Efficiency of TI's motor control kit

Hi Emondo,

Would you pls let me know which build level you are running?

Second, why do you need to apply a constant IqRef1, it should be taken care by speed PID in the given FOC scheme.

Regards

Hi! Bill,

Thanks for your reply, and my answers to your instructive questions are as the following:

Q1.  Would your pls let me know which build lever you are running?

Ans.  #define   BUILDLEVEL LEVEL6

Q2.  Second, why do you need to apply a constant IqRef1, it should be taken care by speed PID in the given FOC scheme.

Ans.  I’m still learning how to use IqRef1 correctly and efficiently. I think it’ll be safe to set IqRef1 as high as 0.8 to carry the increasing load without losing the set speed too much. It seems that I can't paste my plot of torque and speed here. If possible, I'd like to send you the related measurement data. 

In addition to instructing me how to operate TI parameters correctly, would you possibly afford me some your experimental data for comparison and reference?

Regards

Hi Emondo, Bill

I have a few thoughts and questions on the subject that might help clearing things up:

1. From my experience measuring efficiency is not trivial task as any error in measurement can quickly lead to incorrect results (in either way + or -). Thus proper (certified) equipment must be used.

2. Emondo how did you measure the torque? I have no experience with small dynamometers, but if you used one with range to e.g. 10Nm you can not expect it to give you accurate readings in bottom 10% of measuring range, so there might be significant measuring error involved. The same goes to DC current and voltage measurements. How did you measure them as any EMI (from the three phase bridge) can (I have witnessed such a case) cause inaccurate current or voltage measurements.

3. Bill. As far as I am familiar with TI code structure if Emondo is in build level 6 the IqRef is never used, so it does not mater what value does it have.

4. As far as I can tell the efficiency we are talking about is efficiency of the whole system inverter and machine. It would be nice to know which is the weakest link the inverter or the machine, but my guess is that the machine is operating far from nominal operating point so most of the losses are in the machine.

Regards, Mitja

Hi! Mitja, Bill, and any other interested experts,

(By the way, Mitja, it’d be nice if you may delete the extra letter “d” that you used to put on in my name!)

The following responsive answers are my reply, suggestion/comment, and request:

1. From my experience measuring efficiency is not trivial task as any error in measurement can quickly lead to incorrect results (in either way + or -). Thus proper (certified) equipment must be used. (Mentioned by Mitja)

Ans. The efficiency of the driver with the motor is crucial especially in the application of electric vehicles, in which we want to design TI’s Piccolo with reference to the TI KIT, as mentioned above. Thus origins the current efficiency-related argument.

2. Edmondo how did you measure the torque? I have no experience with small dynamometers, but if you used one with range to e.g. 10Nm you can not expect it to give you accurate readings in bottom 10% of measuring range, so there might be significant measuring error involved. The same goes to DC current and voltage measurements. How did you measure them as any EMI (from the three phase bridge) can (I have witnessed such a case) cause inaccurate current or voltage measurements.

(Mentioned by Mitja)

Ans. For measuring the efficiency of TI's “Dual Motor Control and PFC Developer's Kit” concerning only about its controller and one appended driven motor, we use DC power source and KYOWA toque meter TP-2KCE, of which the rated capacity is 0.2 Nm. And we’ve already subtracted the overhead power.

3. Bill. As far as I am familiar with TI code structure if Edmondo is in build level 6 the IqRef is never used, so it does not mater what value does it have. (Mentioned by Mitja)

Ans. We need Bill or any other experts to verify/endorse this.

4. As far as I can tell the efficiency we are talking about is efficiency of the whole system inverter and machine. It would be nice to know which is the weakest link the  inverter or the machine, but my guess is that the machine is operating far from nominal operating point so most of the losses are in the machine. (Mentioned by Mitja)

Ans. What an expertise viewpoint from Mitja! Because of not having the efficiency data about the driven motor appended with this TI KIT, I just showed my experimental record from the beginning for all your review. Bill, any TI or other interested experts, we do need some experimental data related to this TI KIT other than mine, to make sense of the efficiency disaster number < 30%?!

Regards

I can confirm that in the final build (5 or 6 with dual axis) and with lsw=2 (closing the outer speed and inner current loops) that Iqref is NOT used, and Idref = 0, so shouldn't matter.

I disagree with how you are trying to measure efficiency though.  Do you have any other motor / control efficiency numbers calculated in this fashion?

Hi! Chris,

How about showing us some TI’s experimental data about the efficiency issue? If TI people don’t feel proper to release these essentially crucial data, or if there is no such data at all in TI for this TI KIT, may any other experts possibly announce some to conclude this post?

Regards

Hi Emondo

I do apologize for my mistake spelling your name. It has been corrected.

Thanks for the info on torque meter I have never seen ones for such a small torque and this info on supplier will sure come handy in the future.

Given that you will probably not get any additional info on the machine I can recommend few thing I would try (not necessary in this order):

1. I don't know how is the rest of your laboratory equipment, but you should be able to get approximate estimate of electrical power, that the machine receives ( thus measuring machine efficiency) by measuring two phase currents, and two phase-to-phase voltages with a scope. If you have a good scope you should keep the sample rate very high in order to correctly sample voltages. Once you have the data import it in  to matlab or some similar package and post process it to get electrical power.

2. You can measure the machine stator winding resistance and you can get at least estimate on copper losses if you also know phase current.

3. You can measure the machine losses/efficiency in generator mode at the same operating point that you tested in motor mode (have the brake act as the motor, measure speed and torque and put pure symmetric ohmic load on stator of the machine to get approximately the same current that you got in motor mode). This efficiency will be the top efficiency that you can get in motor mode at the same point (there are additional losses in copper due to skin effect and there are no eddy and hysteresis loses of higher frequency content of supply current - due to the switching nature of the supply)

If you do any of these measurements please do post the results

Regards, Mitja