Should I Replace Just Two Blown Diodes in a Bridge Rectifier or all Four

[Humber888]

I think I know the answer to this but seek confirmation. I am repairing the low voltage linear power supply in an old HP oscilloscope and have found that two of the four rectifier diodes (in a bridge) feeding the -12V regulator circuit are open circuit - probably because two of the metal bodies were just touching. The original diodes are old metal 'top hat' types of 3A, 50V PIV rating. I would normally go for something like the 1N5401 as a modern replacement in such an undemanding situation but find I only have the UF5401 ultra-fast version in my stock. This is slightly overkill but should not cause any problems.

With 1N5401s, I would probably just replace the two blown ones, but suspect it would be better to change all four if using UF5401s. Does the panel agree?

[Ed_Dinning]

Not worth the risk of not doing it. Possibly better if you can get a complete bridge and mount it on the chassis for better cooling

Ed

[paulsherwin]

Diodes are cheap. If two have failed, then the other two may be on their last legs too.

[stevehertz]

From the learned answers so far, it would seem that replacing all four is the thing to do. FWIW - wrong as it seems! - I'd only replace the ones that are faulty.

[Sideband]

Originally Posted by stevehertz

FWIW - wrong as it seems! - I'd only replace the ones that are faulty.

No....false economy! I used to think the same BUT a couple of times I did just that and either one or the other diodes would fail soon after. 40 years experience in the trade has proved that changing all four is the most reliable way and avoiding call-backs. As has been pointed out, diodes are cheap....only pence usually so not worth scrimping. Consider also that when a diode (or two) in a bridge go short, there is a tremendous current passed through the smoothing cap as the poor thing passes AC for one half-cycle. Before any fuse blows the two remaining diodes pass a current well beyond their design, tens of amps probably. That is not going to do them any good at all. Replace all four with the same type or better.

[US7IGN/P]

New diodes will have different characteristics and the bridge will become "crooked". Change them all.

[fetteler]

Yep, change 'em all
During the failure the diode pair in the 'good' leg will have been stressed.

Steve.

[Radio Wrangler]

The ultrafast diodes are NOT overkill. They avoid the problem of elderly diode types whic don't turn off cleanly and make RFI. Once you come across this, you'll be surprised at the amount some simple power rectifiers make, even if they're not in anything that receives radio signals, you don't want them in the same house/street.

Just treat it as good housekeeping.

David (woken by a rainstorm)

[PaulDarzi]

Originally Posted by Radio Wrangler

The ultrafast diodes are NOT overkill. They avoid the problem of elderly diode types whic don't turn off cleanly and make RFI. Once you come across this, you'll be surprised at the amount some simple power rectifiers make, even if they're not in anything that receives radio signals, you don't want them in the same house/street.

Just treat it as good housekeeping.

David (woken by a rainstorm)

This makes sense, and looking at the poswer supply of (in my case R606-MB) Roberts radios, they place capacitors in parallel with their diodes in the bridge - I did wonder if it was to reduce RFI, but haven't seen it done in modern sets maybe due to using faster diodes?

As for replacing two or all four diodes, do we assume that a mains spike or short-time transient took out the pair that failed?
(Thinking one half-cycle of mains was over the top, but maybe not the next, so only two diodes were blown by the transient)
Just trying to gain an understanding of the "why" for this failure.

[Station X]

Failure of one or more diodes in a bridge rectifier is quite common. I too would like to know the failure mechanism.

[Humber888]

Well that's 6 to 1 in favour of changing all the diodes, so pretty conclusive. That was my plan anyway, primarily (as stated above) because of possible over-stressing of the remaining diodes and the mismatched characteristics between old and new. I need to dismantle the PSU block further to be able to unsolder the old diodes. I will report back when the unit is repaired.

Many thanks to all.

Mike

[Humber888]

In part answer to #9 and #10 of the thread, I did say in the original post that it looked as if two of the diodes were just touching (all metal bodies). It is unlikely that the downstream regulator board caused the fault as the fast blow fuses on its inputs are all intact. So that just leaves old age, the big reservoir capacitor or a wiring short. I will check the latter as I go.

[mark_in_manc]

Originally Posted by Radio Wrangler

David (woken by a rainstorm)

Ohhh, you have my sympathy. My half-fixed was-covered-in-kitchen-fat Hitachi radio-cassette street find (subject of perhaps the least memorable thread ever to appear on UKVRR) is currently doing sterling service providing FM static noise for just such nocturnal reasons (tinnitus in my case). In fact it's a nuisance that it pulls in so many stations; clean static is hard to find!

OK, back to diodes. I'm a tight-wad, so reading this thread is a good corrective for me.

[HamishBoxer]

I have always changed all if one failed in a bridge, no come backs that way.

[paulsherwin]

I've just checked eBay and you can buy 10 UF5408s for less than £3 including postage. 1N5408s are £1.50 for 20. It's what the Americans call a no-brainer really.

[Sinewave]

I'd suggest replacing smoothing capacitors too, as they may either be aged/incorrect value and have stressed the rectifier diodes.

[stevehertz]

Originally Posted by Humber888

Well that's 6 to 1 in favour of changing all the diodes, so pretty conclusive. That was my plan anyway, primarily (as stated above) because of possible over-stressing of the remaining diodes and the mismatched characteristics between old and new. I need to dismantle the PSU block further to be able to unsolder the old diodes. I will report back when the unit is repaired.

Many thanks to all.

Mike

Not really, if you read it, my "only change two" response was a rhetorical one because previous to the thread I knew no better. Given the much better knowledge levels than me on the forum, I would always change all four now. So no, I would not advise changing just two.

[Humber888]

All 4 diodes changed, ancillary items checked and LV PSU refitted to scope. Everything is now working and all dc supplies are within spec. I do not replace the big Sprague smoothing electrolytics on sight as previous experience suggests they are quite reliable unless the pressure release pip has aged and leaked some of the contents. What caused the diodes to fail? As nothing downstream was faulty, I guess it's down to either age or two diode metal bodies touching - although why this should occur now after years of being like that is not obvious.

Again thanks to all who contributed to my further education.

Mike

[Hybrid tellies]

If any of the bridge rectifier diodes fail, open circuit, short circuit or leaky we always changed out all four diodes.

[Chris R]

On the same theme - I have just restored a Dansette Musique record player and in the process I had to change the original (copper oxide?) bridge rectifier. I used a new KBPC 802, which I had in my spares. I believe this is a silicon device.
A number of bewhiskered transistors and all the electrolytics on the amplifier board were changed. The resulting sound quality is very good - except that that there is a 100Hz buzz when the volume is at higher settings. This disappears when the volume is reduced, so it appears to be picked up in the early stages. I changed the quite substantial (4700 microfarad) smoothing capacitor, to no avail.
I wonder if is the effect described by Radio Wrangler and if so, is there a more suitable bridge rectifier?