Isolated PFC Topology choice, 800W. - diyAudio
Go Back   Home > Forums > Amplifiers > Power Supplies
Home Forums Rules Articles diyAudio Store Gallery Wiki Blogs Register Donations FAQ Calendar Search Today's Posts Mark Forums Read

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Thread Tools Search this Thread
Old 24th May 2010, 07:44 AM   #1
Slobban is offline Slobban  Sweden
diyAudio Member
Join Date: May 2009
Default Isolated PFC Topology choice, 800W.


just completed a 5kW powersupply for our high power product, altough my company decided that our next product must be cheaper! currently my company has focused on high performance, and not low cost, and I'm not very experienced in low cost segment.

so to make an isolated, regulated, universial mains supply with PFC, +/- 100v output and about 800w continously.

the cheapest possible way must be making a single stage with PFC and isolation in one converter. and as I see it, there is only 2 topologies feasable:

- flyback, suffers from high peak currents, high voltage stress, large input filter.
- isolated boost (current fed boost-push pull). snubbing? is this even possible? suffers high voltage stress, input current rush needs to be handled.

I'm kind off liking the idea with the isolated boost, since it's so simple in theory. thinking of having to add the buck transistor but only using it during startup, to relieave the design of the usual startup resistor and releay. the buck transistor might be a simple P-FET driven at low speed (20kHz) and could also be used as on/off switch. up and running the reflected output voltage is about 400v. using a simple Atmel 8bit processor for the Avg.mode current control. Im thinking to not use a flyback winding on the inductor, the stresses on the transistors are great enough as it is. 1200v igbts will be used to handle the basic 800v stress +spikes. all the control and regulation is the easy part.

for cost, I think that the input inductor added in the isolated boost is proberly matched by the cost of the DM coil needed as input filter to the flyback. the dual transistor is no problem since the current stress is less.

during simulations, I found that the push pull is actually easier to snub than the full bridge variant, mainly since during cross conduction the leakage in the primary windings actually carries half the current already, which is not the case for the full bridge. meaning that only half the current adds to resonances.

however, even when setting the coupling to 0.998 (leakage of 1uH) in SPICE the amount of dissapation in the snubbers are still like 50W, and it's abit too much to use in an enclosed box.

is there any other cheap topologies possible to use? is the isolated boost really feasable for this kind of design? is there any way of clamping this design without involving floating drivers and such?

I realize that the best design is proberly a boost PFC with a LLC resonant halfbridge, but my guess is that it will be too expencive. it would have 3 magnetic component, together with 3 switches, and some king of inrush limiter.

  Reply With Quote
Old 24th May 2010, 06:15 PM   #2
luka is offline luka  Slovenia
diyAudio Member
luka's Avatar
Join Date: Sep 2005
Location: in Slovenia :)
Send a message via Skype™ to luka
Do you think you could send me some pictures of that 5kw model?
home page @ @ 24/7 all year long
  Reply With Quote
Old 25th May 2010, 08:05 AM   #3
Slobban is offline Slobban  Sweden
diyAudio Member
Join Date: May 2009
my precious work is owned by my company, I could never send pictures on this out, for many reasons.

however, the supply is pretty simple. the PFC is a buck+boost type (dual ramp control, avg current mode, hard switching), delivering a maximum of 5kW to a cap bank (400v 10mF). the cap bank stores the peak power. it is allowed to be discharged to 50% of the voltage during the bursts (thus utilizing 75% of the energy stored).

next is an isolated phase shifted full bridge, generating the +/-200v maximum 300A pk. it can delivier 60kW pk. there are no real energy storage after the DC/DC converter so it handles the audio program up to about 2kHz (then the LC filter takes over and the music peaks are averaged), meaning that to be able to produce a 20Hz sine wave of 30kW the peak power will be 60kW.

but basically it's no magical supply, it's just textbook examples of designing DC/DC converters.

the PFC topology is basically ripped off the web (buck and boost using the same coil), but we needed to make our own regulator, since we didnt find any ic to control this type of PFC.

the DC/DC topology is widely known on the web. sadly it could not be afforded to utilize the resonance during turn on, since added leakage in the transformers would yield too much snubbing on the secondary rectifiers (yea I know, it could have been a discrete inductor with clamp diodes but it did not fit), altough we benefit from the topology very much during idle (discontinous operation). the supply only consumes 10W in idle which is extremly good for this kind of high power supply.

the board measures 425 x 185 mm (height ~70mm) cooled by 3 60mm fans (yes, it gets pretty hot when driven to the maximum).

  Reply With Quote
Old 25th May 2010, 09:51 PM   #4
diyAudio Member
Join Date: Jan 2007
I was just working on something that reminds me of your problem. Mine was converting a 60hz wall wart to a float battery charger. But the problem was absorbing the spikes from the leakage inductance flyback in the power transformer. You might be able to do something vaguely similar even though your frequency is much higher, requiring much more complexity. The spikes are transfered to the output load instead of being dissipated. Probably would need a gate protection zener on the MOSFET gate.
Attached Images
File Type: jpg 1amp-batt-chgr.jpg (81.1 KB, 284 views)
Attached Files
File Type: zip (1.2 KB, 47 views)

Last edited by Electrone; 25th May 2010 at 09:58 PM. Reason: added comment
  Reply With Quote
Old 27th May 2010, 09:46 AM   #5
diyAudio Member
Join Date: Jan 2007
I'm pretty sure now that there is a big problem with the idea I proposed. The way it could seem like it could work would be because the switching losses dissipate energy. That wouldn't be helpful, then.
  Reply With Quote
Old 27th May 2010, 11:30 AM   #6
Slobban is offline Slobban  Sweden
diyAudio Member
Join Date: May 2009
aah, ok... now I see... it took me some time to realize what the circuit did... you burn up the spike in the fet. it would have been simpler to use an RCD clamp. the amount of energy would be the same. but a resistor have a far better chance of surviving the energy over the mosfet.

thanks for the suggestion though!

there is an intresting connection between the voltages in a flyback that is puzling my mind. if a capacitor is connected over the flyback transformer, through a diode, and is charged to Vin+spike, when the voltage over the switch is Vin*2. when the switch is closed, you would have the same voltage over the transformer. there should be some way of arranging diodes around this cap, to utilize this connection in voltage potentials to discharge the cap back to Vin when the switch closes. I have been thinking of this for some time, but still I cannot find a good way to make this happen.

  Reply With Quote
Old 27th May 2010, 01:23 PM   #7
diyAudio Member
Join Date: Jan 2007
When I noticed that my idea wasn't really helping yesterday, I was thinking about the idea of a diode discharging the cap, too. It seems a separate switch would have to be included to do it. I seem to recall that in theory, an inductor connected between the cap and Vin can discharge the cap, though.

Last edited by Electrone; 27th May 2010 at 01:52 PM.
  Reply With Quote
Old 27th May 2010, 05:58 PM   #8
diyAudio Member
Join Date: Jan 2007
Thanks for expressing your appreciation.

I guess the image I've included below might show something we are trying to think of.

There is another completely different approach, but more complicated, that I have experimented with if this might not be helpful enough. It involves doing the PWM while the transformer is powered by mains AC instead of rectified mains. It eliminates the rectifier bulk capacitor. It involves clamping the transformer after each half cycle. The MOSFETs are configured to operate bidirectionally. I did it using a 60 cycle transformer because I love the ease of slow speed. Maybe it can be done by chopping the AC at a much higher frequency than 60hz as well.
Attached Images
File Type: jpg spike_recovery.jpg (10.2 KB, 217 views)

Last edited by Electrone; 27th May 2010 at 06:17 PM.
  Reply With Quote


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off

Similar Threads
Thread Thread Starter Forum Replies Last Post
isolated or non-isolated psu for UcD in a car? zilog Class D 1 2nd February 2009 02:56 AM
200w to 800w or am i missing something? djmarvellous Solid State 75 18th December 2007 12:29 PM
500W, PFC, isolated, regulated PSU advice theChris Power Supplies 3 2nd February 2007 04:19 PM
fried 800w audiobahn 1needforspeed Car Audio 10 22nd July 2005 06:03 PM
Help , I need a 4CH 800W amplifer rex_wang Chip Amps 10 11th August 2004 01:39 PM

New To Site? Need Help?

All times are GMT. The time now is 03:24 PM.

vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2017 DragonByte Technologies Ltd.
Copyright 1999-2017 diyAudio

Content Relevant URLs by vBSEO 3.3.2