BBC Micro switch-mode PSU blows fuses

[icon]

Hi

My BBC Micro, recently pulled from the loft for the entertainment of the boys, blew it's 500mA PSU fuse whilst playing Chuckie Egg.

Now all the references to this PSU on the Internet are about a slightly different version to mine, which is plagued with exploding RIFA X2 caps. I've attached the schematic for this, though to stress - it's slightly different to mine. I have one made by "Wong's Elect. Co. Ltd", and I can't track down a schematic for it. Mine has an optocoupler between high- and low-voltage sides, for example. I've attached some general shots, so you can see the kind of thing I'm dealing with.

What I've learned to date: there's no obvious short when powered off. The bridge rectifier's fine (works at 10V - might it not work at 240V?). The TO3 transistor (NEC C1358, aka 2SC1358) is fine, I think - B>C voltage drop of 0.51V, B>E voltage drop 0.54V. High voltage electrolytics measure the right capacitance. None of the others look distressed. No other damage I can see.

So I'm a bit stuck. Can anyone suggest what might cause this PSU to blow its fuse? I'm keen to get back to Chuckie Egg.

Thanks
John

[amyk]

I'm assuming a replacement fuse also blows? More pictures of the backside and we can probably reverse-engineer a schematic.  It doesn't look like anything exotic - probably a standard self-oscillating flyback or forward converter.

That attached schematic is horrible --- are those crossed wires actually connected or not!?!?

[sleemanj]

I'm assuming that you know the fault is in the power supply and not something said supply is powering.

[Ton]

I see there is a scr crowbar circuit on the 5v output, and since it is an old computer from the attic, I am kind of suspecting this to be linked, I also assume that the PSU works with no or light load, or it will with new caps.

Old electrolytic capacitors might have partially dried out during attic time, so just booting up and doing small stuff gives the impression everything is fine, but when pushing it with a complex program the total current consumption loads the tired caps so much that the ripple trips the crowbar circuit, and byebye fuse

If I read the pictures right then the lone to220 in the middle of the secondary section is the SCR crowbar, and if it is triggered in the same way as you example schematic then as soon as the 5v reaches the trip voltage the scr shorts the 5v output.

I would check all caps for value and ESR, if this is not possible then replace the caps in the secondary side of the PSU and also replace the caps on the main computer board, if it is hard to replace the caps, then as minimum reinforce as meny as possible of the power caps on the main CPU board with a new good quality low ESR cap in parallel

Well that was my best guess based on the available information .

[Macbeth]

Something that may help isolate the problem - if the beeb isn't powering up then it's the +5V on the secondary. The -5V is only used for the RS423 UART interface and the +12V is for the external floppy drive IIRC. As for Chuckie Egg - good call! I thing I need to find it and run it in BeebEm on my PC. Hours of fun. Also, Repton was a fave back in the day!

[Macbeth]

I'm assuming a replacement fuse also blows? More pictures of the backside and we can probably reverse-engineer a schematic.  It doesn't look like anything exotic - probably a standard self-oscillating flyback or forward converter.

That attached schematic is horrible --- are those crossed wires actually connected or not!?!?

I agree - they should have blobs on the multiple connections or even the old fashioned half-loops jumping over the intervening wires. However, using common sense it's easy enough to see that putting dead shorts across the outputs for example is nonsense, so it makes sense in the end. Maybe someone should put blobs on the diagram.

Make it a competition, we all have a go and vote the most likely! 

[Macbeth]

Here is my quick and dirty attempt. I'm sure I've missed a few junctions!

[icon]

Hi

Thanks for the replies so far. In no particular order:

Please note - the schematic (and it is awful, isn't it?) is NOT exactly the same as mine. I can't find the precise schematic for my "Wong's" PSU.

I *was* sure that I had blown a fuse with the supply disconnected from the board, but now I'm beginning to doubt myself, so I'll solder it back up and check again.

The lone TO220 on the low voltage side is actually a 78M05 which supplies the -5V. The 'output' is taken from what would be the ground pin, if that makes sense.

I was having a go at powering the board from my bench PSU. What would you expect the board to draw from the 5V rail? 1A? 2A? 3A? It goes into current limiting at 2A, but I don't want to just keep turning the wick up!

I will take a photo of the copper side and post it later.

Thanks
John

[icon]

Right

I tried a new 500mA fuse. Bang. In the interests of science, I tried a new 2A fuse, with it hooked to a multimeter in min/max mode. It drew just over 500mA at start up, but stabilised at around 1mA! There's nothing at all on the low voltage side. I had a poke around with a scope - I see the bridge rectifier doing its thing, but everywhere else I look I only see some variation of a 50hz (UK) wave form of a variety of different shapes - I presume the base of the big NPN transistor ought to be buzzing away at some high frequency?

I've had a go at providing some views of the board, one of which attempts to show both sides at once - contrast's not great between the copper and the solder mask.

Tell me what to poke, and I'll go poke it!

Thanks
John

[grumpydoc]

What would you expect the board to draw from the 5V rail? 1A? 2A? 3A?

2-3A is probably about right.

When the PSU current limit kicked in, what was the voltage?

[sleemanj]

I would replace the electrolytics, if nothing else for a couple bucks you remove them from the equation.

[planet12]

The original PSU was specified as 3.75A on the 5V rail. You might get away with less if you don't have it loaded up with peripherals and sideways ROMs.

-5V allows for 100mA.

12V allows for 1.25A.

See section 3.14 on page 25 of the model A/B service manual, a PDF of which is available from http://acorn.chriswhy.co.uk/docs/Acorn/Manuals/Manuals.html

If you want to test your supply without plugging it into the motherboard, you'll want to provide a small load on the 5V rail to ensure it doesn't trigger any overvoltage protection. A 10 ohm 5W resistor will provide >10% of rated load, which should be plenty for stable operation. Also has the advantage of mistakes not blowing up your Beeb.

Oh, how I miss my Beeb at times. </nostalgia>

[calexanian]

Why did they even bother with a switching supply? A standard linear would have been smaller and cheaper! Probably because it was just what apple did is my guess.

[planet12]

Why did they even bother with a switching supply? A standard linear would have been smaller and cheaper! Probably because it was just what apple did is my guess.

Some of the early ones did in fact use a linear supply, but it was changed out for the switching one relatively early on. I'm guessing it was a case of a 35-40W flyback supply becoming more economical than a linear - not to mention lighter. By the time you put a margin in for the dissipation in the linear pass elements, you're probably talking a 45-50VA transformer, which is a reasonable sized chunk of iron, and then add to that heatsinks for the pass transistors...

[calexanian]

True. Weight = shipping cost, etc etc....

[icon]

Hi

I've decapped the power supply and I'm ordering new ones, so I'll hold off my cries for help for the moment.

Meanwhile, I've found that the motherboard works fine, drawing about 2.1A. Worst case scenario, I could just use an 'offboard' PSU from an old desktop!

Regards
John

[netdudeuk]

If you want sound you'll need -5 volts as well.

I know this because in December 1981, I took a rare (at that point) Model A to Acorn in Cherry Hinton with a faulty tape interface.  They didnt know what was wrong with it so offered us another machine.  We asked them to check the tape interface on that one before we took it.  They loaded the Welcome tape and I commented that I'd yet to hear any sound from my own machine.  Straight away, both guys came up with 'it's the -5v volts'.  A quick check then confirmed a dodgy connection on that rail and we took back our original machine.

[icon]

If you want sound you'll need -5 volts as well.

Chuckie Egg without sound?! Are you MAD??!!

PC supplies normally have a -5V rail, don't they? Still, that's defeatist talk; besides I've just spent more than the price of a cheap PC supply on a fistful of new capacitors...

John

[icon]

I'm back. Putting all new electrolytics didn't fix anything. I've rechecked the power transistor - detached it from the circuit, hooked it up to a sig gen and oscilloscope; all fine. There are no obvious shorts across any diodes or transistors. But the damned thing just won't oscillate.

In the meantime, I've re-purposed an ATX supply and I'm powering the BBC Micro off that - so I know that works!

I'd be grateful for any pointers as to what might be wrong, or what I can check.

Thanks
John

[planet12]

I think given you've already identified that it's not quite the same circuit as the schematic you've found, the next step would be to reverse-engineer your own schematic... preferably in a tidier form than the original!

The hardest part to replace would be the transformer - luckily that's also probably the least likely thing to have failed.

[Zad]

Check the triggering voltage of the output protection zener + SCR and make sure it is something sane. It is easy enough to rig up a low current variable low voltage supply, if need be a 9V battery and 1K potentiometer.

Try pulling the output filter inductors L5+L6+L7 (one at a time) and see if it still fails. That would potentially reduce the fault to one specific output channel.

[icon]

The hardest part to replace would be the transformer - luckily that's also probably the least likely thing to have failed.

Oh I hope you haven't cursed it! I've been puzzling out the schematic, and I've got to the transformer. On the low voltage side, the taps seem to be wired as adjacent pairs. Assuming this is the case for the HV side too, there's an open circuit between two adjacent pads:



Does this mean I'm doomed?

And in the meantime I've found a 1.0 ohm 1W resistor that has failed open circuit - my excuse for not finding it earlier is that it was wrapped in heat shrink. Grr.

John

[SeanB]

Not the same winding for the primary, there they often leave pins unconnected, and have common pins for a few windings and other strange ways all to increase separation between the high voltage spike on the primary and the feedback and supply windings. The heatshrunk resistor likely is a current measuring unit in the primary side, it being open will result in fried silicon and blowing fuses.

[planet12]

I totally cursed it. That's how I roll.

You might need to remove the transformer to figure out the winding scheme. You'd then look at the wire terminations and match things up by looking at wire gauge, insulation colour, and so forth - along with checking the resistances of the windings (and inductances if you can - you can work out winding ratios from that).

Other than that, keep pushing through the rest of the circuit and update us with what you get. It's likely some of us will be able to take a pretty good guess at the transformer once we see what the rest is like.