IGBT and MOS in a linear regulator

On a sunny day (Sat, 09 May 2009 11:53:34 GMT) it happened "maxi" wrote in :

I dunno, I use LT spice. Also, I use a 'generic opamp', and just any MOSFET I could find.... I just checked if perhaps in my circuit the + input of the opamp touches the rails (opamp supply lines), some opamps reverse output in cases like that, I added a diode to prevent the + from going negative, just to see if it makes any difference: ftp://panteltje.com/pub/limiter_600v_with_diode.gif No difference with this opamp. Try running your spice with a different opamp perhaps?

Jan if you want i send you the LTspice file of mine diagram.

Here the picuture of it

"Jan Panteltje" ha scritto nel messaggio news:gu1rus$u1f$ snipped-for-privacy@news.albasani.net...

I've fopund a model for 1200volt 400 amp igbt and the simulation have the same problem of my circuit!!!!!!!!

Here:

"Jan Panteltje" ha scritto nel messaggio news:gu3se1$g9a$ snipped-for-privacy@news.albasani.net...

Better pic with IGBT(you can find the model on LT spice newsgrpuo or if you want i send you)

Picture Here

"Jan Panteltje" ha scritto nel messaggio news:gu3se1$g9a$ snipped-for-privacy@news.albasani.net...

On a sunny day (Sat, 09 May 2009 12:55:10 GMT) it happened "maxi" wrote in :

Seems to work, but I do not know what and why :-)

Does not seem a very reliable circuit, with - input opamp to ground opamp...

On a sunny day (Sat, 09 May 2009 13:37:19 GMT) it happened "maxi" wrote in :

Yes of course, it is not reliable, in fact I do not understand why you do it that way.

On a sunny day (Sat, 09 May 2009 13:32:03 GMT) it happened "maxi" wrote in :

Yes, your circuit makes no sense (at least not o me), I gave the right sort of example, try that with your IGBT model?

Yes your circuit is ok.Mine not

Now i tested live on bread board........and i have a surprise: it clamp but during the to-fff i've a higher frequency oscillation....only with IGBT and not with MOS. But Why? Op amp autooscillation?

Here:

"Jan Panteltje" ha scritto nel messaggio news:gu44a5$raq$ snipped-for-privacy@news.albasani.net...

On a sunny day (Sat, 09 May 2009 15:45:29 GMT) it happened "maxi" wrote in :

Yes, could be the opamp needs some compensation, look here I use a 22pF capacitor between output and - input: ftp://panteltje.com/pub/limiter_compensated.gif It makes the little oscillaton at the statrt of the limiting go away. For your opamp the compensation needed may be a bit different. Could be in its datasheet?

PS, you could perhaps also try to add some hysteresis, so that once the threshold is reached, the thing flips over. Positive feedback from output opamp to + input: ftp://panteltje.com/pub/limiter_hysteresis.gif

Thanks Jan,

also with compensation and Hysteresis i've a osclillation @ 3KHz.

The only can thing that reduce the oscillation peak is encrease the resistor from the out of Op amp and the IGBT till 10Kohm "Jan Panteltje" ha scritto nel messaggio news:gu4ce7$71l$ snipped-for-privacy@news.albasani.net...

On a sunny day (Sat, 09 May 2009 19:32:54 GMT) it happened "maxi" wrote in :

Can I have you IGBT model? I would like to try to see if I can do better. Yes a series resistor in the gate of the IGBT can perhaps help a bit.

What happens if you reduce gain of the opamp a bit with a resistor from output opamp to - input?

Ok i'have two model that do the same thing:

BSM 300GA 1160D Siemens

MG400J1US51

"Jan Panteltje" ha scritto nel messaggio news:gu4sdo$t3i$ snipped-for-privacy@news.albasani.net...

On a sunny day (Sun, 10 May 2009 09:56:59 GMT) it happened "maxi" wrote in :

Can you email one to snipped-for-privacy@yahoo.com (remove NOSPAM)? Google does not turn up those.

What I think happens is that IGBT does not like to be in linear mode. That leads to the question: What does you load look like?

stor

Put a capacitor in series with the resistor. You don't want to lose the gain at low frequencies

Structurally, an IGBT is built like a MOS gated SCR. There is a positive feedback built in them, but this feedback is supposed to be much less than one. You can think about them as being like this:

--------+------------ Drain ! !/e PNP -------+--! ! ! !\\ !!- ! ! Gate---!!- \\! ! !!- NPN !--+ ! e/! ! ! ! \\ ! ! / ! ! \\ ! ! ! ---+-------+----+----- Source

The NPN is an unwanted secondary effect of the MOSFET's structure. The resistor is made as small as practical but is never quite zero.

The PNP provides a lot of current gain to the MOSFET structure's drain current and allows the silicon to be used more efficiently.

This makes a good model for thinking about what goes on in the IGBT. The big fat PNP explains the slower turn off because the charge stored in its base must all get out of the part before it turns off. The fact that the gain is in two stages (MOSFET and PNP) also makes it easier to understand why the phase shift of a IGBT doesn't look like that of a normal RC roll off.

The NPN and PNP with a total gain less than one helps to explain some of the extra longish tail on the current after turn off.

This all also helps to explain problems like trying to make an LDO regulator out of one not working well when the drain voltage is suddenly moved from not quite enough to way more than enough. The internal PNP is saturated

Put a capacitor in series with the resistor. You don't want to lose the gain at low frequencies

Structurally, an IGBT is built like a MOS gated SCR. There is a positive feedback built in them, but this feedback is supposed to be much less than one. You can think about them as being like this:

--------+------------ Drain ! !/e PNP -------+--! ! ! !\\ !!- ! ! Gate---!!- \\! ! !!- NPN !--+ ! e/! ! ! ! \\ ! ! / ! ! \\ ! ! ! ---+-------+----+----- Source

The NPN is an unwanted secondary effect of the MOSFET's structure. The resistor is made as small as practical but is never quite zero.

The PNP provides a lot of current gain to the MOSFET structure's drain current and allows the silicon to be used more efficiently.

This makes a good model for thinking about what goes on in the IGBT. The big fat PNP explains the slower turn off because the charge stored in its base must all get out of the part before it turns off. The fact that the gain is in two stages (MOSFET and PNP) also makes it easier to understand why the phase shift of a IGBT doesn't look like that of a normal RC roll off.

The NPN and PNP with a total gain less than one helps to explain some of the extra longish tail on the current after turn off.

This all also helps to explain problems like trying to make an LDO regulator out of one not working well when the drain voltage is suddenly moved from not quite enough to way more than enough. The internal PNP is saturated