MESA 5i23 7i44 7i49 7i37 8i20

kenny wrote:

Is there any procedure to setting comutate motors, parametrize resolver and basic setting PID? (step by step)
I dont need fine tuning, I need basic + - starting parameters.

The first thing to do it to get the motor turning smoothly in one direction and at the same speed in the other direction.

To do this disconnect the PID from the 8i20.current pin. Then set this pin to various values from the command line.
Only when the motor turns smoothly, and at the same speed in both directions, can you start the PID tuning.
The resolver module gives very good velocity data, so checking that 0.2 amps and -0.2 amps gives the same speed is fairly easy.

kenny wrote:

Ok, but still I dont understand how I make correct comutation on the BLDC servo.

Do you understand how commutation is meant to work on a brushless motor?

These parameters are alright for my HW?

setp bldc.0.scale -559240
setp bldc.0.poles 10
setp bldc.0.lead-angle 0
setp bldc.0.encoder-offset 0

I have no way to know what is right for your motor. I think that as your motor and resolver have the same number of poles you probably want to use

setp bldc.0.scale 1677216 (or -1677216)
setp bldc.0.poles 2
setp bldc.0.lead-angle 0
setp bldc.0.encoder-offset (adjust to suit)

In any case you want bldc.rotor-angle to go from 0 to 1 ten times per rev. Once that part is working then it comes down to positive or negative scale and encoder offset.

kenny wrote:

Last information from TG-drives manufacturer:
My motor have one pole resolver.

That doesn't fit with the earlier statement:

variable: hm2_5i23.0.resolver.00.angle is changed value between 0 an 1 every change mag. pole of motor (10 per rev.)
variable: hm2_5i23.0.8i20.0.0.angle is changed value between 0 and 1 every one turn shaft of motor

The resolver.angle comes directly from the resolver driver, independent of any scaling factors.
If that is the case then you should use 16777216 or -16777216 as the bldc.scale and 10 as the pole count.

Zero point settings from TG-drives manufacturer:
W phase to +
V phase to -
U phase no connect
Then resolver is in zero position.

Mesa assumes that zero is with U +, V - , W -.

U	V	W	Angle	Offset
+	-	-	0	0
+	0	-	30	1398101
+	+	-	60	2796202
0	+	-	90	4194304
-	+	-	120	5592405
-	+	0	150	6990506
-	+	+	180	8388608
-	0	+	210	9786709
-	-	+	240	11184810
0	-	+	270	12582912
+	-	+	300	13981013
+	-	0	330	15379114

So, looking at the difference I would expect the required encoder offset to be 12582912, -12582912, 4194304 or -4194304. (some of these are equivalent)

Well, no voltage on the wire at least.

If you feel like it you could check this with a current-limited PSU and halmeter.

I had a sudden thought this morning ...

Looking at the code on line 627 of the bldc component:
git.linuxcnc.org/gitweb?p=linuxcnc.git;a...6222c928c61a38c#l679

It is apparent that the offset is subtracted before the "poles" calculation.
This means that it would be correct if you measured the offset. (ie, rotate the motor by hand to rawcounts=0, then energise U+ V- W- and note the new rawcounts) but that the table above needs to be divided by 5 for a 10-pole motor.

It might be simplest to measure the offset, really. Though be aware that I am not sure whether to use the number returned by the test, or its negative.

This is a more complete table, and I think it finally explains to me the -950000 offset that my motors need to move, it seems to hint that the 6-pole motors have resolvers aligned to U-V+W-.

U	V	W	Angle	2-pole	4-pole	6-pole	8-pole	10-pole	12-pole
+	-	-	0	0	0	0	0	0	0
+	0	-	30	1398101	699050	233016	58254	11650	1941
+	+	-	60	2796202	1398101	466033	116508	23301	3883
0	+	-	90	4194304	2097152	699050	174762	34952	5825
-	+	-	120	5592405	2796202	932067	233016	46603	7767
-	+	0	150	6990506	3495253	1165084	291271	58254	9709
-	+	+	180	8388608	4194304	1398101	349525	69905	11650
-	0	+	210	9786709	4893354	1631118	407779	81555	13592
-	-	+	240	11184810	5592405	1864135	466033	93206	15534
0	-	+	270	12582912	6291456	2097152	524288	104857	17476
+	-	+	300	13981013	6990506	2330168	582542	116508	19418
+	-	0	330	15379114	7689557	2563185	640796	128159	21359