1
\$\begingroup\$

In a separate but related question, I asked how EEs normally wire a large number of SCRs in parallel for high current applications. For this 1 example (sample size = 1), I found a single hockey puck to be more costly than an equivalent array of smaller devices.

That question made me wonder, "Is the work of ganging them in parallel of such difficulty that that justifies the price differential between lots of parallel SCRs and an equivalent hockey puck?"

For the purposes of this question, "equivalent" means equivalent off voltage and equivalent on current. For instance, a 10 x 10 array of 100 V, 100 A SCRs can handle the same current and voltage as a 1 kV, 1 kA hockey puck.

The specific scenario that motivated this question is an application powered by 3 kV DC which will conduct 10 kA into an inductive load. On time is 1 ms followed by > 10 ms off time.

My question is twofold:

  1. Is it typical for a single hockey puck to be more expensive than an equivalent array of smaller devices? (I'm asking because, while I might be able to find 10 or 20 examples myself, that is small compared to decades of experience that career power electronics people have.)
  2. If so, is the reason most likely due to lower demand, or is the work of ganging small devices in parallel so costly that you normally save money by going with the hockey puck? Is one way normally cheaper than the other, or is it one of those things that swings with market conditions? For example, "This week/month/year, hockey pucks are cheaper?" or vice versa.
\$\endgroup\$
12
  • 2
    \$\begingroup\$ ((Uninformed) gut feeling: Paralleling is at least one order of magnitude simpler than increasing voltage handling.) \$\endgroup\$
    – greybeard
    Commented Feb 28, 2023 at 5:52
  • 1
    \$\begingroup\$ With negative tempco’s and negative resistance with 10kA loads might fail sequentially like fire crackers on a string very quickly without compensation or protection and thermal heat sinks for the large power loss of 10kA * 1V (?). Perhaps IGBT bricks might be considered. Mercury valves were once used for this that were 1m tall. \$\endgroup\$
    – Hoagie
    Commented Feb 28, 2023 at 6:22
  • 3
    \$\begingroup\$ "10 x 10 array of 100 V, 100 A SCRs" should only handle 1 kV 1 kA, no? \$\endgroup\$
    – Matt S
    Commented Feb 28, 2023 at 8:31
  • 1
    \$\begingroup\$ @Hoagie 1m isn't that big for big mercury valves. There's no scale on these valves at Haywards, NZ, but given the comparable capacity to the later thyristor pole 2 valves... they big. The 2005 condition assessment report also makes for interesting reading. 0.8 Henry 1200A inductors, anyone? \$\endgroup\$ Commented Feb 28, 2023 at 9:28
  • 1
    \$\begingroup\$ @SomeoneSomewhereSupportsMonica I love those Kiwi’s way of thinking. The Land of Jurassic Park. 0.8 H/2 * 1200A^2 is just a Panzer sized tank. \$\endgroup\$
    – Hoagie
    Commented Feb 28, 2023 at 14:03

2 Answers 2

2
\$\begingroup\$

Thyristors don't really like to be paralleled because it's like paralleling diodes each with a slightly different voltage drop. You need extra junk to make sure they current share properly or else things go awry. The one with the lowest voltage drop will cause all the current to go through it which will cause it to melt or explode.

\$\endgroup\$
2
\$\begingroup\$

For instance, a 10 x 10 array of 100 V, 100 A SCRs can handle the same current and voltage as a 10 kV, 10 kA hockey puck.

Herein lies your problem. A 10×10 array of devices has 10 columns of 10 rows.

The ten columns will create 10 paths for current, giving you at most an increase in current by a factor of 10.

With ten devices per column, they can withstand up to ten times as much voltage.

Your design can handle up to 1 kV, 1 kA. But you need 10 kV 10 kA. You need a 100×100 array.

I suspect the added expense of the extra 9900 SCRs will make the integrated "hockey puck" devices more cost efficient.

\$\endgroup\$
4
  • \$\begingroup\$ Thank you for pointing out the typo. It has been corrected. \$\endgroup\$ Commented Mar 14, 2023 at 15:33
  • 1
    \$\begingroup\$ @JamesStrieter "The specific scenario that motivated this question is an application powered by 3 kV DC which will conduct 10 kA into an inductive load." If your design no longer needs to use 10 kA or well over 1 kV then I think it's a fundamentally different question now and my answer no longer makes sense. \$\endgroup\$
    – Matt S
    Commented Mar 14, 2023 at 16:25
  • \$\begingroup\$ The 10x10 array was just illustrating what I mean when I use the word "ganging". That's it. Below that, the question states that I want 3 kV, 10 kA total, which suggests 3 in series, 100 in parallel if all devices are 100 V, 100 A. \$\endgroup\$ Commented Mar 15, 2023 at 11:25
  • \$\begingroup\$ 30 in series, actually. Nonetheless, is your 3kV 10kA SCR really over 3000 times more expensive? Digikey tells me I can buy a 3kV 10kA SCR for just 10-100x the cost of a 100V 100A one. \$\endgroup\$
    – Matt S
    Commented Mar 15, 2023 at 12:37

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