5
\$\begingroup\$

I'm looking at a number of thyristors (for example, this N2593MK160). Each has a critical rate of rise of on-state current. This limitation surprised me; I don't understand why the derivative of current would matter. Isn't it the excess heat that damages parts? And isn't that strictly determined by the current, not the rate of change of current?

But that's really a side question. My real question is, how important is it that I stay in that limit? Is the 300 A/μs an instantaneous limit, or a limit on the average over any microsecond? In my application I'm expecting 12000 A/μs at it's peak (though I don't know how long that will last, whether less or more than a microsecond. I suppose if it's longer than a microsecond, my question is answered, since either way I'm way out of spec).

The highest on-state current thyristors on Mouser all have critical rate of rise of currents orders of magnitude lower than what I'm expecting in my application.

EDIT: I didn't expect this to have such relevance, but I see why it does now. I'm trying to build a linear motor. I want it to be "low voltage" (200v), because it freaks my wife out to have an 800v .1 farad capacitor. (to be honest it freaks me out too). Low voltage means I need low resistance to get the power I need. Low resistance means I need thick guage wires in the inductor surrounding the rails (I'm thinking a quarter inch). I'm hoping this thing isn't massive both in volume and weight, so I don't want more than at tops 10 turns. the 12500 A/μs was based on 4 turns around the rails. at 15 turns it's 3300 A/μs, so I don't think it's likely I can solve this problem by adding turns to the inductor.

FINAL EDIT:

Disregard the estimates of dI/dt above. I have found my ODE solver has a massive error in it (or rather, there was a massive error in the ODE I gave the sovler). The comments were correct, my inductor will easily limit dI/dt to within spec. But I learned a lot through this mistake, so thanks everyone!

\$\endgroup\$
11
  • 1
    \$\begingroup\$ ”my application I'm expecting 12000 A/μs” This sounds like the job for some other semiconductor than a thyristor, probably IGBT or MOSFET. \$\endgroup\$
    – winny
    Feb 25 at 12:25
  • 2
    \$\begingroup\$ How important is it that the thyristor work more than once in your particular application? \$\endgroup\$ Feb 25 at 14:23
  • 1
    \$\begingroup\$ Could you show your circuit, or the equivalent network surrounding the thyristor? \$\endgroup\$ Feb 25 at 14:30
  • \$\begingroup\$ I'm trying to make a linear motor. So far my design is 3 parts, though I know I'll add more later: a capacitor, the thyristor, and an inductor. (The rails can be thought of as part of the inductor). I need incredibly low resistance to get the current I need at 200V, so the windings on the inductor must be wide wire. This limits the number of turns I can do to below 10 ish tops and preferably 3 or 4 \$\endgroup\$ Feb 25 at 14:47
  • 2
    \$\begingroup\$ @JosephSummerhays Comments can't be edited. If you make a mistake or omission in a comment, and it does not have any upvotes to lose, you can just copy it, delete it, and make another comment with the desired edits. Otherwise, it's probably better to just add an addendum comment. But think twice before adding important information (such as the specifics you disclose) in comments at all. It's much better to add them into the original question near the bottom and tag or add a pointer in a comment. \$\endgroup\$ Feb 25 at 15:02

2 Answers 2

6
\$\begingroup\$

Turning to ON state doesn't happen uniformly inside the thyristor. It starts somewhere and spreads gradually to the whole area of the semiconductor. Current growth rate limit is specified to prevent too high local current density which could create a hot spot and kill the device.

How important it's to stay in limits? If you have too high rate amperes/second in your circuit a part of the semiconductor in the thyristor can melt. Does it melt if the rate limit is exceeded say 50% for a nanosecod? You can find it by making large scale tests or the same physics simulation analyses that semiconductor manufacturers must do to design their products.

Too high current growth rate is generally prevented by having an inductor in series. In many cases the circuit can have enough inductance without adding it as a separate part (motor, transformer, wiring). But that is not generally true in high power circuits which are constructed to create pulses by discharging capacitors.

The series inductor or a part of it can have saturating core to prevent harms caused by too high magnetic energy which must be dissipated or redirected to somewhere when one wants to turn the thyristor off. In addition the reduction of the inductance (caused by the saturation) may be preferable because after a while the voltage in a capacitor (under discharging) can be low enough so that less inductance is enough to keep the current growth rate acceptable.

I must admit that designing such circuit and verifying it before building the prototype is tricky; far beyond the ordinary linear circuit analysis.

About your case:

Redesign. We cannot help without knowing anything of the goals, boundary conditions nor the existing design of your application. I guess it's not the finest idea to connect in parallel tens of smaller thyristors, each in series with a saturating ferrite core choke. We also do not have a slightest idea is it possible to apply some switch which does not have the max. amperes/second -limit, like spark gap, thyratron or saturating transformer.

\$\endgroup\$
8
  • \$\begingroup\$ This is such a fantastic answer. Thank you so much. If I have a follow up question, can I ask it here in the comments, or should I post elsewhere? I have an inductor in the circuit already. It's purpose is to generate the magnetic field to increase the efficiency of the Lorentz forces. When you say to use a ferrous core to soak up the "magnetic energy", do you mean the magnetic field strength is reduced, or the energy transfer due the changing magnetic flux is reduced? \$\endgroup\$ Feb 25 at 15:05
  • \$\begingroup\$ Saturating core, when it's still not saturated, can have high permeability, so few turns of wire can create enough inductance to keep di/dt low enough in the circuit. When the number of the ampere turns reaches the saturation limit the core cannot store more magnetic energy. As a current growth rate limiter the saturated core inductor is no more than an air core inductor. Hopefully it + the rest of the circuit prevent too high di/dt. When a capacitor is discharged to a load the cap voltage drops and less inductance can well do the job. \$\endgroup\$
    – arnold8a
    Feb 25 at 16:43
  • \$\begingroup\$ @JosephSummerhays you already have an inductor? How much in henries? And how high voltage you are going to have over it? Why you do not use high enough inductace to keep di/dt in the limits with it ? ( by recalling that di/dt = V/L) \$\endgroup\$
    – arnold8a
    Feb 25 at 16:48
  • \$\begingroup\$ I have no idea how many henries, I was going to build it myself. it will be about .5 meters long, .01 cm wide in the inside winding and have 4-10 turns, be made of 1/4 inch copper wire. So... how many henries is that? I got my number for dI/dt from a script I wrote of differential equations. It was based on the backvoltage (equal to derivative of flux), source voltage, and resistance in the system. too long to go into the specifics in the comment section. \$\endgroup\$ Feb 25 at 17:38
  • \$\begingroup\$ Inductance = Flux divided by current. Use only SI units and get the inductance in henries. A drawing would be interesting to see. 0.01 cm doesn't give much space where the flux can have substantial density (assuming I guess the geometry right). Have you something which can move in that 0.01 cm wide space? \$\endgroup\$
    – arnold8a
    Feb 25 at 19:42
0
\$\begingroup\$

Suggest overcoming the issue with a little testing. Your 12,000 A/us is a wee bit high for a thyristor. I'm looking for such switches also. But your voltage is usefully low.

Suggest looking into breaking your cap bank into smaller chunks and using a bunch of switches in parallel to share out the di/dt.

Even pushing things https://ieeexplore.ieee.org/document/896169 getting more than a few kA/us might be a challenge.

If you're building the type of linear motor I think you are, use the armature coming into contact with the rails as the switch. Starting a railgun with the sabot stationary and using external switching is asking for serious rail damage.

New contributor
LightningPhil is a new contributor to this site. Take care in asking for clarification, commenting, and answering. Check out our Code of Conduct.
\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.