3ph Parallel Thyristor Rectifier Not Sharing Current 2

[AG-PPI]

Good afternoon all,

I have a problem which has really melted my brain this afternoon and I am hoping someone can offer some insight.

My company had recently designed and built a three phase thyristor bridge rectifier for our customer.
One of our customers requests was that each thyristor had redundancy, so for each thyristor position in a normal bridge rectifier, we have two thyristors sharing in parallel.
To keep general assembly size down, we supplied to our customer 3 individual phase limbs. Below is the basic circuit diagram of each phase limb with individual thyristor fuses and snubber circuits.

However, our customer is reporting that the thyristors are not sharing current equally, and in some cases, not at all.
Our customer sent one of the phase limb assemblies back to us for examination, and sure enough, they are not sharing.
One thyristor always takes full current, the other takes nothing.
We have removed the thyristors, heated them up to around 100'C (around 110'C junction temperature), run full PIV tests on them. They are fully functioning. We cannot find any fault with the thyristors.
We have even pulled thyristors from a different manufacturer out of stock and still get the same issue.
We have tried triggering the thyristors simultaneously, and individually one before the other, but to no avail.

Oddly, one thyristor ALWAYS takes full load, we'll call this 'Thy A'. If we trigger the thyristor which does share first (Thy B), it comes on and takes full load current. But the moment we trigger 'Thy A', then 'Thy B' switches off and all the current goes through 'Thy A' again.
I feel like we are simply missing something that should be obvious but it is escaping me completely.

Anyone have any suggestions?

 

[waross]

The thyristors are not matched.
While they are within spec, there is enough difference that one hogs the load.
At zero current flow, one SCR turns off as is expected.
It has been years since I have encountered this and I forget the details, but I have heard anecdotes of hand matching devices to much closer tolerance than allowed by the spec sheet.
I seem to remember a circuit that caused enough current to flow to avoid turn off.


--------------------
Ohm's law
Not just a good idea;
It's the LAW!

 

[jraef]

The inherent problem in paralleling thyristors is that they are constant voltage drop devices, but finding two that are the exact same voltage drop is virtually impossible. So the one with the lower VD, no matter how small of a difference, will hog the current. Then you combine this with the fact that thyristors have a negative coefficient of resistance, as the one carrying more current heats up, it has even LESS resistance, which exacerbates the problem.

But, you can parallel them from a system redundancy standpoint, just don't expect them to SHARE current equally, they never will.


" We are all here on earth to help others; what on earth the others are here for I don't know." -- W. H. Auden

 

[Compositepro]

As I recall from reading electronics literature 40 years ago, you will have the same problem with two diodes in parallel. The solution was to place resistors in series with each diode. As little as 0.1 Ohms may be sufficient. It does waste some energy as heat. Perhaps these days there is a more efficient solution.

 

[AG-PPI]

Thanks everyone for the replies, much appreciated.

I kinda suspected that the forward voltage drop difference would have been the issue here.
I thought perhaps that if they were close enough then they would share but unequally, but it seems not to be the case here.
The thyristors did actually arrive as matched pairs from the supplier with a printout of the test results, but it would seem they weren't matched closely enough.

I would try to add in a small series resistor, but these are big 2500A 'hockey puk' thyristors, so that would be very challenging.

Thanks again though for the answers!

 

[LionelHutz]

Only way I know is using a center tapped reactor, each SCR feeds into one end and the output is the center tap. I think a dual winding transformer configuration could also be used.

 

[Compositepro]

That makes perfect sense. But that description sounds a little more complicated than saying to use inductive reactors in series with each scr rather than resistors. Reactors are commonly used to limit AC current where resistive losses are undesirable. My point is that the center tap is not an essential feature, but rather an artifact of how the reactors are wired together.
Or is it an essential feature? Is there some advantage to having a common iron core between the two?

 

[waross]

A comon iron core may force a balance by transformer action.

--------------------
Ohm's law
Not just a good idea;
It's the LAW!

 

[LionelHutz]

Is there some advantage to having a common iron core between the two?

Yes, the configuration lets transformer action force them to share equally.