BBC Micro vcc short

[GrahamN]

First time I've come across this - I am working on an Issue 4 BBC Micro and there is a dead short between vcc and ground on the board.

I've removed all socketed chips and tried using my bench power supply and a thermal camera to try and see if anything is getting hot, but with no success.

I'm close to scrapping the board unless anyone has any ideas of how to isolate the area where the short is? I'm guessing a shorted capacitor, but am at a loss as to how to find it without removing potentially a large number of them (I can't find any damage or obviously stressed parts.

Any thoughts welcome!

[SiriusHardware]

By dead short, do you really mean 'dead', like no discernible resistance? Zero ohms? In a case like that the short is nearly always a 'metal' short like a solder whisker or solder splash. When components like ICs fail they rarely fail to zero ohms and if you are lucky will heat up.

Capacitors can potentially fail to absolute zero ohms if there is a failure of the insulation separating the plates but in my daily experience even capacitors more usually fail to a measurable low resistance, sometimes only one or two ohms but not dead short.

I have a couple of BBC Bs but they are out of the way at the moment so I can't recall whether the supply decoupling capacitors are Tantalums but if they are those are famous for failing very low resistance.

[SiriusHardware]

I've just found an image of a BBC motherboard online and I see that while the majority of the decoupling caps are non-electrolytics there are a relatively small number of the notorious blue tantalums as well.

Desolder one leg of each of the tantalums and see if your short goes away.

I believe conventional wisdom is that those blue Tants should be replaced on sight, in any case - not just in these but anywhere they are found.

[julie_m]

Tantalum capacitors are definitely a "don't ask, just replace" part.

Capacitor technology has kept improving steadily over the years; and whatever reason (leakage current, ESR, physical size .....) anyone might have had for choosing specifically Tantalum parts 40 years ago, modern electrolytic capacitors of a decent quality will do at least as good a job.

[GrahamN]

Thanks

I'll certainly try the tantalum capacitors as there are only a few of them.

In answer to your question regarding the resistance, the short gives exactly the same result as does shorting the test leads together (about 0.4 ohms according to my meter). The board hasn't had any work done to it at all and there are no discernible shorts on the underside.

[SiriusHardware]

Even with no rework ever done on the PCB, solder can sometimes grow crystalline 'whiskers' which gradually grow longer and eventually touch another solder pad.

In this case though I would investigate the Tantalums first.

[GrahamN]

Good advice - thank you. I removed 4 tantalum capacitors, checking for the short after each one. The fourth was the charm - the short has now gone.

I'll now refit the chips I removed and hopefully will either have a working board or at least something I can repair without tearing my hair out.

What I don't understand is how I've never come across tantalum capacitors failing like this in the past. I've been repairing computers and game consoles as a hobby for about 40 years and regularly found ceramic and electrolytic capacitors shorted, but never tantalum. It sounds as though I've just been lucky (or unlucky depending on your point of view!)

Anyway, thanks again for the assistance.

[SiriusHardware]

I'm glad that suggestion saved you a bit of time.

As Julie said, you might, while you have access, replace all of the few Tantalums, not just the faulty one, with good quality modern electrolytics of the same voltage / value.

If you think about it the remaining Tants are exactly the same age and type as the one which failed. They could be a problem waiting to happen.

[GrahamN]

And I have a working board - thanks again.

I will go through and replace all the tantalum capacitors as you suggest. Do you think it worth replacing them with new tantalum capacitors? Or should I just go to electrolytic?

[Sideband]

I always remember the quote by a certain Les Laurie Johns in TV magazine back in the 80's... 'Beware the blue tants'!

Use good quality low ESR aluminium electrolytics for replacements.

[julie_m]

One Tantalum capacitor failing is a sure sign that the others are waiting their turn.

In my experience, they wait till the last screw is tightened and lined up with the others; so if in doubt, whip 'em out.

There's almost certainly no need to replace with Tantalum. Modern, quality electrolytics are at least as good as 40-year-old "Tants" were when they were brand new.

[Keith956]

Must confess I've never had a problem with tantalums, despite having a couple of 80's Datron calibrators which contain dozens of them. Maybe I've been lucky.

It would seem that when they do go, they go short circuit, which is obviously catastrophic on power lines, whereas electrolytics seem to mainly go high ESR and leak.

The other thing I've read somewhere is that tantalums need to be derated more than electrolytics i.e. use a voltage rating of at least 50% more than the circuit suggests.

[SiriusHardware]

Do you think it worth replacing them with new tantalum capacitors?

Honestly, no, unless you are concerned with keeping the historically correct appearance of the PCB (and some people are, which is fair enough).

As per Julie, just use good quality modern 'conventional' electrolytics of the same voltage and value.

[Michael Haardt]

It's already solved, but this kind of fault can be traced by some extent with an Ohm meter that can measure below 1 Ohm. The closer you come to the fault, the lower the resistance from traces will be.

Adrian Black shows this on his YouTube channel occasionally.

Michael

[ortek_service]

Originally Posted by Sideband

I always remember the quote by a certain Les Laurie Johns in TV magazine back in the 80's... 'Beware the blue tants'!

Use good quality low ESR aluminium electrolytics for replacements.

The Blue Thomson? ones in Racal RA1792 etc. Receivers are quite notorious for failing - Usually short-circuit, is the usual failure mode for these and tend to just go suddenly and don't just degrade over time.
The Yellow AVX-Kyocera etc. ones tend to be much-more reliable.

Strangely, I've never had a Tantalum fail in a Beeb but I've had them go in quite a few other things (Including getting very-hot and burning-up, sometimes damaging the PCB).
Maybe Acorn de-rated them better (It's often good practice to at least double the working voltage) - and you only need to exceed their rated voltage slightly for them to break-down.

There was also (Post Office?) recommendations about putting 10R in series with them, to limit surge-currents that they are not too-good at handling compared to many Aluminium Electrolytics (Note: Tantalums are technically actually also Electrolytics, due to how that's defined).
But when you could get (expensive) ones with built-in fuses, you do have a feeling they will never be that reliable.

Tantalums did / do have some advantages over Aluminium Electrolytics - like tolerance and stability over time & temperature, although often higher ESR than good quality Aluminium Electrolytics.
But are now often replaced by ceramic capacitors (often due to cost with Tantalum being a relatively-rare material only found in some Countries often in conflict), after ceramics have had a significant increase in their capacitance-density - especially in Surface-mount packages.
However even with a reasonable dielectric there are still fundamental physics piezoelectic effects that can cause a high variation of capacitance with voltage that Tantalums don't have, meaning you actually get rather-less capacitance (can be 1/5th to 1/10th! at full rated-voltage)
Plus sometimes ceramic capacitor ESR can be too low for circuit stability, so an 'Electrolytic' type is with a bit of ESR is required.


Standard Electrolytics do have their own issues, and were originally not always that low-ESR. So you sometimes find (often Green coloured) Low-ESR / 'High-Frequency' special ones. And it was said that the ones in the Beeb PSU were these and standard ones couldn't be used. But I compared datasheets on originals with modern ones, plus measured their ESR, and have found most modern standard ones to measure as good.

Ideally 'solid' / Polymer / Organic Semiconductor (OSCON) types should be used, for much greater performance / life. But most development tends to occur on Surface-Mount types, with leaded ones mostly still conventional 'wet electrolyte' types that have a fairly-limited lifetime.

And there's now a Niobium oxide replacement for Tantalum (but mainly in SM packages?). As "Niobium is a widely available material and has several advantages over tantalum including a non-short circuit failure mode".
https://www.kyocera-avx.com/resources/mlcc-tantalum-interchangeability/

[ortek_service]

Originally Posted by Michael Haardt

It's already solved, but this kind of fault can be traced by some extent with an Ohm meter that can measure below 1 Ohm. The closer you come to the fault, the lower the resistance from traces will be.

Adrian Black shows this on his YouTube channel occasionally.

Michael

You may also find that the (virtually s/c) one starts to get quite hot / smokes a bit. So may be able to detect this (particularly with a thermal camera, if access to one) after powering on for a short while - Assuming PSU doesn't just trip, due to excess current. Although could be a risk of PCB tracks burning-out / damage to PSU if current is too much.

The Beeb circuit / PCB is unusual in that there are 3 separate +5V power feeds to it, to minimise voltage drops on a large 2-layer PCB. And I believe these are all isolated on the PCB, so it won't work if one is missing, but it does make tracking down the fault by resistance measurements a bit-easier as will only be around a third of the Beeb's approx 100 IC's

[SiriusHardware]

For a component to heat up, it has to have a little bit of resistance so that it will dissipate some Watts. In this case it seemed that the guilty Tant had gone genuinely zero-ohm short circuit so it was impossible for any voltage to develop across it, therefore it was never going to get warm.

As far as the +5V rail being separated into three 'zones', I'm not to too sure about that - the 5V supply certainly is distributed by 3 x black wires and 3 x red wires landing on different areas of the PCB but I think all of the +5V feedpoints and all of the 0V feedpoints are still commoned up on the PCB - I'll have to check.

[TonyDuell]

I think early 'issues' of the BBC micro PCB had the grounds commoned but the +5V inputs separate. This was for the 'table-burner' linear power supplies which had separate 7805 regulators for the 3 5V feeds.

Of course the Astec SMPSU supplies all 3 5V wires from the same output

Later issues of the BBC micro had the 5V inputs linked on the PCB.

[SiriusHardware]

Thanks for that clarification - mine are issues 4 and 7 respectively, so that will be why I haven't encountered the 'split' arrangement.

[ortek_service]

IIRC, VCC(1/2/3)y are still separate on the Schematic, but not sure what Issue board that was meant for (There seemed to be mainly one version, that was in the back of the Advanced User Guide. But there were different issue Service Manuals, so maybe these had different versions of the schematic?)

However, as the schematics and the PCB were created separately, rather than on a full PCB design E-CAD system that used a common netlist to verify both against, then there's going to be a few slight differences / errors.

Even on the Linear supply, they were still common-ed together at the PSU end and didn't have separate regulators. So it was just a way of more-evenly distributing the power / grounds across the PCB to minimise voltage drops on that (as well as effectively tripling the wire CSA, by kind of paralleling the three wires / splitting the load across these)