Smooth transition from V/Hz scaler mode to FAST SpinTAC velocity control

I'm not sure if not getting a reply means this shouldn't or can't be done.

I've seen references to a "sensorless rev-up" where the motor is started in V/Hz and then transitions to sensorless FOC. This seems like a good approach to try since I have had no success at all using force angle to get a smooth startup.

For an ACIM what angle/control alignment need's to be done to transition from V/Hz to FAST velocity control?

Thanks.

You are looking at transitioning from V/Hz to FOC because the start-up capability in FOC is not good?

I would first look at improving the start-up performance.  First let me ask, once started up, are you getting good performance during full speed and load?  If start-up is compromised it first makes me think to make sure the parameters and settings are correct.

With an ACI motor you really don't need to use ForceAngle.  By inducing the rotor flux with Id current you should get an acceptable start-up with FAST able to track at very low speeds.  Attach your user.h file with your response so I can double check.

Hi Chris,

Yes, I'm looking at transitioning from V/Hz to FOC to improve startup at low speeds.

The motor performance is very good other wise, including full speed and full load.   See my attached user.h.

Based on your comment about not needing ForceAngle, I have been testing starting without V/Hz or the ForceAngle feature.    I still see an unacceptable amount or speed overshoot at low speeds, and the current waveforms aren't smooth on startup especially with light loads.

Below are graphs showing light and full load startups.    The graphs are 200 samples at 10 msec intervals.   The graphs show: FAST angle, FAST speed, IqReference out of ST Velocity Control, and the reference into ST Velocity Control.  Also speed indication from a 2K line incremental encoder.   The encoder data is sampled at the same rate as ST is running (1 mSec) and scaled to electrical Hz similar to the FAST speed reading for comparison.

The FAST Angle starts increasing positive immediate after startup, but for some reason the FAST speed indication goes negative for the first ~175-250 msec.  During this time the independent encoder shows speed increasing.   The Iq reference wildly overshoots because the FAST speed is negative and not increasing.  Why would the FAST speed be negative if the FAST angle is increasing and also an independent tach shows the motor is accelerating?

No Load Startup:

     Full Load Startup:

7711.user.h

are you using the encoder in the control loop, or just as a reference?

are you running RsRecal right before start-up?

if not - you should if possible - but try lowering your Rs value in user.h by 20% and see if you see the same issue. If the Rs being used by FAST is > than the actual Rs, the initial orientation for FAST will be 180 out of phase.

did you say that you tried these start-up tests with ForceAngle enabled also?

The encoder is for reference, to see what the motor is actually doing.

Offset calibration and RsRecal are being performed right before start-up. Rs is being calculated at about 4.67 ohms each time. I tried disabling RsRecal and putting a 10% lower value (4.2 ohms) in user.h. The FAST estimated speed still goes negative for awhile on startup, like shown on the graphs. The only difference with Rs artificially lowered is the motor takes slightly longer to get moving, as indicated by the encoder speed.

Lowering the value of Rs didn't help the FAST speed estimate starting negative.

Have you tried with ForceAngle enabled?

The 'force angle' (if set up correctly) should force rotation at startup in the correct direction.

 

Hence to consider the negative  estimated speed issue the following check list is advisable:

- are voltage and current ADC offsets measured corectly

- is the voltage measurement correct: which means looking at the hardware , correctly matched components  for all three phases, gain set correctly so  that the ADCfullscale voltage is  what you think it is

-same issue for the current sensing circuit,  if shunts are used check if components are matched and the gains in use are so that the

  ADCfullscale current is what you think it is

-check the measured resistance of FAST against a Ohmmeter to see if it matches. It is preferable  that  the value used by  FAST  is  EQUAL  or LOWER then the actual value. 

- check on noise issues, on the measured signals. this can be an issue of the shunts are not set up correctly, i.e Analog and Digital OV linked at the correct point.