"Carbon" New Drsstc / Projects / Forums

Forums

4hv.org :: Forums :: Projects

« Previous topic | Next topic »

"Carbon" New Drsstc

1 2 next

Move Thread

LAN_403

Reaching

Wed Aug 15 2007, 07:27PM

Registered Member #76 Joined: Thu Feb 09 2006, 10:04AM
Location: Hemer, Germany
Posts: 458

Now, that i have more time, i wanted to build a new, big drsstc.

First attempt was to build a powerful discrete brick driver to power the fuji 1mbi400-06 (600A/800APulsed 600V).

That seemed not to be easy, cause the gate capacitance of one device is 84nF.
I wanted the input signal optocoupled so that i dont need a gdt or other driver transformers in the drsstc.

heres what i came up with by now

Carbon1

Heres the whole highside driver for the fullbridge. the power part such as the optocouplers are fully shielded to prevent any em interference. Powered by 2 print transformers i can drive 4 of the big fuji igbts up to 100khz cw! without big problems

Carbon4

Closeup of the 4 identical push pull drivers, nearly the same as a discrete ucc driver ic, but much more powerful, much more forgiving.

Sigout100khz

Heres the output signal of one stage without igbt connected. div5v/cm 5ÂµS/cm 100khz

Sigout100khzmitlast

heres the gate signal of one of these heavy igbts with 84nf gate capacitance, at 100khz cw!
For sure, the signal is not bad but not nice. i plan to get a resonant frequency of around 50khz or so, so the gate waveforms should look much better at half the frequency.

So give it a go. More to come soon. No Specs for the secondary, mmc yet

Experimentonomen

Wed Aug 15 2007, 07:36PM

Registered Member #941 Joined: Sun Aug 05 2007, 10:09AM
Location: in a swedish junk pile
Posts: 497

Niiiiiiice start :)

Marko

Wed Aug 15 2007, 08:02PM

Registered Member #89 Joined: Thu Feb 09 2006, 02:40PM
Location: Zadar, Croatia
Posts: 3145

Hate to say it now, but I would be very wary of those dual-secondary transformers. Insulation between those two windings is usually very poor, but fact that brick is 600V probably saves the day.

Reaching

Wed Aug 15 2007, 08:04PM

Registered Member #76 Joined: Thu Feb 09 2006, 10:04AM
Location: Hemer, Germany
Posts: 458

They dont look like, but they are 4kV tested in insulation, and cost about 4 times as much as a normal print transformer. they should stand up the few hundred volts

Firnagzen

Thu Aug 16 2007, 07:56AM

Registered Member #567 Joined: Tue Mar 06 2007, 10:55AM
Location: Singapore
Posts: 147

Good grief. You built 2 H-bridges to drive a H-bridge? I really admire the work you've put into this, I don't think I'd have the patience!

Reaching

Thu Aug 16 2007, 11:51AM

Registered Member #76 Joined: Thu Feb 09 2006, 10:04AM
Location: Hemer, Germany
Posts: 458

Okay, i made some decisions about the size and wire gauge of the secondary coil and came up with 2 designs. Now i wanted to know, cause its the first time i build such a huge secondary and i dont want to waste my money and time, what data is significant, especially for a huge drsstc secondary

Heres the first design

Secondary diameter 16cm ~6,2"
Winding lenght 90cm ~ 35,4"
Wire gauge 0,4mm~ 26awg
turns 2249
toroid 15x60cm~26,4pf
resonant frequncy 68,7khz
dc resistance 155ohms
Q 309

Heres the second design

Secondary diameter 20cm~ 7,87"
Winding lenght 100cm~39,37"
Wire gauge 0,5mm~ 24awg
turns 2000
toroid 15x60cm~26,4pf
resonant frequncy 63,8khz
dc resistance 110ohms
Q 428

and not to come to wrong conclusions here. its not 2 H bridges. the driver consists of 4 pus pull arrangements to drive each for itself one brick of a H bridge

For any of you who might want to know exactly how it works, heres the shematic of one driver.

The whole design was simulated by sw cad and build up several times. i even build the whole thing to determine the maximum, minimum frequency range, frequncy range of the optocoupler and so on
yeah, and for those who are familar with optocouplers or high side drivers in general. the delay between input and output is 1,6ÂµS. thats a lot, but i think i should give it a go, cause every input signal is delayed by 1,6ÂµS. The optocoupler itself delays the signal 1,4ÂµS.. maybe it will grow to a problem with overcurrent protection, cause the bridge turns off 1,6ÂµS after the ocd was detected. but dont know.

Steve Ward

Thu Aug 16 2007, 06:01PM

Registered Member #146 Joined: Sun Feb 12 2006, 04:21AM
Location: Austin Tx
Posts: 1055

the delay between input and output is 1,6ÂµS. thats a lot, but i think i should give it a go, cause every input signal is delayed by 1,6ÂµS. The optocoupler itself delays the signal 1,4ÂµS.. maybe it will grow to a problem with overcurrent protection, cause the bridge turns off 1,6ÂµS after the ocd was detected. but dont know.

That is a seriously huge delay. Dont worry about the OCD, it shuts off at a normal switching event. But, you will be hard switching big time, all the time. Transformers dont look so bad compared to those slooooooow opto-isolators (my signal transformers only add a few nS delay). Id suggest you look into newer opto's, you can get some with only 60nS delay times.

Lets take a look at how bad the switching will be with that delay time. 50khz has a half-period of 10uS. At 16% of that period, you will be switching sin(180*.16) = 48% of the peak current! Thats really quite bad since large DRSSTCs need at least 1kA for decent results (so hard switched 500A). This makes dealing with voltage spikes nearly impossible and also lowers your power-factor a good bit (so lower efficiency accordingly).

Fix those optocouplers! Im also not convinced the rest of that gate driver (with the discrete transistor buffer) is all that fast either. I tried hard to get my total propagation delay (from CT feedback input, to gate driver output) down to something like 150nS (less than 1/10th of just your optocoupler). Even switching from 15V to 5V logic (since its faster) made a very noticable difference in switching noise.

I suppose it probably would work to some extent, but i personally would not use it.

Reaching

Thu Aug 16 2007, 06:56PM

Registered Member #76 Joined: Thu Feb 09 2006, 10:04AM
Location: Hemer, Germany
Posts: 458

the Delaytime is not within the bridge itself steve. the complete bridge is delayed 1,6ÂµS. so just every igbt within the bridge has the same delaytime and that means no hard switching for 1,6Âµs. the delay is just between the input signal (Oprocoupler) and the output signal. so i drive the gate driver with say a 100khz signal, there is 1,6Âµs delay between the input 100khz and the output 100khz.

i tested the bridge as it is (loose wiring) at 50khz from a signal generator on cw with an bridge input voltage of 100V. The current draw of the bridge without load was around 100mA which is quite good.

No Delay between each igbt, so for safety i should add a bit deadtime

Sulaiman

Thu Aug 16 2007, 06:57PM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3140

I haven't used 4N26, but I have used 4N25 which is similar.
Have a look at these application notes

and

to see how a major speed improvement can be made with minimal changes.

Also, what's the purpose of R4,R5,R6,R7 compared to removing all four and
connecting both bases of Q2 and Q3 to the collector of Q1?
i.e. a little speed lost here.

Reducing R2 and R3 to 1 kOhm would also make it a lot faster
the resistors would dissipate 400mW and 200mW respectively, not much for the speed gain.

Steve Conner

Thu Aug 16 2007, 07:22PM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

Well, big IGBTs can have turn-off delays of almost 1us themselves, if not more for those old Fuji Electric monsters. So you can increase that 48% to 100% :P This is why I got so interested in PLLs. They can dial in a phase lead to compensate all the delays in your circuit. (it takes a cycle or two for the loop to settle though, so don't worry, no time travel needed...)

The Toshiba TLP250 is a fast optocoupled gate driver that comes to mind.

1 2 next

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob