BLDC and fanuc redcap parameter problems?

More
27 Nov 2015 21:39 - 27 Nov 2015 21:48 #65887 by gmouer
I am attempting some bench testing before a VMC retrofit. I have a fanuc 5S motor, AMC BE25a20AC drive, mesa 7i77 and a parport breakout board for the tests. I realize other mesa boards will be needed for the actual retrofit, this is just to prove the redcap can operate under linuxcnc using the BLDC component. Running 2.7 master under debian wheezy.

My test setup : Drive in velocity mode, running open loop velocity mode with no positioning loop yet. Using onboard pot for speed control.
Drive has red/green led which indicates fault as well as valid commutation signals.
Using encoder feedback to drive for velocity loop.
Fanuc 4 bit graycode and the 3 hall outputs are all ran through the parallel breakout board.
Encoder fed through 7i77 and displays properly.

BLDC setup
loaded with cfg= fHT
no other parameters loaded or modified from defaults I found the pattern set to type 19 by default which does give a green led on the AMC drive. BLDC is being used for comm code conversion only, fanuc gray 4 bit to 3 industry std hall signals.
There is that youtube video of a fellow doing the exact same bench test and he claims to have used pattern 9 or 36, those patterns show as non valid on the amc drive. I tried all 48 patterns and none gave smooth running. Some patterns ran others didn't but none ran smooth. Tests were limited to a couple hundred rpm max.

Any suggestions on getting the motor running smoothly? Trying all motor power combinations along with all 48 comm patterns is obviously not a option. I appear to have good signal quality, bldc is code converting. I must need other parameters set or something. Knowing which BLDC pattern I should try would help a lot.

Thanks!
George
Last edit: 27 Nov 2015 21:48 by gmouer.

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

More
28 Nov 2015 01:20 #65901 by andypugh
It is probably possible to work it all out by logic alone, but that is likely to need information that isn't readily available (such as what each drive manufacturer and motor manufacturer defines as the zero angle)

So, if it was me, I would still be trying trial and error.

There are not that many motor phase combinations, and running through the bldc hall pattern codes is fairly quick. (you do _not_ need to restart between each iteration, you can change them on the fly from the halcmd prompt).

If I had infinite spare time I would love to write a utility to analyse motors and work out the optimum bldc component numbers. It is definitely feasible, I just have too much else to do.

The Fanuc 4-bit scheme isn't a perfect fit to the 3-bit Hall pattern scheme. It is possible that if you hook up an encoder and turn off the "T" option that things will get smoother. (ie, generate Hall from the encoder, not 4-bit Fanuc)

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

More
28 Nov 2015 01:33 #65903 by gmouer
I did run through all 48 combo's using the halcmd window in "show hal configuration", without restarting. I did not try the motor UVW combinations for each pattern.

I was not aware that the fanuc 4 bit conversion was not a perfect fit, hopefully results better than I have now are possible. I have the fanuc encoder hooked up and feeding into the 7i77 giving me a axis dro, so the encoder is there and that method is possible if just code converting does not work well.

Do I need to set any other parameters for just code conversion? Like the number of poles?

I guess I am off to try a whole lot of BLDC patterns and motor wire combinations, not looking forward to that one !

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

More
28 Nov 2015 01:44 #65905 by gmouer
Another question,

The guy on youtube said he had used both pattern 9 and 36, both which worked well just with opposite rotation.

Have the BLDC patterns and their numbers changed over the different development versions of BLDC or have they remained the same all along?

If the same, I will start off using pattern 9 and 36 and experiment with motor combinations for those patterns first.

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

More
28 Nov 2015 01:49 #65906 by andypugh

I did run through all 48 combo's using the halcmd window in "show hal configuration", without restarting.


If you open a terminal window and halcmd -kf you get the same effect, but with tab-completion and the up-arrow saves you re-typing the whole previous line.

I did not try the motor UVW combinations for each pattern.


There are only 6, so it isn't that horrible.

I was not aware that the fanuc 4 bit conversion was not a perfect fit

Basically there are 16 Fanuc codes and 16 Hall patterns (they are not a binary set) and 6 into 16 doesn't go.

I think the blcd component knows exactly where each Fanuc edge is as an angle (it certainly knows this for Hall, but it has been a while since I wrote it) so effectively any transition can be used as an index to switch to using the encoder as the master reference.

Interestingly, I am currently trying to decide if my rotary axis motor need to stop doing that. It actually seems to have more torque some days than others. I am wondering if that is because one or more Hall-edges are out of place, and that I get iffy commutation sometimes.
If so, the solution is to use "hi..." mode and switch at encoder index. But that does mean finding where the encoder index is magnetically.

I wrote the component. I know the problems. I can't necessarily solve the problems.

Do I need to set any other parameters for just code conversion? Like the number of poles?

Depending on the mode, that can be critically important. However I don't _think_ it matters in "h" or "f" modes as those are in "magnetic space" rather than "motor revolution space"

I guess I am off to try a whole lot of BLDC patterns and motor wire combinations, not looking forward to that one !


It doesn't take that long with halcmd.

If you want to get very posh, a PyVCP panel and an up-down comp would make it faster.

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

More
28 Nov 2015 01:59 #65909 by andypugh

Have the BLDC patterns and their numbers changed over the different development versions of BLDC


No.

If the same, I will start off using pattern 9 and 36 and experiment with motor combinations for those patterns first.


I think it is actually quicker to run through the software patterns than to re-wire the motor.

NEVER rewire a motor with the motor drive powered and enabled. Some don't care. Some blow up immediately.

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

More
28 Nov 2015 03:39 #65916 by gmouer
Well, I lost another round in the battle.

The AMC BE25a20AC drive has a red/green status led. That led goes red if there is not a valid commutation pattern. AMC shows in a video rotating the shaft with the power leads disconnected and verifying the led remains green. I did this test for all 48 BLDC patterns and about 12 of them gave a steady green led while rotating the motor shaft.

Next, I tried all 6 motor power lead combinations for each of the 12 BLDC patterns that gave a solid green led. None of the combinations gave a decent running motor. The best I got was quite rough.

I am assuming for the fanuc 5S that 120 deg comm is the correct setting on the AMC drive, not 60. Not a lot of info of that sort around on fanuc motors.

As for using the encoder for commutation, I have read all that I could find on the process but the BLDC docs do not detail the process much, only the pins/parameters.

How the youtube guy got that motor running with patterns 9 and 36 I have no idea. He used the same drive that I have but did not provide much details. Patterns 9 and 36 give a red led on the drive when I tried them. With a red led, the drive is in a fault mode and cannot be ran.

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

More
28 Nov 2015 03:47 #65918 by andypugh

I am assuming for the fanuc 5S that 120 deg comm is the correct setting on the AMC drive, not 60.

I would have assumed 60, actually.

How the youtube guy got that motor running with patterns 9 and 36 I have no idea. He used the same drive that I have but did not provide much details.


You could ask him, I found him quite communicative.

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

More
28 Nov 2015 05:50 #65921 by mariusl
Andy
I have a number of motors here that I would like to use but I dont have drives for them. Would it be possible give some sort of schematic on how to use this component with discrete bridges or drives.
I am trying to follow what is being discussed here but have very little insight as I dont understand how things are strung together :(

Regards
Marius


www.bluearccnc.com

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

More
28 Nov 2015 12:28 #65932 by gmouer

I am assuming for the fanuc 5S that 120 deg comm is the correct setting on the AMC drive, not 60.

I would have assumed 60, actually.

How the youtube guy got that motor running with patterns 9 and 36 I have no idea. He used the same drive that I have but did not provide much details.


You could ask him, I found him quite communicative.


I will try and contact him through the youtube video, that is the only way I know unless you have contact info. As far as I found in searching he is the only one that has come close to getting a redcap running using BLDC. He didn't have a positioning loop and did not understand why the motor had no holding torque but that is a easy task once the motor is running well in open loop velocity mode.

I am going to give it another try today, setting the AMC drive switch for 60 deg commutation. The other non fanuc brushless motors I have used were all 120 deg so I just assumed........ Maybe a bad assumption with luck. Again, I will report back..

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

Time to create page: 0.989 seconds
Powered by Kunena Forum