Finally I can found and Indigo2 teal, but ...

[kikems]

After few time looking for an SGI for test and use COFF or IRIX 5 old apps, I found an Indigo 2 teal at reasonable price. It comes with R4400 at 200 Mhz, 194 Mb ram and extreme GFX board set.

This sgi works really good, and faster for an old config from 1993. Finally I can use Alias3 and TAV 4.x works perfect now.

After 2 days installing and testing apps I have a terrible magic smoke.

Now a friend is trying to repair the Power Supply, if we have luck no other items was damaged.

 

[weblacky]

Did your Indigo2 come from another country? Or was it purchased from someone within your country?

Since it was working for a few hours, this looks like a power supply failure that happens when you have a old power supply from a country that uses 120v system (like the USA) and you run it on 220v system (like you have in Spain).

The autoranging power supply can no longer handle the high voltage of 220v with such old parts and fails, but worked on a lower voltage system still.

 

[kikems]

Comes from Europe ( Germany ), I'm suppose the problem was the years of hardware, as I say the machine works perfect during 2 or 3 days. A friend with good knowledge of retrocomputers and electronics working to repair and recap the power supply. I hope no other elements was damaged, because a fairule in the mainboard or graphics cards will be terrible.

 

[Elf]

My suspicion would also be capacitor failure, with that much smoke. Either one of the bulk electrolytics or the X/Y noise suppression capacitors right at the input. I think chances are better than not that the machine is okay, but hard to say for certain until the PSU is repaired or a replacement is tried.

Nice machine though!

 

[flexion]

Oh no! I'm sorry to hear that.. I was able to smell the smoke just by looking at that video

 

[Irinikus]

That's it! I'm not powering up my Indigo2's again, or any of my older SGI's for that matter until I've recapped them!

I think I'm just going to use my Tezro and Onyx2 from now on!

I really hope there's no other damage to the machine, as there's allot of smoke coming out of the GFX side of the machine! (It's probably just be the result of the fan being stopped!)

I love the response from your cats though!

A "similar" thing happened to my quicksilver G4, which I bought from the US and then kept for a few years before powering it up.

I plugged it in and as soon as I hit the power button, I heard a loud bang and the magic some filled the room, I immediately knew what had happened! (The input voltage selector was set to 120V and I use 220V locally!!!) I sourced a new power supply for the machine, and it now works perfectly!

So If my G4 could survive that, I'm sure the rest of your system is still ok!!!

 

[Jacques]

Oh man, that is not nice. I hope you find a replacement PSU or get the old one fixed, that is a lovely Indigo2.

Your cats, their expression is just sooo cat like.

 

[llopis]

Resurrecting this thread because I'm looking at Kikem's power supply. I know nothing about SGIs in particular, but I know my way around older computers, so I thought I may be able to help.

When I opened and cleaned the power supply, there was nothing obviously blown. One thing I didn't realize is that Kikems never got as far as to turn the machine on. He just plugged in the mains cable and was ready to turn it on when it started smoking. So I figured it would have to be something in the high-voltage side.

Looking around, I noticed one of the big caps had a bulge and a tiny crack. I think that was our smoke source!

I didn't see anything else, so I decided to just replace that cap and the other two big ones next to it. I figured it was either the capacitor that was bad... or there was something else and the capacitor was just a symptom of the problem. But since I didn't see anything obviously shorted or burned away, I gave it a try.

And...

Dropbox - File Deleted - Simplify your life

www.dropbox.com

Since the power supply turns on when one of the signals isn't connected to ground, it turned on, but the same capacitor quickly and spectacularly blew up.

Given that, does anyone have an idea what might be wrong? For something like this to happen this quickly it seems it has to be a pretty noticeable problem, but I didn't notice any shorted diodes for example. Anything I could check?

Thanks.

 

[llopis]

I just found a shorted diode. It's a 1.5KE200CA, which is a transient voltage suppresor. I wonder if it was destroyed in my recent explosion, or it was the cause of the original smoking and subsequent explosion. Hmm...

 

[weblacky]

I don’t visually remember the Teal PSU guts. But overly high AC filter cap voltage (maybe that’s what you have) could be cause by a shorted PFC circuit in a SMPS.

I heard this somewhere, don’t remember where.

I also don’t remember if they have PFC on the HV side on this PSU, if they do check for a shorted Coil and the the caps in the PFC circuit too.

Basic trick, don’t know what PFC is or if there is one? If you see beefy coils interleaved between caps that are really close to the AC cord connector? That’s PFC. I doubt these had active PFC. but if you see a semiconductor between the coils in the PFC, check that too.

 

[weblacky]

Where was the diode, HV or LV? A shorted diode on the LV side normally should cause the PSU to shutdown on OVCP or blow a fuse on the HV. It shouldn’t usually cause a cap to smoke that you just replaced.

However, a failed secondary diode (if it was on the LV) is very often caused by shorted or failed secondary output caps (same thing in PC SMPS).

In all truth, with a PSU this powerful and old, you really need to replace any and all electrolytic caps as even those squat/small ones do important stuff.

But failed output filtering will first take out the secondary diodes, then possibly damage the transformer.

Infact, given the smoking primary filter.... maybe that’s what happened? If a secondary diode was run too long as shorted, it will saturate the secondary coil in the main transformer (that it’s attached too) with the large reverse voltage spike that a working diode was supposed to use/suppress.

So, you may have an overdrawn the primary filter cap(s) from a shorted secondary winding in the main transfer, due to a shorted diode on the LV side caused by aging secondary output caps on the DC output.

Just conjecture of course if the diode you found was in fact a fast recovery or Schottky diode.

 

[weblacky]

Check the secondary transformer outputs with a simple ohm meter (at first), getting nothing is normal on an output with a shorted diode. But not a OPEN diode, just a short/closed one.

If that doesn’t work, you need an instrument that measures/shows basic induction (like a curve tracer, or a ring tester, or an LCR meter). To retest. Also I’d you find an odd reading coil, remove its immediately connect component (often a diode leg) to disconnect it from the circuit to retest. The diodes do a partial disconnect in the correct polarity, so using that direction with your meter you can get something. A short doesn’t have that disconnect so to check the winding, you’ll need to remove the short, then retest.

Also make sure you drain the large caps before all this.

Another two great tests are, with PSU unplugged, shove the ohm meter into the dc side of the connector (using the proper dc ground pins) and see what you see at each voltage pin group does. If you see very low and 0 ohms, that dc rail is shorted on the output stage.

second trick, if you have a nice DC bench current-regulated PSU, set a ideal dc voltage for each rail (in turn, one at a time) and limit your current to say 300mA to start, and shove that back into the correct DC section (3v for 3.3, 12v for 12v, etc) out (using correct polarity) and see what the draw is. Don’t do this for too long. If nothing is taken, that dc rail isn’t shorted. If it takes most of or all the power, quickly remove it. That rail is shorted. Now you’d know where you start, great trick as it detect failed secondary diodes too.

But the more I think about your description, the more I think you may have a failed, main transformer, due to shorted secondary winding. I hope I’m wrong, as you’re not be finding that part easily, but at least it’s not an IMPACT PSU, you’d have zero chance of finding that transformer.

 

[llopis]

I don’t visually remember the Teal PSU guts. But overly high AC filter cap voltage (maybe that’s what you have) could be cause by a shorted PFC circuit in a SMPS.

I heard this somewhere, don’t remember where.

I also don’t remember if they have PFC on the HV side on this PSU, if they do check for a shorted Coil and the the caps in the PFC circuit too.

Basic trick, don’t know what PFC is or if there is one? If you see beefy coils interleaved between caps that are really close to the AC cord connector? That’s PFC. I doubt these had active PFC. but if you see a semiconductor between the coils in the PFC, check that too.

The diode I found shorted is on the HV side, right next to the filter caps and a big linear voltage regulator.

You can see the location here after I removed it:

I'll check for coils with problems with the coil. When you say shorted coil what do you mean? Coils appear as shorts to DC probes. Did you mean a broken coil? Otherwise how do I check for a shorted coil?

Thanks! You've given me something to look into.

 

[weblacky]

Assuming you have 3 digits or more of resolution on your ohm meter, coils (inductors) do not show up as a dead short, unless that circuit is shorted.

while not gospel, large inductors shouldn't be close to zero ohms more around 1 Ohm+. Too many ohms, like 11 ohms+ would be a problem, too. You should see something above 1 Ohm likely?

short refers to either several shorted windings (due to stabilizer goop or corrosion) or completely shorted.

normal coils don’t look like a short unless they are such Low inductance your meter isn’t sensitive enough to differentiate from 0.

 

[llopis]

Assuming you have 3 digits or more of resolution on your ohm meter, coils (inductors) do not show up as a dead short, unless that circuit is shorted.

while not gospel, large inductors shouldn't be close to zero ohms more around 1 Ohm+. Too many ohms, like 11 ohms+ would be a problem, too. You should see something above 1 Ohm likely?

short refers to either several shorted windings (due to stabilizer goop or corrosion) or completely shorted.

normal coils don’t look like a short unless they are such Low inductance your meter isn’t sensitive enough to differentiate from 0.

Ah, that makes sense! I usually check tiny coils, not power supply-level ones like these.

This one gives me 1.8 Ohms across both terminals, so it's probably fine, right?

 

[weblacky]

I’d imagine. Given that nothing is marked, you’ll want to go by comparison a lot. Test similar-looking ones against one another. Something short will be way off.

Also I can’t quite see in your picture. Can you tell me the specifications of the new capacitors you put in for the filter capacitors? I think the text in the side said only 200v? I am I reading that right? Please confirm.

 

[mamorim01]

Also I can’t quite see in your picture. Can you tell me the specifications of the new capacitors you put in for the filter capacitors? I think the text in the side said only 200v? I am I reading that right? Please confirm.

They do look like 200V parts, this may explain why they blow out so soon if they are located on the high voltage side of the PSU. The original parts being replaced are 400V rated capacitors, if memory serves, a bit overkill but certainly a better fit for the European grid.

I absolutely dread the perspective of having to repair one of these PSUs, there is just no space to peek and poke at will, and information is so very scarce...

Edit: nvm, I peeked inside one of my PSUs and these 200V rated parts are apparently factory supplied so they are most likely okay.

 

[llopis]

I’d imagine. Given that nothing is marked, you’ll want to go by comparison a lot. Test similar-looking ones against one another. Something short will be way off.

Also I can’t quite see in your picture. Can you tell me the specifications of the new capacitors you put in for the filter capacitors? I think the text in the side said only 200v? I am I reading that right? Please confirm.

I replaced the caps with the exact ratings as the ones I pulled out. There were two 1500uF/200V and one 390uF/400V. The one that originally cracked and the replacement blew up was the 400V one. I figured it was safe enough to use the same ratings, or would you recommend higher all around?

 

[weblacky]

I had to look through my image library on a Teal Indigo2 PSU, disassembled, cleaned, reassembled and yes, those are the original values. It’s not wrong.

While I’m not sure what’s going on here, it’s definite that the 400v is going to be your real line filter (for sure).

Ok, can you confirm (with the now damaged 400v cap removed from the board) that the holes it went into don’t register as shorted with a multimeter?

Please perform two measuments:

1. on ohms mode, touch the two, now empty, cap holes (vias) and post the result.

2. set the multimeter in diode mode, measure the exact same place (again) and please post the result.

After those, please use diode mode between the legs of the rectifier (google instructions if you need them) and post the resulting measurements.

If those look good, we’ll move on to other basics.

 

[weblacky]

After that, my initial thoughts will be to measure things just after the two fuses. I think the heat shrinked thing right after is a varisistor, then the two blue caps may be x-class caps, then I think that box is a filtering cap (but unsure, could be a protection relay, cannot quickly ID the marking on mine) then I think there are ntcs (two black discs).

All of what was mentioned needs to be measured.

I was wondering if the varisitor or the x-class caps are damaged, they wouldn’t limit current inrush and voltage ramp-up. That might destroy a cap at startup?

I’m really used the protection devices working so we’re looking for something a little unusual here.