How Do SCRs, Thyristors, and GTOs Work for Bidirectional Switching?

[essenmein]

TL;DR Summary

Help with basic understanding of these things...

I have very limited knowledge on these things, I can talk mosfet/igbt/SiC all day... lol

Basically I need to make a bidirectional switch, 600-1200V rating, 2-600A capacity.

No PWM or anything, just on and off, once on (or off), stay that way for at least seconds until next command.

I was thinking anti parallel SCR/Thyristor, but curious about drive, does each device need opposing gate/base polarity to allow it to conduct (since back to back)? Can you do this with a GTO? I assume gate bias must be maintained as the current being switched is AC and will go through zero crossings.

Any thing else to be aware of?

Edit/additional:
If take this as an example:
https://ixapps.ixys.com/DataSheet/CLA80MT1200NHB.pdf

This is an ~80Arms device with 1200V blocking.

The Vf seems to be negative temp co at low current, and positive temp co at higher current:

What is the story paralleling such devices? Seems they should sort of share? As one takes more current its Vf becomes positive temp co?

Or should I aim for single devices that can take the full current?

 

[Joshy]

I can't school you. I wouldn't suggest I know much at all, but figured it's worth mentioning that there's a small section in a very popular power electronics book by Erickson in the switch realization chapter (chapter 4) they talk about Thyristors/SCRs and a few of the challenges with it. I personally don't know much myself, but since it's a seemingly popular book... I remember reading that chapter and grabbed my book... there it was :) so hopefully a copy might be within reach?

Something they seem to focus on is that it has a very large gate and this causes several challenges such as current focusing (has to ramp up slowly during turn on); also turning it off by regular gate control is apparently hard and reverse biasing doesn't resolve that problem very well (reverse current also instead of to the gate and causes a voltage drop). GTO seems to be less problematic in these aspects but has lower ratings. I think another similar device they mentions is a MOS-controlled thyristor (MCT).

Sorry if I'm not too helpful.

 

[essenmein]

Funny you should mention MCT, i did my final year project with those, not sure if things changed since then but they were normally "ON" devices, which lead to some spectacular failures.

I did a bit more reading on the SCR/Thyristor devices, GTO seems to be more higher voltage.

I just played with some devices in simulation and once I add minimum holding current I get nasty voltage spikes at the zero crossing (inductive load) due to that minimum not being zero, so that's not ideal, I'm assuming snubbers are used to deal with that.

Then I found this guy:
https://www.littelfuse.com/~/media/...ristor_module_ms0690j_dl1te_datasheet.pdf.pdf

This would imply each gate needs to be isolated and floating:

But this would suggest that thyrsitors in general are not going to be able to do the job, 500V/us indeed, lol.

Oh well, I will spin around back to plan A, back to back SiC Fet...

 

[Windadct]

Yes - you typically picket-fence the gate current relative to the positive cycle. Injecting current when reverse biased will increase leakage and thus losses. So the firing circuit usually has a comparator type "and" to ensure this.

Here is a more indepth Ap Note : https://www.semikron.com/dl/service...iodes-and-thyristors-en-2018-11-19-rev-02.pdf

Rectifiers are typically not parallelized due to Negative Temp Coefficient - it is difficult to ensure the devices share current well. As one get hotter - it's Vf drops and it takes more current... etc. So a larger device is the way to go. There are dual modules in the this range that can be connected as "AC" switch. If it is a one off or limited production I can help. Contact me directly.

These are "rectifier" items and do not typically need high dV/dT. There are INVERTER GRADE THyristors, that cost more and higher losses ( more H/S Cooling)( Since MOSFET / IGBT are for PWM they typically need that high dV/dt. For an AC switch why do you need this ? Also - the higher Vf ( or Rds on* I) of MOSFET and IGBT cause VERY high losses in the surge currents often needed in AC switching.

I have customers looking for 200-400A IGBT/MOSFET breaker type applications every year. Looks good on paper, until it doesn't. I have to chuckle inside when it comes up . This is major global research guys and they spend 6 months to figure out what I say up front. Granted a solid state breaker is more involved than basic AC switch. Basically if you need 2x or 3x OL type currents MOSFET /IGBT become impractical.

( I am assuming you do not want to turn off sub-cycle?).

 

[essenmein]

The dv/dt capability I was concerned about is the (I assume) off state critical dv/dt, but after reading more carefully the gate is open in this condition, they can probably handle larger dv/dt if gate is shorted.

I can't go into to too much detail but it would be seeing inverter output voltage in the off state, and since this is a SiC based system, very very high dv/dt, like in the 6-10 kv/us.

And ideally I want to be able to turn on or off anywhere, but I was willing to wait for a zero crossing if it meant significant lower cost.

 

[Windadct]

This is like a UPS bypass - they use standard Thyristors for this all the time, so one side "sees" the PWM signal in the off state.

 

[essenmein]

Hmm interesting, so how do they handle that, low impedance on the gate holding it off?

I think I'm going to keep investigating thyrsitors, when you do the math on the number of 25mOhm SiC die you need at max heatsink temp for bi directional 250Arms switch its eye wateringly expensive!

 

[essenmein]

Here is a more indepth Ap Note : https://www.semikron.com/dl/service...iodes-and-thyristors-en-2018-11-19-rev-02.pdf

Perfect, thanks for this link! Some good reading in there.

I don't know why, but stuff like this makes me smile! :D

 

[Windadct]

Ah - you mean the High dV/Dt on the Cath as the SiC/IGBT device turns on? In the design you are thinking of, does the thyristor see the full pulse across it? And are there system parasitics that filter this - and then - yes the Trigger circuit also behaves like a snubber.