[SOLVED] Looking for off the shelf SMPS ic, EXternal FET < 500 VDC down to 350 VCD

[hughmanoid]

I am looking for something similar to an LM2575 BUCK only with an external witch.
I am starting a new post since I am focused on just finding the right IC.

Input power ramps up from 0 to 450 VDC at up to an availabel 15 amps and back down to 0.
The load is a VICOR DC-DC converter rated to run from 200 to 400 VDC, 300 VDC nominal.

Requirement 1 is to act like a series pass regulator that limits output to about 350 VDC
and will still pass current during long periods of dropout down to 200 VDC.
Output current will pass through an inductor and be filtered by a large cap.

Requirement 2 is to start bucking from 375 VCD to 450 VDC.

No need to boost. Just a simple switcher with an external P or N FET or IGbT.
Toss in a bulky toroid, heavy duty high current diode, large 600 VDC cap, and
a couple resistors in a feedback divider.
I can power the chip with 15 volts or so, just need to switch high voltage at
a few amps or more with multiple FETS.
The closest IC I found is an LTC4366 surge protection IC that clamps surges and turns off
a series pass fet before it overheats.

I am hoping to avoid futzing around with discrete comparators, gate drivers, pwm generators
since I don't want to re-invent the freewheel....

Thanks

 

[mtwieg]

Okay, so to clarify you just want a buck regulator which regulates to 300VDC, but when the input drops below that it's acceptable for the output to droop too and lose regulation.

You won't have to build your own pwm controller, but you are going to have to make your own gate drive circuitry, since I'm not aware of any controller which can switch 450V. Unless it's an external PFET. But you'd probably be much better off with an N channel IGBT, so you'll need a high side gate driver, and a general purpose PWM controller. If you use a bootstrapping gate driver, you'll need to limit the duty cycle to <100%, so you'll get a bit of voltage dropout no matter what, but it doesn't sound like that's a big issue. The only potential issue I see is if you need current limiting, which is tricky to implement with high voltage buck regulators.

 

[hughmanoid]

Thank you. Can you suggest a general purpose PWM controller?
Target output voltage is around 375VDC.
I can provide 2 separate isolated 15VDC rails if necessary for gate and PWM controller drive.

------------- more extraneous info ------------------------------------
I have several 1200 or 600VDC high current (to 45A) Nfets, IGBTs, Pfets, and diodes.
I have several toroids to test with (40 to 180 uH) inductors rated a several amps.
I have a few large 600 VDC 1uF to 40 UF caps to test with but may need more.
I can always add more to reduce ripple but need to test at light loads < .1A first.
I have several simple optoisolators and a couple low side/high side gate drivers.
I am alos looking at SSRs and totempole opto gate drivers as well.
I plan to use back to back zeners to clamp the Vgs to +-12VDC or so.

Yes, droop is OK. All values are approximate. Hope to draw .5 A to 5 Amps.
Primary DC-DC converter runs at reduced power output from 1?? to 199 VDC .
Nominal voltage is 300VDC. Full power conversion from 200 to 400. Transients over 425 and POOF
I can provide 2 isolated 15VDC low current (.2A or less) supply for any logic or gate drive.
I can use a comparator(s) to shutdown below 150VDC and disconnect at 400 VDC.
I have a large selection of zeners (4V to 300V) to stack in series if I need a high voltage reference.
I can crowbar / blow a fuse if FET fails (short) and output exceeds 400VDC with a 600VDC 60A SCR.

I will be building a test setup to run off rectified 110 AC line voltage (165VDC) and then graduate to
rectified 220V (2 phase dryer outlet) 315VDC. I have a few voltage doublers and triplers to test
the 450 VDC at very low current (10mA?). I also have a 110VAC to 800VAC? transformer that may
have been used for a neon sign that rectifies to 1100 VDC @ 30 mA.

Hopefully I won't end up on the TV show ''1000 ways to die''....

 

[mtwieg]

There are tons of possible chips to use. For applications like this where you don't know exactly what you need, I'm a big fan of basic control chips like the UC2842 and UC2843. Can be configured to a huge frequency range, current mode or voltage mode operation, very simple internal operation. The UC3824 is even better, and has a soft start and shutdown pin, plus you can use your own voltage reference for the error amp (the UC3824 is pretty expensive though).

 

[hughmanoid]

Thanks for the suggestion. I ordered a few of the less expensive ICs to build bucky.
From what I understand, A Pfet on the high side is more common but has a higher Rds on.

I understand Fets can be paralleled to load share because their current decreases as they heat up.
Apparently normal transistors do the opposite and burn up as they carry more and more of the load.

Generally, do IGBTs load share like fets or pop like their bipolar siblings ?

 

[mtwieg]

For paralleling FETs or IGBTs, you have to look at the characteristics of the specific device, and its operating point. Many FETs can't be paralleled due to a high negative coefficient of Vth, or a negative Rdson coefficient. Switching losses pretty much always have a positive temperature coefficient, so watch out at high frequency operation. Most FETs can be paralleled fine, but you have to check and make sure, especially at high frequencies.

For IGBTs it's more complicated. Many IGBTs will have static temperature coefficients that are positive in one region and negative in another. Punch through devices have a negative Vce tempco while non punch through devices have a positive Vce tempco (most modern IGBTs are NPT). IGBTs also have positive tempco for switching losses. Unless an IGBT is explicitly specified as being suitable for parallel operation, you should do some calculations yourself.

 

[hughmanoid]

Finding the right inductor and filter capacitor(s) may fall through.

I may try an industrial strength IGBT module (half bridge) as a simple two stage / cascaded
source follower with 380 volts worth of zeners on one gate and 420 volts worth on the higher gate.
Dropping 80V at 5A means buring up 400W (200 per collector).

Not sure how they will need to be biased or operate when runniing in drop out mode (375 volt or lower supply)
but I will test it out on smaller fets at safer voltages.

Just ordered this IGBT module on eBay:
MG100Q2YS1 absolute maximum ratings: (1)collector-emitter voltage:1200V; (2)gate-emitter voltage:±20V; (3)collector current:IC:100A, ICP:200A; (4)forward current:IF:100A, IFM:200A; (5)collector power dissipation:800W; (6)junction temperature:150℃; (7)storage temperature range:-40℃ to +125℃; (8)isolation voltage:2500V; (9)screw torque:3/3N.m.

Also posted a plea for help on TI's forum. Their forums seem to be moderated by TI staff and hopefully they or
other forum members can help. Most of the power regulation products I stumble across are geared for portable,
low power or automotive apps.

Brian E.