Burst mode PSU and EMC?

[cupoftea]

Hi,
We have an offline PSU with 40vout (39-41V). Its in Hysteretic mode….hysteretic comparator “looking” at vout.
Hysteretic comparator goes via Digital isolator to ON/OFF the primary FET driver/controller. When fully loaded at 400W, then it switches all the time, continuously, and gives around 40Vout…..when loaded at say 200W, it switches in Burst mode , ayk, obviously. When in a burst, the peak current is the same as when on continuous 400W,
Our consultant tells us that when its bursting ON/OFF….etc, to give just 200W output…then its conducted and radiated emissions will be worse than when on continuous switching with 400W load.
Surely this could not be so, would you agree?

 

[dick_freebird]

I disagree if you are talking about a classic spectrum analyzer
which averages power as it sweeps through.

But I could see some possibility of intermodulation products
which might throw you a surprise depending on amplitude
and frequency, of.

And when you get to single pulse variable frequency the
question about "which side of the frequency mask's
allowable-amplitude-transitions do I fall on, in this moment?"
that could hold some surprises as well. Maybe the comparator
/ isolator is too slow for variable frequency single pulse.

But in some recent charge pump work, I've observed that
bursting does increase output voltage ripple over continuous.
If VOUT ripple is the main source of your EMI then maybe
there's an issue.

I kind of expect that the full-scale voltage on the switch node,
(spraying E-field) and current spikes on the edges (spraying
B-field), are more likely the deal.

 

[cupoftea]

I kind of expect that the full-scale voltage on the switch node,
(spraying E-field) and current spikes on the edges (spraying
B-field), are more likely the deal.

Thanks, this is exactly what i think about it....and as such, i wonder why Burst mode type PSUs are so villified?
They have fantastic no Load to Full Load transient response, and vice versa.

 

[cupoftea]

Though a story is becoming...because its amazing how a massive number of Power electronics company staff wrongly think that a burst mode SMPS is worse at EMC , than a standard frequency compensated SMPS.....which if course, is totally not true....neither does it fail flicker.

I once worked at a co, where they had a 600W UV lamp, which was on a gantry going back and forth, every time it hit the gantry end, the 600W was suddenly turned off, then suddenly turned back on again some 2 seconds later......the full load transients made the output go up to >60V.....and it eventually blew up the control PCB.......they were asked why they didnt use a Burst mode SMPS....which would have alleviated the problem....they said , because Burst Mode would fail EMC....which of course, is nonsense...they are pretty much just the same at EMC as a standard frequency compensated SMPS.

(ayk, a Burst mode smps, which has a hysteretic comparator at output for regulation.....has a certain fixed peak current setting.....if you incrementally increase its load...you will eventually get to a stage where its output drops a little , and it goes into continuous mode..not bursting.............if you use it for a power level which is just a bit lower than dropout, then it is in burst mode, but its peak current level is pretty well the same as what you'd get with a frequency compensated SMPS.)

Also it does not fail flicker.

--- Updated May 16, 2023 ---

Take the following nonsense written about Burst mode...

https://www.electronicproducts.com/burst-mode-operation-a-double-edged-sword-for-power-supply-manufacturers-users/

They say the filter woudlnt work....burst mode switching in bursts at a certain frequency will give no problem to a standard emc filter.

 

[cupoftea]

Also, an offline Burst mode PSU can avoid serious output voltage overshoot and undervoltage following transients.....and do so with extremely minimal output capacitance...not so for a frequency compensated SMPS with an opto in the feedback path.....Burst mode 2 tran forward , 400W..and you dont even need output 'lytics...low value film cap will do.