SCR Bay Repair

If you really "have to make them like New", you'd better replace the lot (all microswitches and all SCRs).

The only advantage you get from cleaning electronics is the heat transfer is a little more predictable. (Depending on the ventilation and skills of your technicians, by cleaning and disassembling you may have lost critical evidence of this circuit and components.) Disassembling an SCR bank and not replacing all of the SCR is just making it less reliable not more reliable. Save the old SCR as spares. The micro-switches should be a trivial cost in this process, replace the whole fuse assembly with opto-coupler based fuse indication. No moving parts or long term unused contacts to oxidize.

I would take four points into consideration for your decision:

1. What's the cost of an unscheduled outage if a thyristor fails?
2. What is the cost to disassemble the SCR, replace a failed part, then reassemble the SCR and return to service?
3. What is the expected lifespan (MTBF) for the thyristors and microswitches?
4. How much do the switches and thyristors cost?

If an unscheduled outage causes a serious financial penalty, replace all parts.

If the cost of replacing the parts is significantly more than the total cost of the additional new switches and thyristors, replace all parts. If the parts are expensive relative to the cost of installation, thoroughly test the old parts and keep the ones with pass.

If the parts have reached 90% of their expected life, replace them.

Actually, the recommended minimum PIV (Peak Inverse Voltage) rating of an SCR is 2.5 x the line RMS voltage, some say 2x the PEAK voltage, not the RMS, which works out to roughly 2.8x the RMS voltage. I would not use SCRs for a 600V RMS line supply that have a PIV rating of anything less than 1600V, then 1800V is very common on quality systems. But assuming his rectifier was properly rated before it was taken out of service, and he matches part numbers for replacements, that should not be an issue.

The good point you and SHOCKHISCAN raise however is that SCRs cannot be simply replaced with similar devices, they must be replaced with EXACT devices, or those recommended by the OEM ( if they still exist).

Thanks for clearing that up a bit more accurately than I did.

Installing US built computer equipment in the early 80's here in Germany made me aware of the problems (then!!), I was posting from memory, which can be a little less than accurate!!!

We at the time, replaced all the 600v SCRs with 1000v SCRs and realised a great fix, but with a massive amount of work.....

I will remember your infos for any future events, thanks.

Erring on the side of a too high PIV appears to be the "best" failure to commit.

Hi Andy

the scr we are using is this one

http://www.ebay.co.uk/itm/Westcode-N540CH18-SCR-/181149373854

WOW, that thing is HUGE!!!

Much beefier than any I have ever worked on myself, though of course the principles are the same!!!!

Never forget that on ebay they are many forged components, so although your testing will find a defective SCR, but even a forged one out of China will still pass that test......but may fail in normal usage.

Maybe you need to be thinking about a "full-blown" test(er) that explores the full allowed parameters to allow each device to be properly tested individually before installation.

Its a sad state of affairs that companies are forging almost everything nowadays.....even flight safety airplane parts.....

Thank you very much JRaef for your answer . It has been very helpful

we have tested all the SCR loose on the bench , using a kit which works on 12 volts Dc , we connect leads to anode cathode and the gate , a push button on the test kit is pressed and if the SCR is good a bulb lights up if bad the bulb wont light up .

and Yes we have got a firing transformer and capacitors .

we have tested all the SCR loose on the bench , using a kit which works on 12 volts Dc , we connect leads to anode cathode and the gate , a push button on the test kit is pressed and if the SCR is good a bulb lights up if bad the bulb wont light up.

Never heard of a test like that, but it sounds as though you are only testing the gate itself, not the actual device capability under compression. "Good" and "bad" in SCRs are only clearly different at the extremes. That means this kind of test will only weed out the spectacularly failed SCRs, they may still not work when you put them in, clamp them, and expect them to turn off when you want them to. The anode to cathode resistance test under proper clamp pressure is the only thing I would consider valid.

Good luck!

A word of caution in addition to all of the excellent advice already posted.

Timing signals and firing of individual SCR units is critical to controlling voltage output level, waveform shape, and in preventing "out of phase" firing that will destroy the bridge and/or components.

SCR transient "on" and "off" times as well as "forward" and "reverse" resistance are different from one application design to the next.

It is critical that you replace the components with devices that accurately mimic the design specifications and failure to do so will result in poor performance and/or failure(s).