Resonance of LC output stage of Full-bridge?

[eem2am]

Hi,

I was recently sacked from my job as a SMPS designer at a new start-up Class D amplifier company

Was i sacked because my voltage mode full-bridge SMPS with +/-40V rails would have resonated at the output LC resonant frequency when audio pulses at that frequency were drawn from it?

It was a dual rail output , so there were two LC's.

Each individual LC had a resonant frequency of 630Hz.

Vin was = 90-265VAC...it had a voltage doubler link for USA/Japan mains...(no PFC).
Output power was 330W.
Type 3 compensation.

For the first few months they were lauding me as there new messiah......then all of a sudden they were ranting at me as if i'd just stolen the whole year's profits...then sacked me.

I think that they showed my schematic to a SMPS colsultancy...and the consultancy told them something.....but what they said i dont know.

Do you think it was the danger of the resonating LC output stage of the Full-Bridge?

[!]

Maybe they found it wasn't suitable for Europe with no PFC?

It all seems very strange, they should have had a list of requirements that if met, completed the job. Then it's a question of whether you are their employee or a contractor. If your design did not meet the requirements, they should have clarified them and you change the design. I don't think it had anything to do with where your design would resonate, unless you stated something in error then only after production began and there was time and money lost on a product that didn't work, should they have no tolerance - but still tell you why you are terminated.

Why not just ask them? Something does not add up, people are seldom fired without any cause given. Perhaps there is nothing wrong with the design and they just decided it was cheaper not to pay you anymore once they had it.

[RJM1]

I don't know the design specs. for your power supply but 630HZ seems way too low.
You might want to look at this app note for type 3 compensation.
http://www.google.com/url?sa=t&rct=j...OxaxdQ&cad=rja

[Eva]

Designs like yours have been around since mid 1980s, and yours does not seem to have anything innovative in it. There are thousands of SMPS (and designers) like that.

On the other hand, single-switch integrated-magnetics PFC stages with universal input and isolated output are becoming more popular, and they can be less expensive to produce and far more efficient than conventional push-pull SMPS. In fact, I'm researching this topology for a new line of class D modules intended to have a very good features/performance to cost ratio.

[DF96]

You certainly don't want anything in a PSU to resonate anywhere in or near the audio band. You should have realised this and checked for it. How many years experience do you have in designing PSUs? I noticed that you have been asking lots of questions on here about SMPS etc. Was your employer aware of this?

[eem2am]

Thanks to all..... of course it was always the company's right to sack me at any time they chose, i've no problems about that.

Eva:
I appreciate your comment that it was not innovative, and i am relieved that you are saying this kind of thing is usual, since i really was aiming for the most common, cheapest possible design.

The customer didnt want PFC becuase they assumed extra expense for it.....i must admit, i am wondering if a Boost PFC stage followed by a current-mode two transistor forward would have turned out cheaper.?

RJM1
thats a great app note, though my design was TL431/Opto isolated, so i used the formula on page 292 of basso to get the error amp transfer function.......i used the formula on page 226 of same book to get the full-bridge small signal transfer function.

...then got bode plots from these two.

of course when i did atan to get the phase, i get discontinuities (sudden phase jumps) , i think due to my use of the atan function...so i think i need to adjust the phase angles accordingly.

PFC is not a requirement for audio under 600W anywhere in the world....as long as you pass mains harmonics at 1/8 audio its fine.

[Eva]

Single switch PFC with isolated output can be cheaper, smaller and more efficient than other approaches with 2, 3 or 5 switches. Switch and transformer utilization is very good. There is even the option to use a bidirectional primary switch (back to back MOSFET pair) to get rid of the diode bridge completely and gain extra efficiency/power.

Any power filter resonance in the audio band is not a problem as long as it is damped, either by all parasitic resistances or by a regulation loop.

[nigelwright7557]

Working for none technical people is always a minefield.
Just what hapens if for instance someone criticises one of your designs ?
Do they believe you that the design is good or believe the other person ?

I was in a similar situation to you in that I did some design work for a company then was paid off. I found out from anotehr employee that I had cracked all their problems with my new designs and was no longer required ! I had put myself out of work by working hard and doing the job right.....

[eem2am]

Eva:
I see what you mean now about single switch PFC....do you refer to the signal stage Flyback PFC?

I thought about this, but the thought of such a flyback , with my output power of 330W.........well i wondered about the peak current levels, and the EMC situation?

I see that Onsemi chip called NCP1651 is single stage flyback...

http://www.onsemi.com/pub_link/Collateral/NCP1651-D.PDF

....but it says its only for up to 250W, so i shyed away from it for this 330W application.....maybe i was wrong here...the customer certainly was encouraging me from the start just to do a cheap flyback.........but i declined for the reasons i have given.

....though it haunted me throughout....and i kept thinking that with big enough heatsinks it could have been workable?

[darkfenriz]

Man, consider yourself lucky, find yourself an employer with considerably better mental stability.

Considering the SMPS, the resonance can be very detoriating for the amlifier, especially if it is rated for 'just for' the voltage provided by a SMPS.

It could be damped with proper control loop tuning or even transfering the poles/zeros in front of an optocoupler ( so that thenonlinearities transfer just feedback factor).
All in all I believe you were very close to making it dead stable. That's what TYpe 3 compensation was invented for!