Large DRSSTC [update] / Projects / Forums

Forums

4hv.org :: Forums :: Projects

« Previous topic | Next topic »

Large DRSSTC [update]

first previous 15 16 17 18 next

Move Thread

LAN_403

Linas

Sat Apr 06 2013, 04:57PM

Registered Member #1143 Joined: Sun Nov 25 2007, 04:55PM
Location: Vilnius, Lithuania
Posts: 721

just small update.
I connected all electronics, and finaly made synchronization board for all protection, (saturation, overrcurrent, and Ton(max) limiting)
problem is that saturation detections works too well, and just cut IGBT driving signal at 2-5% max power. So i will find out what is the maximum safe rating for CM600 , and tweak voltage divider for saturation detection, and what should work with no problem.
Just 5A 220V and i already don't have enough space

this use Predictor, just i use 0.5R resistor, and 22uH inductor, and timing resistors set to 0 .

for saturation detection fault . MC33153 trigger fault when on voltage is 6.5V, but i have +18Vge. so i should not get any desat detection error, strange. (for Desat detection i used UF4007 diodes, 1K resistor in series, and small capacitor (100pf range to get that spike from turn on/off)
does any of this make sense?
1365276500 1143 FT66676 Vi

Kizmo

Sun Apr 07 2013, 10:58AM

Registered Member #599 Joined: Thu Mar 22 2007, 07:40PM
Location: Northern Finland, Rovaniemi
Posts: 624

Its the module internal inductance that makes the voltage over C-E terminals be far greater than the actual voltage drop over the igbt dies.

This was discussed some years ago and if i remember correctly these big modules can have (depending on current of course) 10s or even 100s of volts over C-E terminals due to their inductance. This is mainly a problem in our drsstc applications where we pass "too high current" at "too high frequency" through these poor transistors :)

Steve Conner

Sun Apr 07 2013, 12:12PM

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

Yes, I saw something like 50-100V across the package inductance of CM600s in my OLTC experiments.

An elegant solution would be to cancel it out with an equal and opposite voltage from a Rogowski coil wrapped around the collector lead. The ghetto solution is to forget about desat protection, which is what I did.

Linas

Sun Apr 07 2013, 03:47PM

Registered Member #1143 Joined: Sun Nov 25 2007, 04:55PM
Location: Vilnius, Lithuania
Posts: 721

Steve Conner wrote ...

Yes, I saw something like 50-100V across the package inductance of CM600s in my OLTC experiments.

I forgot i have inductance ( well, i just assumed it's too low to get any efect for desat detection)
lets makes some calculations:
if i have 6.5V and 45KHz, and lets say transistor Vce is 3V (lets say 1000A)
so 3,5V voltage generation via L.
3.5=I*R
R=2*pi*45k*L
L=38nH
hm, i don't know what to think...

, i just can't say i had 1000A, basically it will trigger when interrupter allow two periods to primary coil, so at best i had 100 or 200A, so that means around 2V and that work out to 80nH.
so i have to assume that i have around 100nH and add voltage divider to make compensation ?

In that case it works out for 4000Apeak i have 113Vl+4Vce and divider should be 18x

Steve Conner

Mon Apr 08 2013, 10:44AM

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

If you set your desat trip to 113V, then it won't be sensitive enough to detect the desaturation.

The voltage across the package inductance is proportional to di/dt, so it's 90 degrees out of phase with the current, hence it falls to zero as the current reaches its peak. That means your IGBT has to desaturate by the full 113V before it'll be detected.

Linas

Mon Apr 08 2013, 11:45AM

Registered Member #1143 Joined: Sun Nov 25 2007, 04:55PM
Location: Vilnius, Lithuania
Posts: 721

Steve Conner wrote ...

If you set your desalt trip to 113V, then it won't be sensitive enough to detect the desaturation.

So i will have simple Current protection. So that is not that bad.
If will be fancy way of measuring Ipeak

Paul Benham

Sun Mar 23 2014, 12:27PM

Registered Member #570 Joined: Wed Mar 07 2007, 03:41PM
Location: Winchester Uk
Posts: 42

I've got some large IGBT modules and they have separate collector and emitter connections for de-saturation detection, I guess they connect back directly to the die.

Maybe its only possible to implement the de-saturation detection with these kind of modules?

Steve Conner

Wed Mar 26 2014, 10:26PM

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

Yes, Kelvin connections make it much easier. Also bear in mind that we are using the IGBT modules at much higher frequencies than they were designed for. In a large inverter drive they might only switch at 5kHz. That gives plenty of time for the L*di/dt voltage to die away, so you can get a more accurate reading of the saturation voltage.

Linas

Thu Jul 17 2014, 11:05AM

Registered Member #1143 Joined: Sun Nov 25 2007, 04:55PM
Location: Vilnius, Lithuania
Posts: 721

Does any one developed better controller for CM600 bridge than Predictor ( which does not for for me somehow ) ?
if yes, can some one link schematics ?

Also in related topic, if club want to use my DRSSTC for show start and end ( so like 5 min run at show start, and 5min run at show end) , how much should i take for this job ? ( since i have to fix my coil, transport 300hÅ¾km away, use one day to set it up, and when after the show go back home again, 300km)

Mads Barnkob

Thu Jul 17 2014, 11:39AM

Registered Member #1403 Joined: Tue Mar 18 2008, 06:05PM
Location: Denmark, Odense C
Posts: 1968

Linas wrote ...

Does any one developed better controller for CM600 bridge than Predictor ( which does not for for me somehow ) ?
if yes, can some one link schematics ?

Also in related topic, if club want to use my DRSSTC for show start and end ( so like 5 min run at show start, and 5min run at show end) , how much should i take for this job ? ( since i have to fix my coil, transport 300hÅ¾km away, use one day to set it up, and when after the show go back home again, 300km)

Are you sure the driver is the problem? It sounds like it could be a similar problem that Kizmo and I had, residual MMC charge that messed up the drivers edge detection.

Topics about it:

and

first previous 15 16 17 18 next

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