High Voltage Motor Not turning smoothly

Gary H said:

The winding resistance as measured from line to line is 3 ohms (so 1.5 ohms per phase). The inductance as measured with LCR meter at 1 kHz from line to line is about 18 mH (so 9mH per phase).

The inductance you measured with InstaSPIN-FOC is significantly different. That's where I would focus first

USER_MOTOR_Ls_d                 (0.06921637)

Gary H said:

When running the lab 2b the motor does seem to stall sometimes even when I increase the current.

Is the motor supposed to stall during this test?

During which test?  During Motor Identification

1. Motor must be unloaded, or only slightly loaded.  The rotor has to spin constantly from RampUp until RampDown state.

Gary H said:

Should I disconnect all load from the motor? Currently the motor is built into a two head compressor, but I do not load the compressor, however, some loading does occur from the friction of the compressor pistons.

It's better if you can. Typically if the compressor isn't loaded you can still ID...but again, if you have one of the motors outside the enclosure just use that for ID.

Gary H said:

When I try to run the motor RPM up to 2000 it stalls. From what I understand the current settings are in amps. If so, it seems these currents are high for what I am used to driving the motor. Typically at max load, I am not driving more than 2 amps, but with this software it seems 3 amps is just barely driving the motor

Let's get proper parameters ID'd first, or if you like, simply load the Ls value you measured with your meter into the user.h (0.009).  BTW - if you know the salience of your motor you can enter it directly in user.h as well, note the Ls_d and Ls_q.

Gary H said:

My main question is… as long as the old sliding mode observer was able to work with this motor should the new FAST algorithm be able to work as well? Do you see anything about this motor that would concern you with your new software?

Yes. No. You will get significantly better performance with InstaSPIN-FOC though.

Gary H said:

My main goal of switching to the new software is to hopefully gain more starting torque and to utilize the stator winding resistance feedback while the motor is running

FAST will be able to take over at a lower operating point than the SMO, so that should help.  Yes, RsOnline will help quite a bit as well.

Since you are familiar with SMO you may also want to look at the eSMO project in controlSUITE.  And since you have an IPM motor you could take advantage of the High Frequency Injection method to estimate an angle at 0 speed and give best start-up capability.  The difficulty in this solution is with the transition from HFI to eSMO angle.  InstaSPIN-FOC handles this automatically.

 C:\ti\controlSUITE\development_kits\HVMotorCtrl+PfcKit_v2.1\HVPM_Enhanced_Sensorless_2803x

Thanks for the quick feedback. I will try to remove all the load from the motor and try again. This may take a few days to get, but I will follow up with what I learn.

With a little help from a hammer and screwdriver, I was able to remove the pistons from the compressor so that the motor could freely spin. I reran the lab and got the following parmaters.

#elif (USER_MOTOR == My_Motor)

#define USER_MOTOR_TYPE                 MOTOR_Type_Pm

#define USER_MOTOR_NUM_POLE_PAIRS       (4)

#define USER_MOTOR_Rr                   (NULL)

#define USER_MOTOR_Rs                   1.47)

#define USER_MOTOR_Ls_d                 0.01135705

#define USER_MOTOR_Ls_q                 0.01135705

#define USER_MOTOR_RATED_FLUX           0.4322564

#define USER_MOTOR_MAGNETIZING_CURRENT  (NULL)

#define USER_MOTOR_RES_EST_CURRENT      (1.0)

#define USER_MOTOR_IND_EST_CURRENT      (-1.0)

#define USER_MOTOR_MAX_CURRENT          (4.0)

#define USER_MOTOR_FLUX_EST_FREQ_Hz     (20.0) 

Much better response!  Once I moved on to Lab 5b, I grabbed another fully assembled compressor and ran it through its passes. Everthing ran wonderfully. I was able to achieve much better performance especially at the lower RPMs. My next step is to move from the development kit to my system board. This will require a board layout change since I will be moving to the TMS320F28027F from a TMS320F28035, but the TMS320F28035 was a little too big for this system anyway. 

Since this is a cost sensitive high volume product, I will also be looking for a lower cost solution for the current feedback. The OPA2350 is a little pricey. I will also investigate how my performance changes if I only use two current feedbacks instead of three in order to save the cost of another op-amp. I will also work with my local TI guy to find the best solution. I understand the Ifbs need to be high bandwidth op-amps, but does the Ifb-Sum op amp need to have a high bandwidth? Any suggestions for cost reductions would be appreciated. 

Thanks for the help.

Gary H said:

Much better response!  Once I moved on to Lab 5b, I grabbed another fully assembled compressor and ran it through its passes. Everthing ran wonderfully. I was able to achieve much better performance especially at the lower RPMs

Thank you for posting!  Good to celebrate success on a forum that is mostly people working through challenges :)

Gary H said:

I will also investigate how my performance changes if I only use two current feedbacks instead of three in order to save the cost of another op-amp.

The third current shunt is only required if you want to use over-modulation.  The default modution of (1.0) pure sinewave is supported with 2-shunt, and it may be possible to extend this to 1.15 modulation as well.  To move from 1.15 to 1.33 the third is required.

Gary H said:

I understand the Ifbs need to be high bandwidth op-amps, but does the Ifb-Sum op amp need to have a high bandwidth? Any suggestions for cost reductions would be appreciated. 

With 3 shunts they do NOT need to be high bandwidth because you are always ignorning the shunt that is in a small sampling window where the bandwidth is required. Typically this is the same for 2 shunt limited to (1.0).  It gets into a grey area with 1.15 and 2 shunt.

Here is our opinion on OP-AMP slew rate design:

My suggestion is to have at least 5V/µs, and have an OPAMP that is pin to pin compatible with a 1V/µs or even 0.5V/µs. Then, go through all the required tests needed in the application, such as speed, transient conditions, input voltage variations, etc. Once the performance and all the tests have passed, only then, I would start lowering the slew rate step by step, so from 5 to 2, to 1, etc.

This is because there are several factors that are taken into account when selecting an OPAMP, such as PWM frequency, maximum duty cycle used, layout, grounding, etc. That is why I am suggesting a higher slew rate to cover all tests before selecting a lower slew rate.

We don't use the Ifb-sum in the InstaSPIN application examples, though it is common to sample this value or run it through a comparator for a Fault / Trip configuration.

Hello,

Gary H said:

Are these jumper positions in the “InstaSPIN Projects and Labs User’s Guide”?
If not, they should be for lazy guys like me that never reads all the directions.

There is in motorware\motorware_1_01_00_13\docs\boards\control\f2802x\docs, but not in InstaSPIN Projects and Labs User's Guide.

 I was also the one that lazy to read all documents and got problem after that :)

Cheers,

Maria