Simple three phase PFC control topology/IC

[carbontracks]

For a project I've been working on (and for my own amusement) I want to build a 30KW AC-DC converter.  From what I've read, at this power level I'm practically forced to use 3 phase PFC (though I'm don't really care about meeting legal standards for power factor or anything).  Ultimately, I want an output that can range between 100VDC and 600VDC at 50A max, but for now I'm just focusing on the PFC preregulator.  My main issue is that I haven't been able to find any good ICs that are meant for 3 phase PFC, in any topology.  The closest I can come is app notes on how to combine 3 separate single phase parts and merge them together at the output, but this seems incredibly complicated (since you need 3 pfc stages, 3 DC-DC stages, and a power sharing monitor).  I'd prefer using a single stage preregulator, like one of the topologies described in this link:

http://scholar.lib.vt.edu/theses/available/etd-08142002-075617/unrestricted/Barbosa_ETD.pdf

I really don't want to have to control something so complex and powerful with an FPGA, so I'd really like to find an integrated solution for this problem.  At this point I'm willing to bend my specifications for that sake.  Any help would be appreciated, thanks.

[dikris]

there isn't any single IC for 3-phase PFC (operating in CCM) as far as I know. You coul build the unit with a simple IC using a single switch when opeating in so called DCM (discontinuous current mode) but the efficiency won't be great and at the power level you want the DCM is not so good (so the 3 single phase ICs ia the best options you have in my opinion)

[carbontracks]

The more I think about it, the less surprised I am that a single-chip solution doesn't exist.  Simply measuring the instantaneous voltage of each leg (without a neutral line, which is common) would require a lot of external circuitry, and then every gate driver would have to be isolated... sounds like a nightmare.

So I think I'm looking at two options.  One is to have three separate bi-phase PFC.  I really want to avoid also having three separate DC-DC converter whose outputs are summed.  Pages 22 and 23 of that PDF give an example of how to modify the front end so that this isn't necessary; just an extra diode in each boost stage.  But I'm wondering if that diode will interfere with the operation of each PFC controller... 

The other option is to use a 3 phase rectifier and then PFC the output of that with a single boost circuit.  I won't get a great power factor, since each line will only be conducting when above half its amplitude.  Still better than rectifying straight into a big capacitor bank...

[solutions]

I may be missing something here, but can't you just combine the three phases with diode rectifiers and voila!, DC? It's that 120 degree phases apart then add sine waves stuff from school that comes to mind.  No caps, PFC = unity.

[carbontracks]

I may be missing something here, but can't you just combine the three phases with diode rectifiers and voila!, DC? It's that 120 degree phases apart then add sine waves stuff from school that comes to mind.  No caps, PFC = unity.

if you rectify all three phases together, then the output of the rectifier never reaches below half of the lines' peak voltage and therefore you won't be conducting for about half of each cycle.  if you threw a pfc on the output of the rectifier, it would "work" in the sense that you would draw current proportional to the rectifier output, but it wouldn't be the same quality as rectifying each pair of phases.

[solutions]

A paper for you if you need control over the output voltage  http://leiwww.epfl.ch/publications/herminjard_zimmermann_monnier_epe_99.pdf

[dikris]

carbontracks, your option 1 should work

[carbontracks]

I may be missing something here, but can't you just combine the three phases with diode rectifiers and voila!, DC? It's that 120 degree phases apart then add sine waves stuff from school that comes to mind.  No caps, PFC = unity.

I've looked at the vienna topology, and it certainly has its strong points, but it seems that the control method necessary won't work with typical PFC controllers.  I'm not sure about this, because I can't come across any info where the control circuitry is explicitly described in more than pretty formulas or block diagrams.  I have no idea how such control schemes are implemented in hardware.  On the other hand, doing three separate bi-phase stages will definitely work with off the shelf controllers.

carbontracks, your option 1 should work

That's what I'm leaning towards right now.  I'm going to try to implement the method in the link I posted, which looks like the following:


I've been able to simulate one biphase PFC in LTspice (using the LT1249 controller), but it can't handle doing any additional phases simultaneously.  Not sure whether it's a limitation of the simulator of something wrong with the way I've set things up.  But in any case, I'm still not %100 convinced that simply putting that diode in each boost stage allows them to work directly in parallel like that....

this would be SO much easier if I could rely on having a neutral line available.  Then that would be an easy reference for all my control crap.  But from what I've heard, the neutral is often not available (since ideally it's not necessary).

also I just found this link: http://www.powerpulse.net/techPaper.php?paperID=140&print
It says that the control circuitry for the vienna topology is more complicated, as I would expect.  Also, it says that a neutral is actually required to connect to the middle of the DC bus... though I think there are ways around this.  In any case I don't think the vienna topology is my answer.

[carbontracks]

So I've tried numerous ways of making the first approach (the one in that picture) work out in simulation.  I'm pretty confident that it simply can't and the paper is wrong.  The only way it could work is if I had transformers isolating the outputs, or the line voltage, and I'm trying to avoid transformers at all costs.  So it looks like option 1 won't work out for me.  So I'm going to fall back on option 2, which is to have one PFC stage operate on the output of a 3 phase rectifier.  It won't get good power factor off a 3 phase input, but I'll just have to deal with that (however it will get a good PF with a single phase input, which is nice).  I'm planning on using a UCC28070 PFC controller, since it has two interleaved boost stages, making inductor and switch design simpler. 

[krish2487]

Hello
Let me introduce myself.
I am a newbie here and the topic just caught my eye.

I m basically an embedded systems and a power electronics engineer with emphasis on UPS systems.
I have about one and a half years experience in this field.

I am not sure if carbontrack is looking for this solution but here it goes.

Usually in my field when the power requirement is this high with power factor being a critical consideration then we go for a 12 pulse or a 24 pulse rectifier.
I wont reiterate what others here know already(better than me , i must add) but the idea is this-

Use the three phase supply (Wye) and form a Delta connection using the same supply. The output from these two discrete sources(Wye and Delta) are given to two independent 3 phase bridges consisting of (usually) IGBTS. The outputs of the two bridges are parallelled.

The IGBTS are triggered accordingly in a specific manner. The power factor with a 12 pulse rectifier is about 0.8 and as the number of pulses increases the power factor increases.

But care has to taken not to cross conduct any arm at any time.
I am afraid i have no circuit or paper solution to present
googling for 12/24/48 pulse rectifier and/or aircraft pulse rectifier might yield more fruitful results.
 

[carbontracks]

Hi Krish, thanks for the reply

Usually in my field when the power requirement is this high with power factor being a critical consideration then we go for a 12 pulse or a 24 pulse rectifier.
I wont reiterate what others here know already(better than me , i must add) but the idea is this-

Use the three phase supply (Wye) and form a Delta connection using the same supply. The output from these two discrete sources(Wye and Delta) are given to two independent 3 phase bridges consisting of (usually) IGBTS. The outputs of the two bridges are parallelled.

So you need a transformer to carry the input power and give delta and wye outputs, correct?  Getting a 3phase 30KVA transformer would be a big pain, especially since I want the whole supply to be somewhat lightweight (like under 200 pounds). 

The IGBTS are triggered accordingly in a specific manner. The power factor with a 12 pulse rectifier is about 0.8 and as the number of pulses increases the power factor increases.

But care has to taken not to cross conduct any arm at any time.
I am afraid i have no circuit or paper solution to present
googling for 12/24/48 pulse rectifier and/or aircraft pulse rectifier might yield more fruitful results.
 

Yeah I've seen a good amount of literature on controlling three phase rectifiers.  Thing is, I'm not incredibly confident in my ability to code the control loops into a DSP or FPGA, so I wanted something that could be implemented with an off the shelf SMPS controller chip.  For that reason, I think a "true" three phase PFC isn't going to happen, so for now I'm sticking to a single phase PFC which can take single phase in and give a good PFC, or three phase in and give a poor PFC.  Maybe in the future I'll try to build a small vienna rectifier or something.  I think it would be a lot of fun.

And as for the status of this project:  I've prototyped the PFC preregulator stage, and it seems to work fine with single phase input at low power (under 1KW).  We ran into a problem in that the inductor cores I bought are way too lossy to work out (made out of powdered iron toroids).  So we're currently trying to make new inductors out of gapped ferrite cores, which are significantly more difficult to build (but should perform far, far better).  Hopefully we'll be up and running within a couple weeks, though.

[krish2487]

Well
in that case i think(i m not sure how practical it is )

you can use a regular off-the-shelf single phase pfc chip like 3854 independently for all the three phases and OR the outputs
then use load share controllers to force the pulse widths for the pfc regulators so that any one of the phases are not unduly loaded, consequently that the magnetics are never abused.

just a thought OTOH

as i mentioned, i have no idea how practical this approach is...

Posted on: November 13, 2010, 03:35:13 15:35 - Automerged
alternatively
if the magnetics size is not a critical factor then you can replace the IGBTs with diodes. The power factor will not be 0.99 but better than a regular three phase rectifier.

[carbontracks]

you can use a regular off-the-shelf single phase pfc chip like 3854 independently for all the three phases and OR the outputs then use load share controllers to force the pulse widths for the pfc regulators so that any one of the phases are not unduly loaded, consequently that the magnetics are never abused.

This was my original thought, but I do not think it is possible without isolation transformers, since each phase needs its own bridge rectifier, but the outputs of those rectifiers will not have a common ground (and therefore can't be summed into one DC output bus).

if the magnetics size is not a critical factor then you can replace the IGBTs with diodes. The power factor will not be 0.99 but better than a regular three phase rectifier.

I think you're describing the approach I'm currently taking, which is to do a single PFC stage off the output of a three phase bridge rectifier.

[krish2487]

actually what i meant was
that you do use the input side transformers and use two independant 3 phase bridges but replace the IGBTS with diodes.
ie generate a star and delta supply nevertheless and then use two 3 phase bridges and parallel the outputs

[shams_iqbal]

hi dear this is post i joined this forum my most recent work area i studied diffrent ways of pfc but still didnt got enough info about vinna rectifier and thinking to use a dsp controller for vinna rectifier  or some trickky way to build this by using nutrel and three phase to collecting splited dc with common nutrel and using two pfc regulator both for negitive and positive side this will give me +400 vdc and -400vdc with ease so what u guys say about this i am almost sure i tried on psim6 but due to sumilation problem didnt got result with this way so requierd real way work ship