Float feed back with phase shifting

[goldsmith]

Dear All
Hi !
I have a float amplifier(Differential class D amplifier) and i wanna take a feed back from its out put. i know that i should use a differential feed back (because of that the out put is float) but my problem is this that the out put filters (butter worth ) will shift my phase and when i got feed back at one time my feed back was positive and at the other time my feed back was positive . as a matter of fact my feed back phase is variable and i don't know that what should i do for constant the phase of out put voltage.
Is here any suggestion?
Each reply appreciated .
Best Regards
Goldsmith

 

[goldsmith]

Again Hi
I need urgent help. is it possible that you guide me please?
Thanks
Goldsmith

 

[erikl]

... at one time my feed back was positive and at the other time my feed back was positive ...

With such info from you -- how could anybody help?

 

[FvM]

I think, you are mixing unrelated problems in the question.
- a differential output stage requires a differential amplifier to derive a feedback signal. With sufficient bandwidth, it wouldn't affect the phase with in the feedback operation range
- a feedback loop involves a suitable designed characteristic of the "controller" that achieves positive phase margin of the loop. The delay respectively additional phase shift of the class D modulator has to be considered.

 

[goldsmith]

Dear erikl
Hi
thanks for your reply. my mean is that the phase is not constant per frequency
Again Thanks for your reply

---------- Post added at 15:34 ---------- Previous post was at 15:31 ----------

Dear FvM
Thank you for your attention and reply.
My problem is with my out put low pass filter. my filter created this problem for me that the phase is various . How can i optimize that ?
Thanks in advance
Goldsmith

 

[mtwieg]

So wait, are you talking about a class D RF amp? I don't see how you could possibly use feedback to control it, without a phase/power detector and a PLL or something like that. Could you post a schematic?

 

[FvM]

are you talking about a class D RF amp?

Rather a class D audio amp, I think.

I'm not an expert for state-of-the-art of class D audio. I've seen reference circuits for class D, that connected the feedback path before the output filter, e.g. with National LM4651. In this case, you shouldn't have inductors with considerable non-linearity. You may thus prefer to close the loop over the output filter. But then the loop gain must roll-off with a single pole before the low-pass filter cut-off frequency. Or you need to partly compensate the output filter in the feedback network which gonna be difficult due to parameter variations.

 

[mtwieg]

Okay, if it's a bridged class D (class S, whatever) amp, then you'll need to give some info. What is the desired bandwidth of the amp, what's your switching frequency, and what kind of filter requirements do you have on the output?

Taking feedback prior to the filter is an easy option, but you will still need lots of lowpass filtering on that signal. The only real benefit is that you can make that filter out of active RC filters, which is much easier than the LC amp output filter. But yes, any significant nonlinearity in the output filter will make the feedback irrelevant, unless you can implement some fancy predistortion or something.

 

[goldsmith]

Dear friends!
Hi
Thanks a lot for your valuable replies . My frequency is about 250KHZ . and i used a low pass filter in butter worth arrangement. i think i should usa a RC lpf filter in parallel with the lc filter and then get feed back , is it right? i think in Lc filters out put phase can increase up to 180 degree but in rc networks , out put phase can increase up to 90 degree and it has not problem for negative feed back , is it right?
Thanks in advance
Goldsmith

 

[goldsmith]

Are the things that i said correct?

 

[FvM]

You didn't talk about filter order. 90 degree maximum phase shift refers to a 1st order low-pass. I previously mentioned, that the pulse width modulator introduces a delay, which shows as an additional phase shift. In so far, a 1st order loop filter doesn't guarantee stability, but it should ease to achieve it. The disadvantage of a feedback skipping the output filter is that you loose the option to reduce filter born distortions.

 

[goldsmith]

Dear FvM
Hi
Thanks for your reply.
So what do you suggest here?
Thanks

 

[goldsmith]

Hi every body
was things that i said correct?
Thanks