BBC Micro Power Supply advanced repair

[dmantione]

Hi!

My first post on this forum. I am normally an active member of the Commodore community, but some time ago I bought a BBC Micro on Ebay. It is another interresting 6502 based computer after all, quite a different take on a computer than i.e. the C64, but that is actually what makes it interresting.

It turned out my purchase was a bad luck and I had to some advanced repairs on the power supply and also on the machine itself. Of course this is part of the hobby are some hours, but it also was a real challenge. I'd like to share my repair with you and gather feedback.

I was prepared for some power supply trouble, as this seems to be a common disease for these machines. But I started the repair and kept replacing part afwer part. In hindsight I would have thrown away the power supply and replaced with a new one, but for its time it is a reasonably advanced power supply and therefore there is some historic interrest in keeping it original. So I set myself to work on it couldn't know broken it was.

Let's start with the broken components that I found:


From left to right:

• C1 mains filter capacitor
• C2 mains filter capacitor
• R12 current limiter?
• R11 current limiter?
• Q2 Main switching transistor
• Q1 Auxiliary transistor
• D7 rectification diode
• SCR1 part of overvoltage protection circuit

C1/C2 are well documented and have been replaced with new X2 capacitors. The other components are less well documented and if you have feedback on my choices here this is welcome.

I replaced R12 with a 4.7 ohm 1W resistor. Because I believe the fuse of the power supply will provide protection in case of short circuit, it is a normal resistor and not a special safety resistor. Same for R11, it has been replaced with a 0.75 ohm 1W resistor. I did mount these resistors elevated so they can dissipate their heat easier if they get hot:



For the main switching transistor Q2 I found a suggestion to use a BU208A, so this is what I used and it seems to work. I not however that this transistor is rated 700V, while the original 2SC1942 is rated 800V, so it is a downgrade, and I have measured that the transformer generated spikes up to 540V at the collector of the transistor. So, while it is in-spec, this modification reduces the safety margin.



For the auxiliary transistor Q1, I have used a PN2222A, this seems to be close enough in specification with the original 2SC210 to me.

Rectification diode D7, originally an S10SC3M, was also broken and did provide little more than some resistance. I have replaced with an SBL2040 which seems fine to me, but I am puzzled why the manufacturer did use differed types of diode for D7 and D8, this makes no sense to me, as both rectify the output of the transformer. I very much like to know the reason, as this may affect the choice for the replacement diode. For the time being, the SBL2040 works just fine.

Lastly, for the thyristor SCR1, I did use a TNY612, which appears to have the right properties.

After all the work, it finally worked. I let it run a halogen lamp for an hour and after it did that succesfully I could connect it to the mainboard again:



I did create a SCART cable out of a Commodore serial cable and testing:



How much bad luck can you have? The machine had broken RAM. Now I'm quite experienced in diagnosing broken RAM chips on Commodore 64s and that knowledge helped me to diagnose the problem quickly on the BBC mainboard as well, compared to the effort of repairing the PSU, this was minor. I have temporarily used a 4164 RAM chip as repacement, I'll replace it with a proper 16 Kb RAM chip later, but at last, success!

[eelco108]

C1 and C2 are the usual suspects (and obviously so), and Q2 can be, but how did you conclude that the other components were broken?

I've only done six power supplies so far (takes some time anyway), but usually C9 and all the other electrostatics are the best ones to replace.

Old caps could mimic some of the issues you think you might have spotted with the regulators, for example. Looking at the pictures, did you replace all big caps anyway (including C9)? They look new ....

[dmantione]

Q2 was a short circuit, so I guess it blew R11/R12 when it did bang. When testing Q1 with a multimeter diode test, it also didn't behave like a transistor, so was broken as well. When this was repaired, I noticed the PSU was switching, but there was no voltage on the output of the transformer. The cause proved to be that D7 was conducting in both directions, so the output capacitors would never get charged. While desoldering, I also desoldered the other TO220 components out-of-circuit, and SCR1 did fail testing.

I indeed did replace one of the output capacitors, forgot to mention this. It wasn't broken, but I didn't like its ESR so I decided to replace it. All other capacitors were testing okay in capacity, ESR and leakage current.

[eelco108]

Thanks for explaining - that was one hell of a broken power supply!

Maar hij doet 't weer

[tricky]

I've had my hands on over 30 PSUs of Beeb's that were new to people, 28 were just C1, C2 and C9 for good measure, one on a master needed the above and a new 1 ohm resister and one is waiting in the repair bin. The symptoms of the master were that it would only work after being off for a day or after I had discharged its caps.
I would say you have been unlucky, but a great custodian of the beeb.
RAM faults do seem to be more common, but of the machines I have worked on, only one B+ (mine ) has had RAM faults.
Two machines have had blown buffer chips, two have had dead or dying 6522s (one beeb had both dead), two have had dead D7002 A2D chips and one had dead OS and BASIC ROMs and lots of ICs and a few keyboard cables needed reseating. Quite a few of the dodgy chips were in a single beeb, which I suspect had beeb a donor for several transplants.
In my experience, as the price of Beeb's rose, so did the number of "untested" (faulty) Beeb's being sold on eBay.

[dsdf323]

Thanks for the detailed analysis!

I'm working on a PSU I just received that's not outputting anything. I've swapped all the capacitors as I usually do and nothing. So I was going to look at Q2 next, although both the BU208A and 2SC1942 are no longer available so will need to figure out an equivalent, but I'll also run through the other parts too.

[1024MAK]

I'm working on a PSU I just received that's not outputting anything. I've swapped all the capacitors as I usually do and nothing. So I was going to look at Q2 next, although both the BU208A and 2SC1942 are no longer available so will need to figure out an equivalent, but I'll also run through the other parts too.

On one electronic suppliers site, a BU508AW is shown as an alternative to the BU208A. Please note that I have not compared the relevant data between these transistors. So please do your own checks before buying and fitting.

Mark

[dsdf323]

On one electronic suppliers site, a BU508AW is shown as an alternative to the BU208A. Please note that I have not compared the relevant data between these transistors. So please do your own checks before buying and fitting.

Thanks Mark.

I put this PSU aside for a while and have now come back to it. Still puzzled. I had it and a working one side-by-side to try to compare a bit easier.

The faulty PSU still shows nothing on the LV side of the transformer. Checking on the HV side I can see 103V on all points, whereas on the working PSU I see 103V and 118V (or was it 115V).

Where the working PSU shows 115V, the faulty PSU shows 103V, which is also where Q2 is. So I'll look again at that replacement.

Brett.

[dsdf323]

I've found a "local" (i.e. Australian) supplier (Wiltronics) for the BU508A which comes in the same TO-3 case as the BU208A to make the switch easier, so let's see ... it's coming from regional Victoria to regional NSW ...

BU508A in TO-3 case

Brett.

[richlane]

Hi, this is my first post.

I can see this thread is fairly old but it's interesting to me as I am in a position where I have a non-working BBC power supply which I am trying to fix.
Is anyone still around who commented on this thread?

After getting white smoke with a machine I acquired (I had no previous BBC knowledge) I started looking into this and have now replaced the two RIFA caps and all the electrolytics in the PSU.

Then for a while it all worked but after a few weeks I now get an intermittent issue on the 5V rail. Looking on a scope I can see the 5V rail is constantly collapsing to around 1V and then restoring back to 5V. Obviously the BBC fails to boot. BBC is fine (with sound) with a bench supply on the 5V and PSU 5V disconnected. So the -5V is OK. Didn't check the 12V but I probably should do. The BBC main board is taking about 2.2A from the 5V bench supply.

Next step is to replace some semiconductors which is why this thread is interesting.

Anyone have any suggestions?

Regards
Richard.

[Boydie]

When I had something similar, it was Q4 that was funky. IIRC its job is to produce a stable output, the value of which depends on the resistance of one of its other connections.

I think I’ve already posted somewhere about what component this is - despite appearances it’s not a transistor and is itself unlabelled and not detailed on any circuit diagram I’ve found. I took a lucky guess from an Apple II PSU diagram (also made by Astec) and it worked.

[1024MAK]

Hi, this is my first post.

Hi Richard

We prefer members to start a new topic/thread for hardware repair rather than adding to an existing topic/thread. It saves confusion.

Mark

[richlane]

thanks I will look into Q4 and try to test it on my PSU. I looked waveforms again on a scope today and in fact all outputs (5, -5 and 12) are not stable. 5V is switching at about 12Hz. This is also intermittent, sometimes the rails are stable and the bbc boots ok.

[richlane]

point taken Mark will do in future

[GoodToGo!]

The most recent BBC I received had the usual failed C1 and C2 capacitors. However once replaced the 5v rail stayed low (around 3-4v). Eyeballing the rest of the electros revealed no abnormalities. I was convinced it was a secondary side issue. However many posts I read kept referring to C9. It looked normal by eye, however there was the typical signs of electrolytic leakage around it. Testing it on removal revealed correct capacitance and the ESR was reasonable at 0.3ohms. Replaced it anyway with a new one (0.1ohms ESR) and the power supply returned to life. 5v rail rock solid. I'm still amazed at how critical this one capacitor is to the feedback......

Cheers,

GTG!