Backplane issue, or have I missed something?

[Chr$]

Hmmm, well my 9v seemed steady and good, but there's something odd going on. I removed the TIP112 transistor that was supplying the 9v rail. It's a bit of a bodge by the previous owner. With that removed, the backplane doesn't seem to work at all, in that no cards plugged in are recognised on boot.

The TIP112 was mounted on a chunk of aluminium, which is grounded, but I don't think it actually should be because now when I re-mount the TIP112 there is a short and the PSU won't even start. There was a plastic thing around the screw so perhaps the TIP112 was never grounded (i'm sure it was touching the alu gnd though). This is my amateur diagram of the arrangement:

There are 3x 1N4005 diodes in series between pins 2 and 1. Pin 2 gets 12v from the anode of C37 on the motherboard. And at pin 3 there is 9v, which goes to the 9v rail on the backplane card.

I assume there must be a better way of getting 9v to the backplane?

And before I removed that and with just the QubIDE plugged in it showed ROM version 3.10. The GC had v2.49 but I also tried v2.27, which is the one it originally came with.

[Pr0f]

the pin outs are B C E in the order left to right - and tab is connected to pin 2 (collector)

More worryingly there is a diode connected between C and E which would provide about 11v to your 9V circuits !

Dataasheet attached:

tip112.pdf

(124.44 KiB) Downloaded 93 times

Ignore the comment about the diode - the pdf shows the diode correctly oriented - the piccy on the web has it the other way around.

The 3 diodes act a dropper to drop about 0.75v x 3 - 2.25v with the voltage drop of the darlington pair in the tip112 this will be about 9v on the emitter with the collector at 12v.

[Chr$]

tab is connected to pin 2 (collector)

So the tab was never intentionally grounded then as that shorts the 12v input on the collector. That's odd, it looks like it's right against the metal in this original pic, even though the maker did use a plastic washer. I'll have to investigate that aspect further, but it still doesn't work with both cards attached.

[Pr0f]

Usually when fitting a transistor or regulator in those TO-220 packages - you would use a Mica washer - which are thin and see through -so not always obvious, these conduct heat very well - but will insulate electrically. Also there would be an insulating bush that allows a bolt to pass through the tab without contacting it - you can normally get a kit of both bits. Some newer kits use a silicon material which is a bit thicker, but also conducts heat well.

[ppe]

There are 3x 1N4005 diodes in series between pins 2 and 1. Pin 2 gets 12v from the anode of C37 on the motherboard. And at pin 3 there is 9v, which goes to the 9v rail on the backplane card.

I'm not much of an electronics guy but that looks like a kind of an emitter follower where emitter voltage follows base voltage. Ve = Vb - ~0.6V. The diodes supply the reference V to base: Vb = 12V-3*(Vf of 1N4005) = 12V-3*~1V = 9V so theoretically your 9V rail should see something around 8.4V ? Depending on actual values of individual components.

So, it's kind of a homegrown voltage regulator

I assume there must be a better way of getting 9v to the backplane?

Might make sense to replace the whole thing with a proper regulator. If heat is not a problem and you just want a quick fix then the classic 7809 plus a couple of caps would work but you could go fancy and drop in a modern LDO or a switching regulator with any associated support components.

Someone who actually knows electronics probably can correct me if I'm talking absolute nonsense.

[Chr$]

It was spot on 9v output. But as Pr0f says, I think it had some kind of insulation that I have carelessly lost so I will replace with a 7809, which can conveniently be attached the the big metal sheet behind the motherboard as gnd and a heatsink combined.

[Pr0f]

A 7809 like the TIP112 circuit will need to dissipate a few watts - the current limit for the darlington transistor is specified as 2A, so I am assuming the 9V supply uses less than that - probably closer to 1A - so with 3 volts dropped - that's 3 watts as heat.

An LDO regulator wouldn't provide any real benefit - but maybe a switch mode supply that regulates 12V to 9V might be more efficient, but if you've got the metal work and are not worried about a bit of wasted heat, then the regulator is good option - plus the tab is connected to the common terminal pin2 - which would be ground. Solder small value capacitors between common and input and common and output - close to the regulator - helps prevent oscillations - 1uF to 10uF would be fine. It probably mentions this in the datasheet for the regulator anyway. a 7809 would be rated at 1A, so may be close the limit on current - best to watch out for that. I think the qubide boards take no more than 100-200ma from my measurements - depends if you power a solid state disk off the 40pin connector pin 20.

[Chr$]

A 7809 like the TIP112 circuit will need to dissipate a few watts - the current limit for the darlington transistor is specified as 2A, so I am assuming the 9V supply uses less than that - probably closer to 1A - so with 3 volts dropped - that's 3 watts as heat.

An LDO regulator wouldn't provide any real benefit - but maybe a switch mode supply that regulates 12V to 9V might be more efficient, but if you've got the metal work and are not worried about a bit of wasted heat, then the regulator is good option - plus the tab is connected to the common terminal pin2 - which would be ground. Solder small value capacitors between common and input and common and output - close to the regulator - helps prevent oscillations - 1uF to 10uF would be fine. It probably mentions this in the datasheet for the regulator anyway. a 7809 would be rated at 1A, so may be close the limit on current - best to watch out for that. I think the qubide boards take no more than 100-200ma from my measurements - depends if you power a solid state disk off the 40pin connector pin 20.

I found and ordered a max output 1.5A rated 7809. The metal plate is massive so that'll be fine for heat dissipation. The HDD is an old school whirly one with separate power from the PSU. I'll add the caps when it comes and will report back here.

Thanks for now.

[Pr0f]

All the best getting it working