Signal noise in Boost PFC?

[Faringdon]

Hi,
{EDIT....if you are a top  power supply  consultant, seeking interesting posts to enhance your horizons, then this post is not for you..this is just a PCB layout issue....END OF EDIT}

Doing 1kW Boost PFC  with UCC28070A. (390vout, 220vac, 60kHz) The current sense is via 50:1 Current sense transformer (CST).

{EDIT-----I suspect the PCB layout is bad, and wish to confirm this with a mod wire.....prior to getting the PCB re-layed out.....---END_OF_EDIT}


The CST (and its reset/rectifier circuit) is some 15cm away from the Control PCB containing the UCC28070A (burden resistor is near controller). To save complexity, I just routed the Current sense current to the controller with a PCB trace over its ground plane. On second thoughts, the power stage switching currents flow in this ground plane aswell, so it’ll be noisy. As such, I think I will “augment” the current sense tracking with a twisted pair kynar wire from the CST reset circuit, to the controller. (ie, some 15cm).
Do you think this will cause more noise issues?
I can’t find out by scoping it, since all I have to probe the burden resistor with is a Micsig DP10007 diff probe…..and with its two huge “dangling clips”, the waveform from it is obviously just common mode trash filled.
I really need an isolation transformer, but it would need to be rated  for 1KVA and they are expensive. AYK, With isol  tran I could use a coax probe direct into the scope and get a great waveform.

Though anyway, what about the twisted pair kynar?
Surely just as good as if i had used star grounding of the CST output/GND on the PCB , back to the controller?

UCC28070A
https://www.ti.com/lit/ds/symlink/ucc28070a.pdf?ts=1660244875404&ref_url=https%253A%252F%252Fwww.google.com%252F

[jonpaul]

sorry entire post and project makes little sense.

Why do you waste our time?

jon

[Faringdon]

OK, thanks anyway,

I suspect the PCB layout is bad, and wish to confirm this with a mod wire.....prior to getting the PCB re-layed out.....

Its talking about taking the output feedback divider resistor point back to the controller with a twisted pair. (instead of PCB tracks).
On the PCB, the feedback divider point wasnt taken back to the controller as a star-grounded pair....so i am thinking of doing it with an externally added twisted pair wire.
(ie fixing a layout issue with a mod wire , to confirm it works, prior to getting the PCB re-layed out, with the preferred  star-grounding method.)
What i am wondering is, will the twisted pair mod wire bring loads of noise issues of itself.?

(i am actually  thinking of having mod wires for all three feedback signals, VINACP, VOUT and Current sense. At the moment, the ground of these signals is just the ground plane, going back to the controller. -I thought that would be sufficient, but  it seems not.).

(Learned Info for anyone doing 1kW Boost PFC for the first time.....just do a single board first, containing just the Boost PFC on its own, so that you can get that bit right first....the size of the power components at 1kW, mean that you cant get tight layout like you can with a much lower powered converter, so PCB layout things, like  having a separate star ground for the feedback signals, apply......ie, not possible to just rely on the ground plane.)

Why do you waste our time?

Apologies Jon Paul, i have edited the top post now, so that top consultants such as yourself, no longer waste their time here.

[jonpaul]

Farrington your PCB note needs photos of layout and schematics to get any cogent response

j

[Faringdon]

Farrington your PCB note needs photos of layout and schematics to get any cogent response

Thanks Jon Paul,
If anyone wants, i will send them the Layout/Schem  file for the 4 layer main PCB,  and the three daughter PCBs (2 x double sided, 1 x 4 layer)......then the kind person would have to work out how they fit together, (which is certainly possible due to lots of helpful silkscreen annotation), then they would have to  evaluate, after they have downloaded an Eagle Pro 7.4.0 viewer. Sorry, i dont believe anyone would want to bother with all that, as such, the above question of the principle, but please pass  it up if its too much time, or not felt appropriate..
In any case, for the next PCB, i have decided to take the current and voltage feedback signals back to the controller using dedicated ground/signal pairs, instead of just relying on using the two ground planes.

[jonpaul]

Farrington:  That is one of many engineering issues you seem to beg to be  solved for free, rather than paying for expertise or learning the trade.

Very disappointing waste of everyone's time...

Either  Just post everything as images or exit the forum and finally do your own learning and research.

no one else will do that work for you.

j

[Faringdon]

Thanks, would you agree that realistically there is only one single way to find out whether diff-pair'd signals, or signal-over-ground-plane , is the best way to go?
No book can tell this....
...Only doing 2 PCBs, one with diff-pairs, and one with "signal over ground plane", and comparing the results, will tell  which one is best, or whether both are workable.

And also Only a person with significant experience in power electronics could  have sufficient knkowledge to offer insight on what might be the right way to go here.......the company i work for cannot afford a power electronics consultancy as there market has been hit by the economic crash.
So far , 307 People have  presumably googled into this thread, and presumably, they are interested in this issue, which is good for the forum?

I have put a warning in the top post so the true consultants such as JP, dont have to  waste their time here.

Sorry but if i post images it wont be followable.....the quickest way would be to use the free eagle viewer...if somebody wants i will post the layout files here.

If im honest , the place i work for is going to be purchasing the power supply from China. -But they kind of want to get into it a bit so they kind of have more idea what they are buying. Eg, if we pay for the layout and  schems, will we be able to realistically modify it ourselves in future etc etc, or get someone else to do it perhaps even, if it was  a small mod. We have to purchase from China, we cannot afford any other  way. But its wanted to get a feel for what we are buying, and we cannot afford to pay a consultancy.

If we can cobble something together ourselves we will at least get some idea of BOM cost, and whether we are being overcharged, and why, and what for,  etc etc.

At the moment we cant even find consultants that are available.......they all seem to be on military projects, for which they are being payed enormously...way more than we can afford.
I imagine a great many of the 307-so-far viewers to this thread are in the same boat as us.

[temperance]

On the twisted pair / coax subject: the answer you are looking for can be found in any good linear circuit analyses book. The question you have to answer goes like this: how do the elements in my board layout interfere with the error amplifiers in my PFC controller.

Besides, when you started this project you've stated it's for yourself and you even have to solder the board together in a garage somewhere. Sometimes you write in "I" form and other posts but later on you switch to "we". Now you are stating this is for a company looking for lowest possible BOM cost so they can compare this with offers from China and the don't have money to spend on an engineering consultant.

So, this company wants a >1KW SMPS at the lowest possible cost in low quantities because if it would be high volume, they would have a budget to spend on engineering. Something doesn't add up here. But as usual, you will ignore such comments and questions raised by others and ramble on in the "would you agree style".

In low volume productions product price is only a small part of the total cost. If I where to evaluate offers from China I would prepare a very different set of questions. Quality assurance will quickly eat your budget.

Been there done that when my customers end up in misery. I have some pictures of a burned down garage if you like caused by a 300W battery charger. Or pictures of a resonant 800W low cost SMPS which all failed with IGBT's exploding at random. Reputation is a very expensive thing.

[NiHaoMike]

Been there done that when my customers end up in misery. I have some pictures of a burned down garage if you like caused by a 300W battery charger.

That was a common problem with universal R/C chargers. It sometimes happens when the user misconfigures the charger or the charger has a fault that prevents it from stopping when it should. And R/C packs usually don't have BMS so no protection against overvoltage.

[jonpaul]

Having  responded to his notes for many such threads, its an Pity that Farrington hides his real agenda (if any) and wastes our time. 

Seems like his ability to think for himself, and  work thru a design logically is impaired.

Since regardless of our efforts and assistance he never  finishes a project and just continues with more questions.

Sorry to waste more time but I would like to get to the root of the problem...perhaps it is not an electronic but a psychological issue ?

Bon Soirée

Jon

[Faringdon]

Thanks,
Just to re-iterate, the topic is about  ~1kW Boost PFCs, and whether or not you aim to take the feedback signals (in this case from vout divider , vbus divider, and current sense transformer) as a  PCB trace over a continuous ground plane, in each case, or whether you aim to take these signals back as signal pairs, ie, each with a dedicated  ground trace, going back,  and star grounded to the controller ground? The controller used here is a UCC28070A, but AYK, can be others.

...And may i ask, what about the sacriligious situation of using just a double sided PCB, and taking the mentioned feedback signals back to the controller with their own twisted pairs?

perhaps it is not an electronic but a psychological issue ?

I can assure you, i have a row of laughing potatoes lined up  on my window sill, each of whom assure me that i am perfectly sane!

[temperance]

I will restate things in your own language.

The answer you are looking for can be found in any good linear circuit analyses book. Would you agree?

@NiHaoMike
that wasn't the problem.

@ jonpaul
I think you will never get to the root of the problem.