HVACIMTR: AC Induction Motor for TMDSHVMTRPFCKIT, Phisical Parameters

Chris,

If you would be so kind, could you tell us what kind of IM model was used during identification process in order to understand relationship between 'USER_MOTOR_Ls_d' and LS. Thank you very much.

Best Regards,

Ljubisa

chapter 6.7 of the UG covers ACI parameter identifcation.

The model that is used is FAST.  We don't supply the details of the model or you would know how to duplicate FAST, and that is our intellectual property.

We split Ls into Ls_d and Ls_q for salient PM motors. If you have the separate inductances (we can NOT identify this today) you can feed these into FAST and it will compensate the estimations appropriately. Otherwise they should be set Ls_d = Ls_q (and that is what is done by default through motor ID).

Hi Chris,

by the way, what about the stator leakage inductance - surely is used by the model solver / FAST ?

Hello Chris,

I do not intend to duplicate FAST!

I am just looking for the correspondence between two published values, namely: LS and Ls_d/Ls_q. Or if LS and Ls_d/Ls_q are equally valid, I am looking for the operator to put between Ls_d/Ls_q and LS. In addition, I think that identification is much more effort than the model itself! Thank you very much.

Best Regards,

Ljubisa

the stator leakage is not required as an external input into FAST. Per Chapter 6.2 Induction motors only require Rr, Rs, Ls, and Magnetizing Current

Ls = Ls_d = Ls_q by default. It is only if you set Ls_q and Ls_d differently in user.h that FAST uses different values.

Hello Chris,

Your assertion that Ls = Ls_d = Ls_q is not right for Induction motors! 

I suppose that Ls_d and Ls_q represent leakage stator inductance. Please check with your identification staff so we can establish correspondence between Ls and Ls_d/Ls_q. Thank you very much.

Best Regards,

Ljubisa

FAST is different. and better :)

You only need the identified Ls, and this is passed to FAST as Ls_d = Ls_q

chapter 6 explains the ID process.

Chris,

I have no doubt that Fast is the best!

Can FAST put the relationship between Ls = 0.316423 (H)  and  Ls_d/Ls_q = 0.05296588 (H)? That is the question!

Thank you.

Ljubisa

Hi Ljubisa,

Perhaps I can help clear up some of the confusion.  As Chris stated, FAST does not need to know the stator inductance value.  Instead, it determines what the motor leakage inductance is (Ls, not to be confused with the stator leakage inductance Lls).  In the above conversation, the leakage inductance is somewhat inappropriately referred to as Ls_d and Ls_q.  For an induction machine, we would expect those values to be the same, since there is little or no saliency.  Ls is the inductance that the PI controller "feels" when it is controlling the stator circuit, and is therefore the value we use to autotune our PI current controllers.  More information about this can be found in my PI controller series which is on our motor control blog site:  ti.com/motorblog.

Ls is defined in textbooks as follows:

where:     L = the leakage inductance

                Ls = the stator inductance

                Lm = the magnetizing inductance

                Lr = the rotor inductance

                s = the “leakage factor” of the induction machine

However, the AutoID portion of FAST does not measure each of these individual terms in the above equation.  Instead, it uses another technique to directly calculate Ls.

Does that answer your question?  If not, please respond back and give me another chance to try and clear up some of the confusion.

Many Thanks,

-Dave Wilson

www.ti.com/motorblog

As a followup to my last entry, for some reason the E2E site did not accept my equation image.  It is written out manually below:

L = Ls(1 - Lm²/(LsLr)) = Ls * s

-Dave

Hello Dave,

Thanks for better explanation. I anticipated that Ls_d/Ls_q represent leakage inductance for IM, and had feeling that nomenclature is borrowed from PMSM technology.

But now if  L=Ls (1-Lm*Lm/(Ls*Lr)) and if we plug the above values then: L= 0.316423*0.13706 =0.043368 (H) , which is different than Ls_d =0.05296588 (H), by ~22% !

If we assume that identification is correct than calculation of leakage factor is questionable here, do you agree? Therefore, question is: What FAST say about values of Lm and Lr?

Thank you, and Best Regards,

Ljubisa

Hi Ljubisa,

Yes, I see your point.  I plugged in the numbers myself, and I got the same discrepancy that you did.

I would assume that the datasheet value for stator inductance for this motor assumes rated flux conditions. For just about every ACIM I know of, the rated flux is well into the saturation region of the hysteresis curve.   However, when we run the leakage inductance measurement test, we use a much lower value of flux so as to keep the torque managable during the locked rotor test.  I have not actually seen this motor, but judging by the fact that this is only 1/4 HP (probably for a mass-produced application), I suspect they are using as little steel as possible, or cheap steel, or both.  If true, it would result in a higher leakage inductance reading at lower flux levels due to the fact that the stator inductance (not the leakage factor) is affected by magnetic saturation.

I realize that this involves some speculation, but it is the best answer I can think of to explain the difference.  The good news is that the leakage inductance value discrepancy does not adversely affect the way FAST operates.  I just ran a simulation using a 30% discrepancy, and saw very little difference in the FAST outputs.  The biggest impact will probably be on the current loop PI coefficient values, which are already set very conservatively.

Does that make sense?

Thanks,

Dave