Thyristor Stack design

[sunrays]

I am in the process of retrofitting a 600 Amps Four Quadrant Thyristor Stack. During this process I noticed that I have to increase the spacing between the two halves of the heat sink to accommodate the bus bars which pass at the center of the heat sink.

One idea that had come up during discussion was to introduce aluminum filler disk (of diameter equal to or slightly greater than the thryristor itself) between the thryristor and the heat sink on both sides of the thyristor. This whole arrangement will be held together by a set of insulated bolt and nut arrangement. Of course I would have to apply heat sink paste at all the contact points.

Is this arrangement acceptable from the point of view of both thermal & electrical conductivity?

Thanks in advance!

 

[itsmoked]

Got a diagram or a picture of this napkin?

Keith Cress
kcress -

http://www.flaminsystems.com

 

[Skogsgurra]

Careful with the compound. You will have another thermal impedance (three, actually; Al/Al interface, bulk Al and Al/Al interface again). Too much compound will be a problem, both thermally and perhaps electrically.

Another thing I would worry about is the pretension. Is this stack running close to rated current? If only half, or so, I wouldn't be too worrie. But, if close to 100 %, I would think more than twice.

No other way of doing it? Other heat sinks?

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[LionelHutz]

If you're worried about the thermal transfer then copper disks would be better.

I sure hope you are not using the white heatsink grease on these connections. It is an insulator and will impede current conduction through the device.

 

[jraef]

I would also be concerned for the bolt pressure (I'm not sure if this is what Gunnar meant by "pretension"). Introducing additional devices means more likelihood of deformation under pressure and possible changes in the SCR conductivity as a result.

Can you have a special bus adapter fabricated to transition to the bus you want?

 

[Skogsgurra]

Yes, Jeff. I should perhaps have used the word "prestressed".

At 600 A, a few flex leads seem to be the best solution. As suggested by Jeff.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[sunrays]

Thank you every one for your time.

I guess I have to go for a different design. Will think of flex leads as suggested by Shogsgurra.

 

[LionelHutz]

You should be able to get some insulating sleeving for the buss bar cheap or get the buss bar dipped in the insulation. Your devices are likely 1" thick so if you're running 1/4" bar you have some space to work with.

Heck, tape the ends for the joints and then go to a motor shop and get a few layers of varnish put on the bar.

 

[Marke]

I would say quite loudly, Don't Do It without doing some proper thermal tests and calculations.

The thermal resistance between the junction and the heatsink is a major factor in determining the maximum rating that you can get out of the stack assembly.

By adding another thermal junction, you are increasing the temperature rise of the silicon and that could be enough to have a major impact on it's long term reliability.

You need to ensure that the two surfaces of the disk are very parallel and very flat. This requires that they are accurately machined.
You need to ensure that a high quality heatsink paste is sparingly applied to the surfaces and that the clamping pressure is correct.

The thermal impdeance added to the thermal circuit comprises one extra thermal junction plus the conductive path of the disk which will have a higher thermal impedance due to is cross sectional area. A copper disk would be a better option than aluminium, but corrosion between the copper and the heatsink would be a potential issue. You would need to plate the copper to reduce the copper aluminium corrosion.

I would suggest finding an alternative route for the busbars.

Best regards,

Mark Empson
L M Photonics Ltd