The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> High-Power Design >> Randomization for PWM DcDc-Converter
https://designers-guide.org/forum/YaBB.pl?num=1219666020

Message started by Berti on Aug 25^th, 2008, 5:06am

Title: Randomization for PWM DcDc-Converter
Post by Berti on Aug 25^th, 2008, 5:06am

Hi all,

I would like to reduce EMI of an existing PWM DcDc-Converter.
Therefore I added randomization, which improved EMI. However, the randomization
signal appears at the converter output (see figure).

Does anybody has suggestions how to improve that?

Thank & Regards

Title: Re: Randomization for PWM DcDc-Converter
Post by NiceFeng on Apr 12^th, 2009, 5:16am

Hi, Berti,

How did you modulate the duty cycle?

I think your output waveform is related to the period of your modulation signal. It seems it is too fast.

Regards,
Feng

Title: Re: Randomization for PWM DcDc-Converter
Post by Berti on Apr 14^th, 2009, 11:27pm

Hi Feng,

The period of the modulation signal is given by the length of the LFSR which is 8bit in my case.
However, in the meantime I found several publications which confirm the presence of low frequency noise if randomization is applied. And since the DcDc-Converter output is finally regulated by an additional LDO this is not really a problem. Especially since the PSRR of the LDO is very good at low frequencies.

Thank you. Cheers.

Title: Re: Randomization for PWM DcDc-Converter
Post by krab on Jul 8^th, 2009, 1:30am

Hi Berti,

I have just started a DC-DC IC design where +-3% variation in the switching frequency is needed for spread spectrum control.

I considered using a 8 bit LFSR just like you but I am wondering if the period of the register (8 cycles) is too short. I plan to use the current clock edge to drive the LFSR and the new state of the LFSR will determine the next cycle frequency.

I am curious, after your research what is the reason why you have low frequency noise on the output?

I am also considering maybe using a low frequency sawtooth to modulate the duty cycle, but I am not sure how I should choose the freq of my sawtooth. Any suggestions?

Also any references on how one might implement this sawtooth modulation on chip will be much appreciated.

Cheers

Title: Re: Randomization for PWM DcDc-Converter
Post by Berti on Jul 9^th, 2009, 12:14am

Hi,

Quote:

I am curious, after your research what is the reason why you have low frequency noise on the output?

I came to the conclusion that the low frequency noise is inherent to randomization. I found several publications which report on that. Basically it makes sense: If you add dither to a system, the dither will typically also appear at the output.
However, since the DcDc-Output is further regulated by an additional LDO in my case, low frequency noise will be suppressed by the LDO.

Quote:

I am also considering maybe using a low frequency sawtooth to modulate the duty cycle, but I am not sure how I should choose the freq of my sawtooth.

The choice of the switching frequency is related to your system (spectral interference planing). Often a low frequency is preferred, which however translated into a large inductor.

Cheers