Looking for a PFC preregulator in the ~1.5kW range

[james_s]

I have a pure sine inverter board that outputs 240V 50Hz from an input of 400VDC. It works fine powered from a variac feeding a rectifier and filter but that requires constant fiddling to regulate the output voltage under varying loads and it's clunky. I'd like to put together a front end to power this from 120VAC 60Hz and an active PFC boost converter looks like a viable option as I do not need isolation. I've seen a few reference designs but they tend to top out at around 300W-600W tops so I'm wondering is there a substantial challenge to scaling this up into the 1.2-1.8kW range? Anyone have any direct experience with this sort of thing? Any particular IC I should take a look at? The UCC28060 looks promising but I don't know whether there's a better solution I should be looking at.

[NiHaoMike]

Just build multiple smaller PFC stages in parallel with the error signal commoned and the PWM phases staggered.

[Pawelr98]

I would look into computer power supplies in the 1+kW range.
Pretty much all of them use a big APFC circuit.
The APFC circuit can be reverse engineered or just "removed" and repurposed (just be sure to make the cuts on the PCB in the right place).
Sometimes such supplies with some kind of a fault can be found on the cheap.

Also, wouldn't it be easier to get some feedback for the inverter itself ?
Inverter without feedback is a poor inverter.
It should be able to regulate it's own output voltage with varying input voltage.

I happen to have an 1000VA UPS unit and it has a simple feedback through optocoupler to regulate output voltage.

I may be no expert on the subject as I have not built any APFC circuit from the scratch but the output of the APFC circuit is DC so using two or three circuits in parallel could be possible.
All would have to be tuned for roughly the same output voltage.

[T3sl4co1l]

Heh, I've got an older design here, some spare boards...  But, I've yet to finish testing, so it'd be up to you to figure out..

(UCC28070 based, haven't figured out yet if the compensation just needs adjustment, or what.)



Tim

[james_s]

I would look into computer power supplies in the 1+kW range.
Pretty much all of them use a big APFC circuit.
The APFC circuit can be reverse engineered or just "removed" and repurposed (just be sure to make the cuts on the PCB in the right place).
Sometimes such supplies with some kind of a fault can be found on the cheap.

Also, wouldn't it be easier to get some feedback for the inverter itself ?
Inverter without feedback is a poor inverter.
It should be able to regulate it's own output voltage with varying input voltage.

I had all of those ideas, however I wasn't sure how well it would work to parallel PFC front ends from something like server PSUs, and reverse engineering a PC power supply sounds like a royal pain in the butt, they're very densely constructed and poorly documented. As far as taking feedback from the output, that might work however I already have the completed and working inverter board and I have put it through its paces and verified that it does exactly what I want it to do when fed 400VDC, so why mess with that? I believe the inverter controller IC does some regulation on its own however it still needs a reasonably constant input voltage.

Also the idea of designing something myself intrigues me as I have never really messed with active PFC, but I'm not sure what sort of challenges I'm likely to face. This will likely be a one-off constructed by hand, not something I'm going to produce.

[james_s]

Heh, I've got an older design here, some spare boards...  But, I've yet to finish testing, so it'd be up to you to figure out..

(UCC28070 based, haven't figured out yet if the compensation just needs adjustment, or what.)

Tim

I'm intrigued, that looks very clean, how far along is the testing? That might be a shortcut if the basic design is more or less sound.

[T3sl4co1l]

Did some hacks to it, last try it blew a fuse (too much startup current?) but I think transistors were okay?  Next test was going to try soft starting or current limiting it.  That was last year.

Tim

[james_s]

Well it sounds like a reasonable place to start anyway, if you've got an extra board I'd be interested in putting one together. I'm assuming that I'm likely to encounter many of the same challenges if I start from scratch, hence asking here in the first place.

[Hydron]

I have some Dell workstation supplies in the ~800W range where the PFC stage is a complete separate board that can be pulled and fired up by itself. The only complication is that the 5V standby supply (also on the PFC board) needs to be loaded a bit - the PFC controller gets it's supply voltage from an auxiliary winding on the primary of the standby supply, so with no load it droops too much (not being part of the feedback loop).

With this sorted all you need to do is short out two pins on a connector that normally goes to the DC-DC board and you're away; the PFC spits out ~390V DC. I've seen this design in two different versions of the supply, and intend to put a couple of the same type of PFC stage in parallel for a project myself (possibly with a controller replacement so I get interleaving and more control).

I suspect there may be other supplies out there with a similar configuration that you might be able to utilise.

Edit: turns out I had taken pics and even made a note to my future self
...though the note references 5VSB while the label on the board only has "+15VSB" and "+5VFP", so maybe not so well done. Vccp is on the connector above Vcc in the first pic.

[james_s]

Do you have a model number for the Dell PSU? I don't want to just buy a bunch of different power supplies hoping to find one.

Tim has kindly offered to send me one of his boards to build however it would be handy to have an alternate plan if that doesn't work.

[Wolfram]

For this power level, a non-interleaved CCM boost converter is a good option. An UCC28180D paired with a nice modern 650 V superjunction MOSFET and a SiC diode would be the simplest solution I can think of. A friend of mine recently built something similar: https://connerlabs.org/?p=404

For the inductor, Sendust or similar alloy powder cores perform very well and are quite cheap, if you can find a supplier that will sell them in small quantities.

[Hydron]

Do you have a model number for the Dell PSU? I don't want to just buy a bunch of different power supplies hoping to find one.

The board pictured earlier is from the top supply in the attached photo, the bottom one is another supply with a similar (albeit slightly less sophisticated) design that would also work stand-alone. If both were available for similar money I'd definitely go for the top one.

I'd guess that the PFC stage from the top supply would be good for 950W or so given the rated output power and a bit of DC-DC stage inefficiency (and note it does need airflow). I haven't looked too hard into the logistics of paralleling two up, but you should be able to read the IC number from previous pics which may shed some light.

Edit: ooh, good to see Steve is back at the coiling stuff; my interest in building/repurposing a PFC is for the same reason. For a larger PFC I'd be going down the DIY route too with superjunction-Si or SiC MOSFETS depending on output voltage and SiC diodes, plus some salvaged inductors and a software based PFC controller. Will need to read the blog posts.

[Pawelr98]

I had all of those ideas, however I wasn't sure how well it would work to parallel PFC front ends from something like server PSUs, and reverse engineering a PC power supply sounds like a royal pain in the butt, they're very densely constructed and poorly documented. As far as taking feedback from the output, that might work however I already have the completed and working inverter board and I have put it through its paces and verified that it does exactly what I want it to do when fed 400VDC, so why mess with that? I believe the inverter controller IC does some regulation on its own however it still needs a reasonably constant input voltage.

Also the idea of designing something myself intrigues me as I have never really messed with active PFC, but I'm not sure what sort of challenges I'm likely to face. This will likely be a one-off constructed by hand, not something I'm going to produce.

Generally speaking even the cheap chinese EGS002 sine wave inverter module which costs 3USD has feedback.
All you need is a bump in duty cycle when the input voltage goes lower.

240Vac has a peak of 340V so the inverter should be running fine for voltages quite a lot lower than 400V DC.

And it may be easier to create a different kind of circuit than APFC to get a nice constant 400V DC.
Voltage doubler from 120V mains should create 340V so we are missing 60V at minimum (under load the voltage will drop down to most likely 300V or so, depending on capacity).
Without PFC functionality this is a fairly easy task as we only need a boost topology or any other type of switching power supply.
Using isolated type (like half-bridge) the circuit only has to handle the 60-100V times whatever current the inverter needs.
1.5kW from 400V gives 3.75A so the circuit only needs to handle 400W at top. That's a far easier task.
And it can be done using easily available parts like pretty much any ATX computer power supply which already has everything you may need to build one.
Or just have a go at 494 or whatever may be your choice.

If the load is slow to change then even mains operated circuit is an option.
A small transformer to boost 120V mains to around 160V (primary to mains, secondary in series with mains, again, transformer only carries the excess power).
Then you only need an SCR rectifier stage.
With an MCU it should be very easy to control as we don't need quick feedback.

Neither of those ideas will be "nice" on mains as neither has PFC.

As for chips.
I have a schematic of a 600W APFC circuit on the table.
Uses NCP1654 along with STW20NK50Z power mosfet.

Also, one thing like to suggest.
Why not use a split inductor (autotransformer) in the APFC stage ?
Lowers the peak voltage at the switching transistor so it can be cheaper and have lower RDSon.
I have read articles about those, mainly for high voltage differences (170V peak 120V mains to 400V I think is quite a lot) in boost topologies.

[cts_casemod]

Hey Hydron,

Are you able to share the Dell PSU model you used? The picture you posted on post #11 seems to be corrupt.
I am also looking for a standalone PFC module I can use to get the regulated 390V rail.

Alternativelly if someone knows of some other model where the PFC bus can be tapped onto, that would be great!

Thanks!

[Terry Bites]

I was just last week reading this.... TIDA-010203. A beast.

[Hydron]

Trying again with attachment (it is indeed corrupted for some reason?)

[T3sl4co1l]

Probably got messed up during the attachmentpocalypse (the [ attach=n ] feature). Were some weird cache issues and a lot of images got brokeded... also why the attach feature was disabled for a couple years, but working again now.

Tim

[temperance]

Heh, I've got an older design here, some spare boards...  But, I've yet to finish testing, so it'd be up to you to figure out..

What are those SMD inductors close to the current sense transformers? Is this some kind of zero current switching CCM PFC with an extra switch and inductor working in DCM mode?

[T3sl4co1l]

Heh, I've got an older design here, some spare boards...  But, I've yet to finish testing, so it'd be up to you to figure out..

What are those SMD inductors close to the current sense transformers? Is this some kind of zero current switching CCM PFC with an extra switch and inductor working in DCM mode?

Who? Oh that was me... you can leave the author in when quoting y'know.  Those were just dI/dt snubbing options, the smaller ones.

Tim

[cts_casemod]

Awesome!

I ended up getting a Dell OU898N power supply, rated for 2.3KW.

I need this PFC to boost low voltage DC (100-120V) to the standard 390V DC Link, so in fact I will only be able to use it for a fraction of it's 'rated' power. I also want it to run... unlike a jet taking off!

This power supply needs a few mods to run below 190VAC and on DC. if someone wants a few tips, please let me know.

I replaced the boost diode with a 16A Silicon carbide, and that reduced the temperature of the mosfets significantly. Next up was the tiny boost inductor. I suspect it was saturating, as it was running arround 100C, so that too was replaced with a larger inductor that can more easily be cooled.

[Hydron]

Great to see you found something that worked for you - I'm certainly interested in any more pictures of the supply or details about your mod, that particular PSU model seems to be available rather cheaply and has a good rating when run on 230V mains supplies.