Parallel Diodes

[Wendymott]

Hi Peeps. This Sunday I had a faulty LED Backlit TV to repair. bear with me mod. The fault turned out to be the rectifier that supplies the LED psu. Now here's the problem. There were two diodes in parallel, in this case SR5200 x 2.
The SR5200 is a 5A 200V device. One was short circuit the other OK. Now in my day...... any components in parallel that were load sharing, would have equalizing resistors to take up any differences between the devices. This has none, obviously for commercial reasons, has anyone any comments on this please.

[Refugee]

That is a common problem. They should not be wired like that but to save cost the manufacturers tend to skimp on the equalizing parts.
The best repair is to salvage a 10 amp diode out of an old computer power supply and replace them with it.

[cmjones01]

Interesting but not surprising. In any switch-mode power supply any extra resistance could have a disastrous effect on efficiency and maximum power output. They'll just rely on the diodes being pretty closely matched, which isn't a problem with modern components as long as they're not over-run and have decent thermal contact. I've used power (3W) LEDs in parallel on the same heatsink, always from the same batch as each other, and haven't experienced any thermal runaway effects in the hundreds of pairs of LEDs I've tested so far.

Chris

[G6Tanuki]

Fitting traditional-style equalising resistors in series with paralleled diodes doesn't play sensibly when you're rectifying the output from switched-mode supplies.
You need to consider the effective parallel-capacitance of the diodes and their 'recovery time' - and think of the AC you're trying to rectify as RF rather than mere 'AC'.

My approach would be to fit a pair of diodes from the same production-batch [adjacent items from a bandolier? they'll probably be pretty well-matched] of whatever diodes the original manufacturer fitted - unless, that is, you *really* want to spend time and do some serious RF circuit-analysis and so prove a better solution to the original.

[SiriusHardware]

A few of us, myself included, have encountered this 'one of several parallel diodes in the LED backlight supply short circuit' syndrome, as shown in this thread:-

https://www.vintage-radio.net/forum/...d.php?t=109045

In my case there were spaces on the PCB for three diodes but only two fitted - the third, presumably, would be fitted in a larger set with more LED backlight elements to power.

I fitted the third diode as well in the hope that three would 'load share' more than two - I also replaced the original surviving diode so that all three diodes fitted were now from the same batch.

It could also be argued that I've actually increased the chance of a future failure since failure of any one of the three will disable the set, whereas before there were fewer diodes which could fail.

[Goldieoldie]

Hi,
You can araldite the diodes together (so they are touching each other ) so they are all at a similar temp which will help

[Wendymott]

Thanks for all your comments. Chris, not sure if the diodes would be "auto inserted" as they are large items, but probably from the same bandolier, thus batch.

I have ordered 4 from an ebay supplier and should arrive weds, Ok Refugee a 10A device would be ok too, but the computer psu's will be of lower voltage; these diodes are rated at 200v, and I know that 60V devices did not work.

I have been associated with TV manufacture in various ways since 1979 and I am never surprised what the designers come up with. They work ok in the lab, even burn in testing, but get them in the field... a certain Turkish manufacturer, who is well known for OEM badging, has made the TV repair engineers a good living with their poor designs.

[Refugee]

I would not expect to see more than 40 volts on an LED power supply. I would have expected a computer diode to be fine. It muse be powering something else that shuts the LEDs down when it fails.

[SiriusHardware]

Originally Posted by Refugee

It must be powering something else that shuts the LEDs down when it fails.

In the example I saw the PSU section using the parallel diodes (one of which failed) was dedicated to supplying the LED backlights only.

Having used a 60V variable voltage / variable current bench power supply to test strings of backlight LEDs individually, I agree that 50V is typical.

However, since one of the PSU parallel diodes had failed short-circuit and posts here on the forum indicated that this was a very common problem, I chose to replace mine (all of them) with diodes which were significantly higher current and voltage rated than the original ones.

[AC/HL]

I've just repaired a caravan power supply for a friend. It has two 1N5351B 14 Volt Zeners in parallel across the output, and one had failed S/C. I replaced them with a new pair, and they seem to run at roughly the same temperature (finger test).

[Maarten]

I agree on the 'certain Turkish manufacturer'. Vestel sure was record holder for designs with the most parallel diodes. I think they used as much as 5 or 6 in some designs. Needless to say, one of those inevitably shorted out and blew the primary part of the supply.

[Wendymott]

Maarten, not only diodes but generic faults, such as the "S" correction cap in the 11AK19 going dry, the 11AK33 Frame circuit, the 11AK07 SMP transformer, I could go on.. but its getting "off piste". OK Refugee, I have not measured the actual DC feed to the LED PSU inverter, but I do know SR560 diodes rated at 60 VRRM were not suitable.

[Refugee]

I have just had a quick look at new diodes and there are plenty 200V parts listed at between £1.50 and £2.00 in TO220 packages that would replace the pair you have found.

[Wendymott]

New diodes arrived and fitted; clamped together with a copper strip and thermal grease.
See how that goes.

[SiriusHardware]

Originally Posted by AC/HL

I've just repaired a caravan power supply for a friend. It has two 1N5351B 14 Volt Zeners in parallel across the output, and one had failed S/C. I replaced them with a new pair, and they seem to run at roughly the same temperature (finger test).

Honestly, I would have thought these would be intended as brute-force output overvoltage protection, and would not normally conduct at all as long as the regulated output (12V?) remains below their threshold voltage.

As such, I wouldn't expect them to be passing any significant current or getting warm at all when the unit is operating correctly. I suspect they are only intended to conduct when something goes wrong and the output gets too high (ie, failure of the main regulating device short-circuit from input to output, in the case of a linear regulator).

[AC/HL]

Me too, but they were regulating. The transformer seems to be sized to produce just over 14V with no load. Hefty diodes (actually a bridge with only two diodes used) and transformer give reasonable regulation. The only resistance was a PCB track which had burnt out. It's rated at 10A, but I couldn't load it so don't know how it sags, only that it's clamped at no load.

[Radio Wrangler]

A lot depends on the circuit used in an SMPS. Some converters need diodes which will withstand reverse voltages equal to the peak input voltage plus the output voltage plus some margin, others may need less. Snubbers waste power but also damp transients and rings. Some people don't design them in....

Diodes and bipolar transistors in parallel without deliberate current sharing arrangements are bad risks which competent designers should know not to take.

Larger diodes (or bipolars), big enough to do the job in one part, can be viewed as a lot of diode areas in parallel. But they are on the same die to better equalise temperature and the semiconductor designer has planned the metal patterns and bond wires to act as balancing resistance, just enough for keeping temperature stable and equalised well-enough.

With transients and pulses you can get fast localised heating, and due to the timeconstants of heat flow, the stability is less, so you can get hot-spotting and very localised runaway unless the current sharing on the die is carefully designed. This is part of the mechanism involved in 'second breakdown' modes.

Multiple parallel devices unless definite action has been taken to force current sharing are just sitting there with a zipper-like failure mode ready to roll at any time.

David

[short wave]

I started a thread on parallel diodes (in a bridge rectifier) some tine ago may be on similar lines
link
https://www.vintage-radio.net/forum/...d.php?t=100385
may prove useful
S-W.

[Michael Maurice]

The other problem is that very high powered fast rectifier diodes tend to be quite large and to make these sets ultra slim, they prefer not to use them.

[Wendymott]

Well it's all academic now. I fixed the fault with two new diodes as described above.
TV died with an unrelated fault, PSU's ok, all onboard mini psu's ok. Scrapped.
Ordered a replacement. Thanks for all your inputs.