(shame) My lack of knowledge of SCR's...

I've got an API console power supply here which has a failed 2N4441 'crowbar' thyristor.

I thought that the ever-wonderful Skycraft would have one, but they've let me down... (SHOCK!!!)

Is there a more common SCR which I might use in its place to effect a repair? -Right now I have it 'snipped' and was able to repair the supply failure which seemed to have caused the SCR to die in the first place, but I don't want to put it back into service without crowbar protection.

I'm afraid I don't know any part numbers to be able to come up with a sub... -any ideas?

-Keith

OnSemi 2N6505 should fit, same pinout, for the critical part dv/dt same 50V/us, allowing more current, TO-220AB case
Farnell #1417099 , current stock 522

Also quite a few 2N4441s on the old ebay too

yeah, it's just a buddy with a vintage API console, and I'm trying to see if there's an equivalent which can get him up & running without having to wait 5 days...

it's the Over-Voltage crowbar. Doesn't ever see over 30V (raw, unregulated, off-load)  so voltage isn't really critical.

I'll have a look and see if Skycraft has 2N6505.

Thanks!

Slim pickings in terms of choice...

The only thing I can find is a "C122D" (click for datasheet).

it's 400V, 8A... same pinout and form factor.

Anyone see a reason why this shouldn't work replacing a 2N4441 (click for datasheet) ?

Thanks,

Keith

If 8A is enough until the fuse blows - why did the SCR break ? Any caps changes causing higher surge currents ?
A more complete datasheet (no mention of crowbar use) "C122".
Some reading on crowbars http://www.onsemi.com/pub_link/Collateral/HB206-D.PDF, starting at pg.97.

SSLtech said:

Slim pickings in terms of choice...

The only thing I can find is a "C122D" (click for datasheet).

it's 400V, 8A... same pinout and form factor.

Anyone see a reason why this shouldn't work replacing a 2N4441 (click for datasheet) ?

Thanks,

Keith

Click to expand...

The C122F would be the ideal substitute @50v

2N6400 looks compatible.......
http://www.onsemi.com/pub_link/Collateral/2N6400-D.PDF
http://uk.farnell.com/on-semiconductor/2n6400g/scr-thyristor-16a-50v-to-220ab/dp/1126367?Ntt=2N6400
I think the 50v part here is the key....a higher voltage part may alter the crowbar sensitivity
as long as the current rating is = or greater than the original there should be no problem

Well I popped it in and it seems to be okay. power supply functions. -The major difference was that the screw hole 'sleeve' is not isolated as it was in the original, so I had to use a shoulder washer and a thinner screw to bolt it onto the heatsink.

Amek power supplies (designs from 30-35 years ago) used to destroy their crowbars... every time they went into OV protect. -Some manufacturers just don't get it right, but I think that in this case, a combination of dried up heatsink grease on the Thyristor (so it didn't cool properly when it activated) and the lack of a DC fuse after the reservoir caps meant that instead of the fuse blowing promptly, the fuse on the PRIMARY side of the transformer (which is going to take a moment to get warmed up) meant that there was excess current flowing long enough for the thyristor to get righteously baked.

Usual plot with O/V failure; Pass transistor goes C-E hard short, then the crowbar triggers. fuse fails to pull quickly, crowbar fails partial-short, hilarity ensues...

Why manufacturers don't put fuses DIRECTLY after the reservoir caps is a mystery to me... Also, this crowbar location is far from ideal... it's AFTER the pass transistors. I'd put them RIGHT after the fuse. Stop everything DEAD... and quite instantaneously.

Thanks for the thoughts. I'll tell him to bag some original parts off of eBay... I don't like the power supply design either mechanically or in terms of the protection circuit implementation, but the console itself is a thing of sonic beauty.

I got the session up and running in time anyhow, so he's all good for now.

Keith

Generally these aren't expected to cycle often, and since they will fail as a short circuit they can be undersized for the reservoir capacity and still be functional.

I would expect a dead short on the output to take out the primary fuse.

Internal fuses are generally added to prevent fire hazard or shock hazard from some other internal component failure. An open internal fuse is as much of a PIA to the customer as any other failure.

I once had a hassle with UL about a sticker for an internal fuse... picky picky picky...

JR

SSLtech said:

Amek power supplies (designs from 30-35 years ago) used to destroy their crowbars... every time they went into OV protect. -Some manufacturers just don't get it right, but I think that in this case, a combination of dried up heatsink grease on the Thyristor (so it didn't cool properly when it activated) and the lack of a DC fuse after the reservoir caps meant that instead of the fuse blowing promptly, the fuse on the PRIMARY side of the transformer (which is going to take a moment to get warmed up) meant that there was excess current flowing long enough for the thyristor to get righteously baked.

Click to expand...

Discharging capacitors directly through thyristors isn't a great idea. When they switch on they start to conduct in a small spot which gradually spreads over the whole area of the die. If di/dt is too high the current increases faster than the conducting area can spread. The result is a high current concentrated in a small area of the die, which melts and causes a short.

This all happens too fast for a heatsink to make a difference - the only thing that prevents it is limiting the rate of current rise (eg. with a small inductor in series with the thyristor).

JohnRoberts said:

I would expect a dead short on the output to take out the primary fuse.

Click to expand...

In a one-rail unit, an appropriately-sized primary fuse should be able to blow... but in an SSL for example, they have MULTIPLE DC rails, all of which up the ante on the demands for the primary fuse, and thus the window between 'too-small-to-not-false-trigger' and 'small enough to be protective' becomes difficult.

SSLtech said:

JohnRoberts said:

I would expect a dead short on the output to take out the primary fuse.

Click to expand...

In a one-rail unit, an appropriately-sized primary fuse should be able to blow... but in an SSL for example, they have MULTIPLE DC rails, all of which up the ante on the demands for the primary fuse, and thus the window between 'too-small-to-not-false-trigger' and 'small enough to be protective' becomes difficult.

Click to expand...

Yup, fuses suck for any kind of precision (and may fatigue over multiple cycles). As i recall the last big dog console power supply I did. the fuses were so marginal, I added status LEDs that would reveal at a glance  A) Green= PS rail up and fuse good, B) red= PS rail up, but fuse blown, and C) off, rail down. Active current limiting can be orders of magnitude more accurate.

Crowbar protection, while literally to protect the circuit node it is placed on, generally ignores the current draw, expecting to blow a fuse (or not). If it doesn't blow the fuse. or start a fire, it has still effectively protected the following circuitry. Agency safety testing surely looks at this (fire/shock hazard). If this is anticipated in system level design, the several crowbars could feed a common upstream crowbar to trip the mains fuse, or just let it all get hot...  

JR

PS: DV/DT issues in thyristors can be mitigated with RC snubbers. If turned on in the proper quadrant a small cap from anode to gate will slow down the switch waveform edge rate.

JohnRoberts said:

DV/DT issues in thyristors can be mitigated with RC snubbers. If turned on in the proper quadrant a small cap from anode to gate will slow down the switch waveform edge rate.

Click to expand...

dv/dt can cause problems when the thyristor is off. If the power supply voltage ramps up too quickly when it's switched on, it can cause the thyristor to trigger spuriously. This is largely due to anode-gate capacitance, so I'd be wary of adding more capacitance between them in a crowbar circuit. A capacitor between gate and cathode however will reduce this tendency.

johnR said:

JohnRoberts said:

DV/DT issues in thyristors can be mitigated with RC snubbers. If turned on in the proper quadrant a small cap from anode to gate will slow down the switch waveform edge rate.

Click to expand...

dv/dt can cause problems when the thyristor is off. If the power supply voltage ramps up too quickly when it's switched on, it can cause the thyristor to trigger spuriously. This is largely due to anode-gate capacitance, so I'd be wary of adding more capacitance between them in a crowbar circuit. A capacitor between gate and cathode however will reduce this tendency.

Click to expand...

Yup my bad, I was thinking in terms of 3 layer devices, a thyristor with cap could trigger inadvertently from a positive voltage spike when off, and once avalanched on, on for good, so no negative feedback.

I have more than my share of bad memories from triacs, but never dv/dt problems. I had some in a power switch application that would spuriously refuse to turn on, even after releasing the smoke from a current limiting resistor in series with the gate. Even the triac manufacturer couldn't find anything wrong with the parts or explain the fault...  I had to replace with mechanical switches since my customers didn't care why they were failing to turn on...

JR

JohnRoberts said:

I have more than my share of bad memories from triacs, but never dv/dt problems. I had some in a power switch application that would spuriously refuse to turn on, even after releasing the smoke from a current limiting resistor in series with the gate. Even the triac manufacturer couldn't find anything wrong with the parts or explain the fault...  I had to replace with mechanical switches since my customers didn't care why they were failing to turn on...

Click to expand...

I've had both di/dt and dv/dt problems. For a while in the mid-90s I was a design engineer at a company that made lighting control gear, and spent many hours trying to work out why the things kept popping or otherwise misbehaving. There seems to be a certain amount of voodoo in the way SCRs and triacs work that even the manufacturers don't fully understand. Nowadays I prefer to avoid them.

johnR said:

There seems to be a certain amount of voodoo in the way SCRs and triacs work that even the manufacturers don't fully understand. Nowadays I prefer to avoid them.

Click to expand...

amen... My gut feeling is these should be mature and well characterized in the physics. The description of how they are supposed to work is pretty straightforward. My personal experience mirrors yours that even the vendors can't explain why, when they misbehave.

Maybe we dealt with the same vendor?  :


JR

JohnRoberts said:

My gut feeling is these should be mature and well characterized in the physics. The description of how they are supposed to work is pretty straightforward. My personal experience mirrors yours that even the vendors can't explain why, when they misbehave.

Maybe we dealt with the same vendor?  :

Click to expand...

I can't remember which vendor it was, unfortunately. I suspect that the problem isn't that nobody knows, but an end user has to go through so many layers of tech support people that it's often impossible to get the required information.

I eventually found out why high di/dt was killing so many thyristors when I read it in another manufacturer's applications manual. (I think it might have been Motorola's Thyristor Device Data book, but I can't check that as it was in a recent cull of dead-tree data books - I wish I'd kept that one).

I've always thought crowbars were a bad idea because they assume the unit has a fuse of the correct rating installed.  If a higher value fuse was installed (for whatever reason) then you have a guaranteed short circuit with inadequate (or no) overcurrent protection.  Following that reasoning, I think it's best to replace the fuse with a 22 caliber rifle shell.  Then, when the crowbar trips, the rifle shell fires, hopefully killing either the unit itself or the person who designed the circuit, before the power supply catches on fire.

Joe