case the motors frequency affects it as well. Thanks Thanos I thougth the effect of interference on usb communication would be like ON or OFF since it is digital and all the handshakes and error correction algorithms etc. If shorter and better cable is the answer, it is done already. Can you elaborate a bit more on how usb comms garbage effect the settings? Is it something like buffer overflow? You said a new release is coming and it will take care of this problem, any hints how? I want to be sure and safe with the platform that's why i am asking. Thanks
|
|
|
Post by tronicgr on Dec 13, 2016 16:06:32 GMT
5m usb cable it too long for many usb devices, especially for usb 2.0. I never tried longer than 1.8m long on the amc1280usb and usually have a 30cm for my tests. The longer the cable the more voltage drop you get on the USB data lines that could lead to garbage in the data. Try shorter usb cable, possibly a better quality too (thicker with shielding) in case the motors frequency affects it as well. Thanks Thanos I thougth the effect of interference on usb communication would be like ON or OFF since it is digital and all the handshakes and error correction algorithms etc. If shorter and better cable is the answer, it is done already. Can you elaborate a bit more on how usb comms garbage effect the settings? Is it something like buffer overflow? You said a new release is coming and it will take care of this problem, any hints how? I want to be sure and safe with the platform that's why i am asking. Thanks In the latest firmware I have added detection for commands to adjust the PID settings while the platform is running. This was a request from the motion software developer Fabian, for his tool and bridge software. Now when I receive packages from the motion software i capture the ID command for position and count bytes after it until I receive all positions. This is done to be able to receive up new position data up to 1000 times per second. If characters that command to change the PID outside that count appear, the controller assumes these are genuine and executes them. The combination to activate this PID change is only two characters, so there is many chances they will be generated by random noise on the USB cable. In noise free usb communications this will never happen. Never seen it myself on tests via digital analyser or oscilloscope. I hope this explain it. I'll be adding additional lock feature in next firmware to prevent this pid change on the fly. Thanks Thanos
|
|
gokhan
Junior Member
Posts: 89
|
Post by gokhan on Dec 14, 2016 13:40:56 GMT
Thanks for detailed explanation.
Can you tell which release have you introduced this PID command detection? Better, shall I load an older version like 2.1b or 2.0(since 2.2 is beta)?
thanks
|
|
|
Post by tronicgr on Dec 14, 2016 17:27:35 GMT
Thanks for detailed explanation. Can you tell which release have you introduced this PID command detection? Better, shall I load an older version like 2.1b or 2.0(since 2.2 is beta)? thanks I think was introduced in 2.1b. You should give it a try with 2.2 beta. The real beta part about it is the new 12bit digital sensors addition, but if you use the analog sensors it should behave the same as 2.1c. Plus you can use the "motion data only" option to ensure it won't change any PID values on fly. Thanks Thanos
|
|
gokhan
Junior Member
Posts: 89
|
Post by gokhan on Dec 14, 2016 18:15:51 GMT
Ok, thanks
|
|