Need Help In Fiter Designing Of Three Phase Inverter...:(

[Airian_007]

Hi,
I am currently working on a three phase motor speed control project i have completed most of the programming task i am using pic18f4431..
switching frequency is 16Khz and varying freq is 5 to 75Hz the problem is i want to design a filter at the output to filter the high frequency pwm signal and want to get a pure sinusoidal signal i did this on individual channels by using low power RC filter and i saw the pure sine wave on scope but only one half filter was directly fed by micro controller that's why i said it low power ....but on output stage i dont know how to calculate L(Inductance) C(Capcitance) as i have read many posts recommending for LC filter at the output.
Can anyone please guide me in this regard.
Thanks in Advance.

 

[moffy]

You need to define the current/voltage rating for the filter as this is important in component selection. The formula for the cutoff frequency of LC filter is: Wf = 1/(LC)^0.5. I doubt you will see a pure sine wave with only a single stage, but it should clean up a lot of the PWM.

 

[tcmtech]

If you are running a stock three phase motor with a typical laminated iron core that was designed to work on the normal 50/60 Hz frequency there is no need for the filtering being it will totally ignore that high of frequencies.

 

[moffy]

If you are running a stock three phase motor with a typical laminated iron core that was designed to work on the normal 50/60 Hz frequency there is no need for the filtering being it will totally ignore that high of frequencies.

Yes, very true. But EMI can be an issue if you have long cable runs from the inverter to motor.

 

[tcmtech]

OP didn't specify wire lengths so I have to consider it irrelevant.

 

[MrAl]

Hi,

I would try it without the filters first too. The high harmonics dont just disappear though, they get eaten up by the core as core loss.
Long time ago i was directly involved with the design of a commercial unit and we did not use filters but i dont remember what the wire run was. I do remember that we varied the frequency as well as the output voltage in order to control the motor speed correctly.

 

[Dick Cappels]

I just did this a couple of months ago. Choose a cutoff frequency for your filter that is above the highest frequency you want to pass but far below the PWM frequency. In your case, 200 Hz would give you a reasonably flat response over your frequency range.

Choose a filter impedance, I don't know the drive capabilities of your PIC but would guess that 200 ohm impedance would be fine (It is fine for the AVR I used

The filter looks like this:

PIC PWM output pin => Filter input resistor => inductor =>output. In parallel with the output are the capacitor and the termination resistor. The termination resistor equals the characteristic impedance (assume you will follow the filter with a high impedance buffer). The filter input resistor is R = (the characteristic impedance) - (resistance of the inductor).

This method will give you the flattest response and the lowest amplitude sensitivity to component variation, and twice the pwm ripple attenuation as a simple RC filter with the same cutoff, but the cost is that your signal amplitude would be 1/2 of that you would get with a simple RC filter.

You can find the inductor you need for a given capacitor or the capacitor you need for a given inductor (more likely case) with the formulas below.

**broken link removed**

The characteristic impedance can be found with

Z = SQRT(L/C)

 

[Airian_007]

Thankyou All for your time.
I will try above solutions and current of the motor is around 70amp and 48volts.
I already drove a three phase motor with very less current everything is working fine I will try to build filter as suggested above this weekend and will discuss the results.
Thanks All...

 

[MrAl]

Hi,

I just realized i was assuming that you were using a synthesized sine pattern. That's the best bet here as then you should not have to worry about filtering.
Also as i am sure you know, there is a relationship between the voltage and what the frequency should be.

 

[Airian_007]

Yes I made that lc filter and results are good since my load is inductive too.
So I decided not to use filter for output stage as the motor itself will act as a filter for hiwm frequency pwm.
I need little guidance on snubber as well.
I made rcd 6 of them across each FET.

My ringing freq was about 5 MHz My switching freq is 15KHz and after some calculations and experimental results I used 10ohm and 20n and ringging almost died out but my cap and resistor are getting hot after sometime load is almost 1.5 amp motor. resistor is 1watt and cap is 100volts rating supply voltage is 70volts should I increase the capacitor voltage rating?
What could be the cause...
Thanks

 

[MrAl]

Hi,

What is "rcd" ?

Are you talking about the snubbers across each transistor?
If you measure the cap voltage it will give you a better idea what voltage rating the cap should be. What does your snubber circuit look like?

 

[moffy]

"rcd" at a guess resistor, capacitor, diode. Maybe a peak clamping circuit with the capacitor as clamp and resistor as disipation, perhaps?

 

[Airian_007]

Rcd is resistor capacitor and diode..
Today I gave a go to my motor 2.6kw 48vdc 30v rms.
Current is about 70 amp but initialy I tested on 35 volts and 30amp supply.
Test was successful without filter at output.
Then I just thouth to skip the soft start feature I coded. And motor took almost 60amp and my high side Fet's are dead with a blast which cost me my gate drivers HCPL3120 also.
I have no data about motor and there is nothing printed on plate other than 30v 2.6kw and 60min.I know what this 60min is its a japanies motor.
Kindly help me how to chose the starting freq and max freq for the motor.
My carrier is 16khz should I bring it to 10khz or less. Is it necessary to add soft start and soft stop as well cant I just put a relay on my dc bus and make and brake contacts to start and stop.
Thanks..

 

[Airian_007]

Rcd is resistor capacitor and diode..
Today I gave a go to my motor 2.6kw 48vdc 30v rms.
Current is about 70 amp but initialy I tested on 35 volts and 30amp supply.
Test was successful without filter at output.
Then I just thouth to skip the soft start feature I coded. And motor took almost 60amp and my high side Fet's are dead with a blast which cost me my gate drivers HCPL3120 also.
I have no data about motor and there is nothing printed on plate other than 30v 2.6kw and 60min.I know what this 60min is its a japanies motor.
Kindly help me how to chose the starting freq and max freq for the motor.
My carrier is 16khz should I bring it to 10khz or less. Is it necessary to add soft start and soft stop as well cant I just put a relay on my dc bus and make and brake contacts to start and stop.
Thanks..

 

[MrAl]

Hi,

A slow start mechanism is a must for these kinds of circuits.
Sorry i dont know the min and max frequency for your motor.

 

[moffy]

FETs really don't like overvoltage and will fail instantaneously. With overcurrent it takes a little longer depending on heatsinking. I would be suspicious that the excessive start up current also induced excessive ringing and a possible overvoltage.
No reason to change your carrier, try starting at half voltage/half frequency of your successful test frequency as your start frequency.P.S. What is that frequency? What is your driver configuration? Am assuming 3 half bridge MOSFETs but could you confirm. I know this might sound a bit conservative, but if I was using MOSFETS in your application I would use 150v breakdown as minimum.

 

[tcmtech]

TO do the equivalent of a DOL (Direct On Line) start with a VFD it needs to be massively over sized in the switching device area to typically to handle at least 2 -3 times the voltage and at least 10 times the current of what the motor normally takes at its full rated load.

Looks like you learned the hard way about the necessity for reasonable ramp up/down times and fast fuses when running induction motors off of a VFD.

 

[Airian_007]

Hi.
Current for my mosfet is 98amp and voltage is 200volts.
I think ringging could be the cause of damage..as initially we tested our driver without snubber..
And my configuration is three half bridges.
Motor data is not available its a japanies motor used in cargo trucks ...
Rated for 1200kg max load.
One more thing I have switching supplies although isolated but on the same core..
One for my controller 3 for higmmosfetmosfets and 1 for low side mosfets.
Is this topology ok or should I have separate supplies on separate cores.....
I have already tested my motor with this topology with soft start and that worked well...
One more thing I want to ask about soft start is....
One of my coleague told me that he was working on 80kva three phase inverter....
And in that soft start is like the angles are increased slowly while keeping the amplitude constant....for example initially angle between three phases is same then slowly phase difference is added.
What I am doing is I am slowly increasing my frequency and amplitude at max freq my amplitude is maximum......or rms of the wave is max lets say 30volts...
Is it fine or how should I soft start my motor..
Again thanks 4 replies...

 

[MrAl]

Hi,

I wish i could remember more about how we did this back in the 1980's, but going by the way a magnetic core responds i would say that the voltage goes down with the frequency, so that if the frequency is reduced by a factor A then the voltage must also be reduced by a factor A. So for example if the normal run frequency was 50Hz and the normal voltage was 100 volts, then reducing the frequency to 40 Hz means you should reduce the voltage to 80 volts.
I would not be surprised to see the slow start keep the voltage lower than that requirement for some time also, as it gradually ramps up to meet the frequency factor. So if the frequency of that same motor was 20 Hz and the normal voltage for that frequency was 40 volts, then perhaps starting at 0 volts and ramping up to 40 volts would work well, then start changing the frequency too if needed.

 

[Airian_007]

Thanks MrAL...
Can you kindly explain a bit about the technique you told above...
What would be the impact if I apply full rated DC voltage and then start increasing my frequency...
And which way is better either by increasing voltage slowly (you mean to say adjusting the peak of my wave via duty cycle). In my application yes I have introduced v/f control ....