Boost PFC stage for highly transient load?

[24vtingle]

Hi,
We have designed a new type of Boost PFC stage.
This one has a lot more than the usual 7-10Hz Voltage loop bandwidth.
As such, when serving a constant load, the mains harmonic emissions will be
worse than a "standard" Boost PFC. However, the load that this PFC will serve
is highly transient in nature...just going from no-load-to-full load repeatedly.
No particular frequency, but could be full load for 100ms...then no load for 1 second, then
full load for 500ms......no load for 200ms.....full load for 300ms....etc etc..

So basically, if you add up the mains harmonics over say a 10 minute period...then our PFC stage
has better mains harmonics emissions than a standard PFC stage.

As such, do you think we can be given a special clause in the regulations?

What would be the point of them standards testing our PFC with a constant load when that never happens
in its real useage?

 

[mtwieg]

Harmonics test aside, good luck getting past flicker testing...

 

[FvM]

You don't need to increase voltage loop bandwidth to handle load transients, just add load current feedforward.

 

[24vtingle]

Thanks, but what if you dont know when the load is about to suddenly go noLoad To FullLoad? How can you feed forward then?

--- Updated Apr 28, 2025 ---

Harmonics test aside, good luck getting past flicker testing...

Thanks, you have a great point.
But one would ask, how do high power class D bass guitar amplifiers get through flicker testing?....they are going no load to full load very regularly.

And what about a densely populated city with everyone owning electric cars, and each charger randomly going from no load to full load and vice versa....thats a mega flicker situation...but no-one cares about that...and so no-one will care about any other flicker either.

 

[FvM]

Feedforward means: measure current add it with respective factor to PFC current setpoint. For load step 100 -> 0%, current setpoint jumps to zero.

--- Updated Apr 28, 2025 ---

Flicker is periodical or at least repeated load variation with 0.5 to 25 Hz frequency. On and off switching of EV chargers won't be rated as flicker.

 

[24vtingle]

Feedforward means: measure current add it with respective factor to PFC current setpoint. For load step 100 -> 0%, current setpoint jumps to zero

Thanks...and incidentally, wouldnt that mean that the load isnt drawing a continuous sine wave?....so bang goes the mains harmonics regulatory pass anyway?

Sorry to sound a nawz, but it makes you wonder the following.......so, In which case, whats the point having a "great PFC with added feedforward circuitry ", when the mains harmonics will be no better than if you just have a simple "fast transient response PFC" that had mains harmonics no worse than the "PFC + feedforward system"?
(at least, no worse when the load is highly transient in nature)

(Obviously if the load is constant then "proper PFC" is the way to go.)

And obviously its sometimes good to consider the "ideal", ie, if one had a Boost PFC with infinite capacitor bank at the output.....then you could handle highly transient loads and have great mains harmonics. (just suppose you have some magic way of making the infinite cap bank at 400V).

Many years ago i found out that the entire class D guitar amplifier industry, even those making 2kW bass amps, werent bothering with PFC because you just cant have good regulatory pass-worthy mains harmonics when your load is highly transient in nature.....like a class D guitar amp.
As far as i know, the reg bodies just turn a blind eye to it.

Anyway, if you are powering stuff off a 1kW Mains 240VAC petrol generator, then PFC really does not matter at all.....other than for a constant load, where the reactive current would be less with a "proper PFC"...but would be perfectly satisfactory with our "fast transient response PFC" anyway...which is massively better than rectifier/cap, and not that significantly worse than "proper PFC".

Our "Jack PFC" draws current right across the mains 10ms half cycle, the waveshape isnt a perfect sine, but not that far off, and its symetrical about the mains peak. Incidentally, it gets more sinusoidal the nearer to max power you are. And its way way better than rectifier/cap.....for eg 1kW petrol generator use...the reactive current isnt looking significant enough to be worried about.

 

[mtwieg]

Thanks, you have a great point.
But one would ask, how do high power class D bass guitar amplifiers get through flicker testing?....they are going no load to full load very regularly.

No idea how audio equipment is tested.

Just to be clear, the flicker testing likely wouldn't be an issue for testing on the power supply itself (if one wanted to certify the supply itself). But it would be an issue for whatever device/equipment creates such transient loads and passes them through the power supply to the mains.

Anecdotally, my company recently had an issue with flicker testing, and we talked with the manufacturer of our PSU about it. The engineers I talked to weren't even aware of IEC 61000-3-2, despite having a CE mark.

 

[dick_freebird]

Feedforward "starting now" is still better than later
or none, as regards the error-transient developed
and the time it persists.

 

[24vtingle]

No idea how audio equipment is tested.

Thanks, nobody does. Friends who actually work in the audio industry also tell me this. The audio industry have a silent agreement not to use PFC and not "grass up" each other about it.

But it would be an issue for whatever device/equipment creates such transient loads and passes them through the power supply to the mains.

Yes its an issue with (an alternative to, which ill come to aferwards) ceramic plate hair straighteners where the ceramic plate gets heated up very hot and stays hot....it would be better to change the design and have pulsed firing of resistive heaters....this would save gazillions of energy over the world, but its not permitted to do burst firing of heating elements, since the flicker laws disallow it.....it somehow doesnt seem right because burst fire heaters (only on when needed to be on) would overall save loads of energy.

The flicker laws are also incredibly non-transparent. What frequency of the bursts and what power level is in the bursting?. We tried to do a burst fire shower once and it was no go but trying to proove it was impossible as the standards were so open to misinterpretation...

Flicker is periodical or at least repeated load variation with 0.5 to 25 Hz frequency

Thanks thats helpful, even then it depends on the power and whether the burst sags off or suddenly goes off etc etc.....and what if the flicker is ok because its usually outside that frequency band but is sometimes inside it for certain "intervals"...etc etc...eg a pulsed hair straightener is sometimes inside that, sometimes outside that because the pulsing is just when the human presses the styling plates together....and then admittedly there is pulsing inside that as the heater regulates the temperature....its not a distinct frequency, but sometimes it may end up being a pretty constant frequency for a few pulses say.
Flicker laws disallow it...but try prooving that its disallowed and you have the impossible job.

 

[scopeprobe]

Thanks, nobody does. Friends who actually work in the audio industry also tell me this. The audio industry have a silent agreement not to use PFC and not "grass up" each other about it.

I don't see the sense behind that at all. Most amplifiers would surely work off a rectified DC which would be heavily filtered for fairly obvious reasons. Dynamic changes in current at the output would still be Sinosoid on the input the only thing that would vary would be Power based on the audio volume. Filtering should be preventing the audio coming out on the line. I could see varying quality of power supply being a factor for different systems but to have some silent agreement is a nonsense.

Given audio would be a dynamic load it would seem sensible to me that validation of any amplifier and its power supply would be with a constant tone of selectable volume and frequency and then tested for its performance.

 

[24vtingle]

Thanks, what you say sounds plausible but i am not sure of you are addressing the point that a class D guitar amp, in its real use by say a heavy metal band at a big stage event , will be a very pulsey load on the mains. And flicker laws may well be failed. Also, the current draw from the mains woudlnt be anything like sinusoidal for the duration of the concert....even if a "proper" Boost PFC'd Power stage had been used.

But yes, do standards bodies test the amps and power supply in that kind of scenario...possibly not.....when i worked in audio, nobody knew how this stuff would be tested by the standards bodies.

Basically you find a distributor in the country where you want to make sales...you pay them some money...then they go to someone high up in their government and try and get you an export licence, and then you either get it or you dont.

 

[scopeprobe]

Thanks, what you say sounds plausible but i am not sure of you are addressing the point that a class D guitar amp, in its real use by say a heavy metal band at a big stage event , will be a very pulsey load on the mains. And flicker laws may well be failed. Also, the current draw from the mains woudlnt be anything like sinusoidal.

But yes, do standards bodies test the amps and power supply in that kind of scenario...possibly not.....when i worked in audio, nobody knew how this stuff would be tested by the standards bodies.

Varying in amplitude over time maybe as the output power changes but i'm sure it wouldn't be pulsing or generating high levels of distortion. Theres a reason why the equipment used at events costs a fortune and i'd put money on it being due to their underlying quality and engineering required to develop the solution such as large inductors for low frequency filtering. Most festivals use off grid power plants etcs.. so good power factor, harmonics and distortion is likely a necessity.