Intelligent Power Modules 1

[ShamilaWije]

Hi,

I'm designing a 3 phase motor drive for a 3 phase BLDC drive. The drive should operate from a 440V 3 phase system and should be capable of handling 5kW to 7kW.

I've been looking for a package which integrates all the IGBTs and the drive electronics from manufacturers such as IOR and Fairchild. They all offer very good solutions but only upto 400V working voltage.

The best solution I found yet is from Mitsubishi semiconductor (Powerex). The part no is PS12036. It integrates all the IGBTs, drive ccts and provides protection such as over current protection, control side under voltage shutdown, shoot through lock out etc.

Do you guys have any suggestions?

 

[sreid]

Can I assume that you are using only two current loops,say Phase R Current and Phase S Current and

Phase T Current = -[ Phase R current +Phase S Current]

 

[sreid]

Error in the last post. The equation should be

Phase T Voltage = -[Phase R Voltage + Phase S Voltage]

 

[ShamilaWije]

I thought your first equation is the correct one. If the 3 phases are Y connected you just have to apply KCL to the point right?

 

[sreid]

KCL is?

 

[davidbeach]

Kirchhoff's Current Law.

 

[sreid]

Clearly the three currents sum to zero in the motor. But with three current loop drives never exactly sum to zero; offsets and scaling differences. This can result in "Current Fights."

Worse, there are an infinate nubber of voltages on the motor phases that will give zero current in all phases.
0 V, 0 V, 0V will give zero current but so will 1 V, 1 V, 1 V. There are ways to handle this but, IMHO, is to use two current loops and drive the third phase with the proper "Compliance Voltage."

There are then no current fights, 0 voltage equals zero current and you don't have to pay for the third current sensor.

And the Intellmod (Powerex) takes care of short circuits for all three phases so the third current sensor is not needed for that function.

I have thousands of servo amplifiers in the field that are implemented just that way.

I have never seen this in any Textbook but it is commom in the servo industry.

Are you going to Soft or Hard Switch and how do you plan to handle the dead band if soft switching (also called center based PWM).

 

[ShamilaWije]

what is exactly meant by 'current loop'? Is it the feedback loop created by taking the current sensor output?

I guess sreid you are referring to self commutation when you say soft switching am I right? If so I'm planing to do hard switching. I am planing to use a IGBT module. Also a constant deadband will be set.

 

[sreid]

PWM amplifiers today all implement a curren loop to drive the motor phases. The loop is a summing junction with a current command input and the current sensor as the current feedback, A Proportional Integral (PI)controller amplifies the difference between I Command and I Actual to drive the PWM Switches.

Almost all three phase PWM bridges are implemented as center based (Soft) switching. This means that for a zero current command, all three phase legs are being driven with a 50% duty cycle with a small dead band to eliminate "Shoot Through" between top and bottom transistors. This causes a dead band which may or not be important in the servo.

 

[blacksea]

sreid: PWM with 50% duty cycle have "bipolar" method. And more modern servo amp use improved "unipolar" one due to lower current ripple.

 

[ShamilaWije]

I agree with blacksea. There are 2 main variations how you can do PWM switching Bipolar and Unipolar.

In bipolar mode you switch from +V to -V and then again to +V. If your duty cycle is 50% then the average voltage would be 0. Since we are changing the voltage by 2*V the there is a higher current ripple than in the unipolar switching. (EMC considerations)

In the unipolar method you switch from +V to 0 the back to +V. Here to achieve 0 average volts you have to give 0 duty. (less current ripple).

There is a variation of unipolar called Advance Unipolar. This is used in ELMO drives

 

[sreid]

A Star for ShamilaWije, Advanced Unipolar (or what ever a particular vendoe calls it) is the best combination of zero crossover distortion and minimum switching power losses.

 

[ShamilaWije]

Thanks for the star sreid.

By the way I measured the capacitance of each phase of our BLDC motor with a LRC meter and found out that there was about 200nF of capacitance in each phase.

By looking at the current wave form going in to each phase I could see two spikes one when the PWM swithes on and one when it swithes off.

So I tried adding some additional inductances on to each phase. (about 1mH). This reduced the spikes of the current wave form out of the power module.

So thanks again guys for the info.

I'll try to post some results in coming weeks.

 

[blacksea]

ShamilWije: reason of current spikes with PWM switching maybe low recovery of free wheling didodes in parallel with switch. As pls check its recovery time.

 

[ShamilaWije]

I thought voltage spikes are the result of higher recovery time. Anyway I'm using a IPM from IRF IRAMY20UP60. This should be able to go upto 15kHz without any problem.