Check your frequency to the DQ/ABC is Wr + Ws. Check that positive Ids, Iqs frequency gives positive speed.

Is the Parks or DQ transform used for your D/Q motor model and controller the same?

Are you using voltage or current controller?

One way to check you DQ/ABC or ABC/DQ is to simulate a 3 phase 50/60 Hz voltage source and input these into the ABC/DQ transformation module. Use the Valpha and Vbeta as the Sin(theta) and Cos(theta). If you get DC it is correct. If you get 100/120 Hz or double frequency the direction of rotation is incorrect. Then just change the sign of the alpha / beta part of the Parks transform to correct this.

Are you using the orthogonal or non-orthogonal form of the parks transform?

Are you using actual or per unit?

Dear Gregory sir

Thank you very much. The situation is due to some problem the dead band and frequency were not proper. But today without doing any modification in the model i run the induction motor and the modulation index too.

My next doubt is i am using ADC to check the Motor current. So in the model sine pwm generation and ADC blocks are there. Should i make these two as compound block or i can run the real time by selecting both and make code generation.

What are the details of the system you are using (Spectrum Digital?).

What is simulation and what is in real-time hardware?

Is the sine wave PWM in simulation the same as in h/w. I mean does the PWM simulate the PWM in the DSP / PWM controller?

Send a block diagram of your simulation in word or PDF.

I rectified the issue sir. First i generated sine PWm using dq-alpha,beta-abc conversion which i said already. To measure the induction motor current when i control it through sine pwm, i used the channel ADCResult0. In this stage the vissim window consist of compound form of sine pwm generation block and ADC channel block. While i try to generate code by selecting both, i was in unable to generate the code. Then after raised my doubt to you, i use compound form of both block then generated code and i can measure the adc signal as well.

Sir, now i am sensing the adc output. Is there any block to measure the frequency of sinewave directly.

Thank you

with thanks and Regards

Rajin M. Linus

If you are using DQ/ABC block you must input the frequency (W = 2*pi*f). You can output this to the DA block. Also the stator currents should be at the same frequency so you can capture both on an oscilloscope

Is your speed and current loop running at the same sampling rate? Are you using interrupts and different sampling rates for the speed, current / PWM loops.

You should do this otherwise all the code will try to execute at the same time in the background.

Till now i am trying with open loop. I need to do FOC control of PMSM. Is it possible to get the frequency from the sine wave and calculate the speed instead encoder feedback. As per your discussion i need to use different sample rate to get the feed back right?

Are you using a PMSM with sinusoidal excitation or BLDC. The former uses sinusoidal currents the latter square wave currents. In a BLDC machine they normally have hall effect devices to control the PWM but in a PMSM you have to use a SE (absolute) and also you have to align the SE to the zero crossing of PHASE A when running as a generator. You can see this by turning the motor by hand and looking at the output voltage on a scope. You can also use a resolver which must have the same number of poles as the PMSM and this gives out sine/cosine wave which you can use for Wr and for the DQ/ABC transform.

Go to microchip web site and they have an example and even code for a FOC for a BLDC motor and IM. I have attached the PDF.

Ok so what you are doing is ok. I understand you are just trying to generate the PWM to see if it works ok without any feedback..

First generate the PWM as you are doing but design so that you can manually vary the inputs to the DQ/ABC block. Set Vd = variable and Vq=0 and also make W (f) adjustable.

Then when you compile and download you can vary Vd and W and you will be able to see the PWM (whether from the PWM modulator / DSP output or inverter and you should see the waveforms vary.

If you set frequency to 0 then it should look just like a DC drive with fixed PWM and as you vary Vd you should see the PWM vary on the scope.as you vary Vd with Vq=W=0

Normally if your PWM is say 5kHz then the current loop would run at 10kHz. If you do this and synchronise the current loop interrupt to the top and bottom of the modulating wave (normally triangle) then you will not require antialiasing filters.

Once you have this running ok in open loop then you can connect to actual inverter and see what the actual currents look like and ensure that you have the correct sign for the current and speed feedback.

see http://www.microchip.com/stellent/idcplg?IdcService=SS_GET_PAGE&nodeId=1824&appnote=en544825

Sir i already generated three phase PWM using DQ to ABC and i have seen the induction motor current through oscilloscope. I can take this current waveform to VISSIm through ADC. My doubt " is any block available to sense the frequency of the sine wave and calculate speed". So that i can use the same concept in PMSM instead of Encoder available with PMSM.

In the example i have seen they are using ADC unit delay. Can you explain what is the reason for the same.

Herewith i have enclosed the waveform which i got through sine PWM in induction motor. 1st one is Pulse of first phase

2nd one is first phase generator current

Thank you so much for your interest to solve my doubt

Ok the waveforms you sent look fine.

Now why do you need a block to calculate the frequency of the currents you show when you already know this. If your controller frequency is say 20Hz then the currents will be the same. In your diagram of your waveform the scope shows the frequency to be 49.02 Hz or 20.4 msec.

As far as the speed estimator goes you will have to develop this block yourself.

I do not see any ADC unit delay, please explain.

Just send a screen shot of your controller

The need for frequency measurement from the sinewave is as follows

We know that the speed of the pmsm is 120f/p. where f- frequency and p- no. of poles.

To do the sensorless control if somehow i find the frequency of ADC output through some blocks in Vissim, i can multiply with 120/p and get the speed. So that i asked that do you know any block in vissim to provide the output as frequency of the input sinewave.

The second one is

While we do the program we have to set the interrupt and delay to get the ADC output etc. So like that should we do any delay block and other thing in vissim.

Dear Diana

Can you explain how to measure the rotor angle from F281x quadrature encder block. The encoder output is 2000 lines for one cycle. I am in unable to think how to get the rotor angle from the same.

An incremental encoder normally has a Phase A, Phase B which are 90 deg out of phase and a reset once per cycle or revolution.

For +ve speed A will lead B.

For -ve speed A will lag B.

All DSP microcontroller encoder blocks consist of an UP/DOWN counter with a reset.

In your case the angle resolution will be 360/2000 or 0.18 deg per pulse or count.

You may use either A or B to drive the counter so the angle at any time is count*0.18 deg. Thus the counter serves as an integrator.

You then use the phase difference between A and to drive the counter to count up or down. ie if Pulse A leads pulse B count up if Pulse A lags Pulse B count down. This is normally done by using a timer to see the time at which phase A and Phase B cause the interrupt on the DSP / micro ports.

Then you use the reset pulse to reset the counter.

The DSP/micro should have input ports, timers etc which do this. They take A,B and Reset as the inputs and all the rest is done in h/w min the DSP. You just have to read the value of the counter from a buffer or holding register and scale to get actual angle.

Thus if you turn the encode by hand you should see triangular wave going from 0-2000 and which should reset to zero. If you turn the encoder in the opposite direction it should start counting down and you will get a triangular counting in the opposite direction i.e. 2000-0

Hope this helps but there is plenty literature on this available. Just Google incremental shaft encoder operation.

Thank you diana. Is it possible in vissim to activate counter etc

If you refer to the TI SPRU065E manual it provides all the above details.

On page 1-5 of the Overview it provides a nice (but complex) diagram of the EVENT MANAGER.

The QED inputs are the CAP1_QEP1, CAP2_QEP2 and CAP3_QEPI1. These are the same as A,B and reset I described above.

As you can see CAP1_QEP2 and CAP2_QEP2 are input to the QEP logic unit which determines both clock and direction which is used to drive GP Timer2 clock and dir inputs via the QEPDIR mux and QEPCLK mux respectively.

All three inputs are also fed into Index Qual logic unit which provided the reset to the GP Timer 2.

Hence all you would have to do is to set the various registers (EXTECONA, T2CON(5,40 and T2CON(15:11,7,6,3,2,0) and once this is done then all you have to do is read the GP Timer 2 register.

Note as they use both the leading and falling edge of A & B you get twice the frequency from the encoder. ie 4000 per revolution not 2000.

Note that TI allow for 2 shaft encoders denoted EVA and EVB the former users GP Timer 2 and the latter GP Timer 4

Hope this helps.

I see you ask questions about both MATLAB/SIMULINK and VISSIM.

See pic. Outputs A,B&Z

Thank You Diana, I understood what you said. In the quadrature encoder for F281x block in Vissim what is the output will be. they have mentioned that T2CNT will cycle back every 2^16-1 counts. If encoder output is 2000 count is 2^16-1 = 2000 count?. Moreover I am in unable to enable the counting f index pulses by providing the tick in the dialog box of Inc/Dec Rev count on index pulse. If i provide tick , i can generate code but i will i run the output file the TI driver error-1143, PTI_ERR_RDY. Can you tell me what is the reason. Moreover without enabling the same if i run the output file T2CNT is increasing for low speed for high speed it is settledown in some value and remains constant. Herewith I have enclosed the output without enabling index and with enabling index in vissim8 (F281x Quadrature Encoder block)

Attached file - Showing T2CNT count increasing for low speed (below 600 rpm) and for high speed (above 600 rpm) remains constant. After that again less than 600 rpm T2CNT count again increasing. (This is the output with out enabling index count)

Attached file- This is the file to show the error while index count enabled

If you are doing 600 rpm it should do 10 rev/sec or 2000*10 = 20000 counts / sec. Without any index it should reach 65535 in about 3 revs. We are not seeing this in the first diagram. Try using smaller time (say 5 sec) in time axis as you might be getting aliasing. What is your PC screen update rate? It might be correct in the controller but on the PC may not be able to keep up in real time.

Did you test on bench before connecting to motor shaft? Just remove the encoder and turn it by hand and see what you get. Try 1 rpm by hand first :)

Remember, when running on controller at say 3000 rpm it will count from 0- 65535 ~ 4 seconds within the controller but your PC will not be able to keep up with this. 50*2000 = 100 kHz and I doubt your PC screen can update at this rate.

You might have to buffer the data or store on DSP and then do post processing.

Dear Diana

My PC screen update rate is now 60 HZ. I am having the provision to maximize it till 75 HZ. Do u have any other sugession to rectify the issues of QEP.

Thank you

With thanks and Regards

Rajin M Linus

You seem to have misunderstood me.

The screen update rate you appear to be referring to is the actual screen or monitor flyback rate. This has to do with your graphics card.

What is your VISSIM sample rate. Please show a screen shot of your System Properties Dialog. Start, Time Step and End

This is the rate at which the controller updates the screen. If you set 0.001 the update rate will be 1 kHz.

Please send screen shot.

Dear Diana

I already set the starting time as zero, frequency 10Khz and end time as 100 sec. After the discussion with you i have changed as frequency as 0.001 sec and end time as 1 sec. Even i do this, after one particular speed, T2CNT value is remains constant. Herewith i have enclosed screen shot. Moreover even i fix the end time as 1 sec, the system keep on running.

Thank you

Ok what speed is your motor running at.

Please turn the motor by hand with no power to the motor you can set time axis to 100 sec.

Can you Skype? Then I can see and we can chat.

My Skype name is gregory.diana

Dear diana

I dont have a skype ID. Even i have it also here the net is so slow. Actually i coupled the PMSG with shunt motor. Today i changed the direction of rotation and checked it. During that time , when the motor rotate beyond 1600 rpm the count will be constant as -5 counts.

Then i checked with different speed as follows

1. When the motor speed is 124 rpm, the expected count in T2Cnt is 248000 counts/min. That is (34 times it reach 65535). But the value became -525/min. (The value = Expected value/472.)

2. When the motor speed is 620 rpm, the expected count in T2Cnt is 1240000 counts/min. That is (19 times it reach 65535). But the value became -2700/min. (The value = Expected value/459.)

3. When the motor speed is 1600 rpm, the expected count in T2Cnt is 3200000 counts/min. That is (49 times it reach 65535). But the value became -2025/min. (The value = Expected value/1580.)

4. When the motor speed is 1000 rpm, the expected count in T2Cnt is 2000000 counts/min. That is (30.5 times it reach 65535). But the value became -4225/min. (The value = Expected value/473.)

So till 1000rpm the value= Expected value/ 460 (Approximately). When the motor speed again increases to 1600 rpm the denominator value increases and the value =Expected value/1580. After the speed of 1600rpm The T2Cnt value is constant. How to rectify the issue.

If i change the direction of rotation, after the speed of around 600 rpm the T2cnt value remains constant.

what happens when you turn the PMSM by hand.

What type and make of shaft encoder are you using.

The Encoder detail is as follows

Position feed back - (Encoder 2000 PPR & Index Pulse)

&

3Nos of hall Sensors Placed with 120 Electrical degree apart

Since shunt motor is coupled with PMSM, i cannot rotate by hand freely. But while i rotate T2CNT value slowly increasing.

When you rotate by hand you must see the full count.

If not then it might not be connected properly.

Can you see encoder pulses which feed the QEP on an oscilloscope.

What is your wiring circuit diagram for the encoder. Are you suing shield pairs.

Do you not have spare encoder to test.

What is the supply voltage for to the encoder and how is it and the controller connected and earthed.

Dear diana

I experienced some error in the f2812x quadrature encoder block in vissim. So i windup the trial and i measured the speed some how using speed calculator block through cap2 output. Now i can get the speed in rev/sec.

Please tell me is there any chance to find the rotor angle from this output. In this regard i have to covert it in to rad/sec and integrate it. After that i have to use MOD ( Integrated value, 6.28 rad) we can get the rotor angle. Can you provide me some help to find the mod in vissim. Otherwise do u have any choice to find the rotor angle from the speed.