Please help me understand\fix this discrete linear power supply

[stj]

Re: Please help me understand\fix this discrete linear power supply

maybe the pot's dirty or bad in some way.

[Agent24]

Re: Please help me understand\fix this discrete linear power supply

maybe the pot's dirty or bad in some way.

Possibly, it's about the only thing I haven't replaced or checked.

And I just realised that the lack of -7v rail which I disconnected today does not explain why it didn't work with the new opamp yesterday, unless the new opamp was DOA perhaps.

Guess I'll take a break until I get new opamps and hopefully if I've done something else really stupid I'll notice it before then!

[budm]

Re: Please help me understand\fix this discrete linear power supply

When you use the 5K pot to get the power supply to put out 5V with 15 Ohms load, you might as well take the Vdrop on that 66 (or 33) Ohms resistor so you will have the ref. to know what it should have for the circuit to work.

[budm]

Re: Please help me understand\fix this discrete linear power supply

BTW, you should have some load (I.E. 100 Ohms) on the output to help discharge the out put caps so when you make the adjustment up and down it will show the new Voltage level faster, right now it has about 45K Load on the output.

[Agent24]

Re: Please help me understand\fix this discrete linear power supply

Finally figured it out. Nothing wrong with the opamp.

Turns out Analog Ground and Digital Ground are supposed to be tied together - but this is not so on the analog board! They are in fact connected at the digital board (verified with ohmmeter) but which I have had unplugged the whole time... most annoyingly, this does not show up in the schematic at all!!

So I connected the two grounds on the analog board, powered it up, and voilà: +5v rail now works just as expected.

Now, if I'm lucky, the original dodgy TIP32 will have been the source of all the problems with the thing.... guess we'll see....

[budm]

Re: Please help me understand\fix this discrete linear power supply

Well, I am glad you got figured out.

[Agent24]

Re: Please help me understand\fix this discrete linear power supply

Well, I am glad you got figured out.

So am I, even if I do feel like an idiot for not figuring it out earlier!

Thanks for all your help with it, I've learned quite a few new things on this one

[stj]

Re: Please help me understand\fix this discrete linear power supply

I've learned quite a few new things on this one

like how much you hate Sinclair ???

[Agent24]

Re: Please help me understand\fix this discrete linear power supply

like how much you hate Sinclair ???

The PCB layout and component placement of this thing certainly leaves a lot to be desired, as does the service manual....


Anyway, thinking about it now, it seems obvious what must have happened:

The original TIP32A (Q302) must have been on the way out, hadn't gone completely open circuit, but was bad enough that the opamp had to turn on Q303 harder to keep the +5v rail up. This meant that Q303 was drawing more current from Q302's base than R303 could handle, so it overheated and the paint burnt off. The supply still 'worked' since R303 was not actually burnt out. How long the thing operated like that for, I have no idea.

After replacing R303 and testing the supply with my dummy load resistor, I probably drew more current than the digital board ever did, which killed off the already sick TIP32A, making it completely open circuit.

I replaced Q302 and fixed the original problem, but now without the digital board connected and the grounds tied together, it didn't work even though nothing was faulty, which of course made no sense.


It was only until I was thinking how I had had to measure half the opamp pins with respect to the other ground connection that maybe something funny was going on and confusing the opamp. Then I discovered that the grounds are actually connected together via the digital board, even though I can find no sign of that anywhere on the schematic!

[Agent24]

Re: Please help me understand\fix this discrete linear power supply

Back on this thing again, making some progress, but now have another confusing power supply issue with the secondary +5VA rail for the DAC\ADC chip.

Since last time, I got the digital section working by replacing U402, a 74HC86 quad 2-input XOR gate. This IC along with several others helps control the chip select lines for RAM, ROM, VIA etc and it was not functioning normally at all. One section was operating more like an OR gate, and the other 3 were just spontaneously oscillating etc (easy to see once I checked it with a logic analyzer). After I replaced that IC the digital section now works and I get the correct output from the composite connector to my oscilloscope (I haven't tested much else yet)


A problem I have noticed is that from first switch on, it does not output a signal. If I turn it off and then straight back on, it works. I have traced this problem to the +5v Analog supply which gives power to the UVC3120 DAC\ADC, An TL04C Opamp, a DS75107 line receiver, and a few other parts which mostly help provide the analog output to an oscilloscope (except for the DS75107 which does something on the digital side). (These parts also operate from the main -5V rail too, which runs from a 7905 if you remember)

The rail in question feeds from the main +7v rail and uses a 78L05 to generate the +5vA rail. So far so good, but it doesn't always come up. When it works, it's perfectly stable at 5v, when it doesn't work, it strangely shows -0.77 volts. (I would have expected zero).

What I tried so far:
1) I tried replacing the regulator with no change in symptoms.
2) I added ceramic decoupling capacitors soldered directly to the regulator's leads and this also did not help. (0.33uF input, 0.1uF output as per datasheet)
3) I decided to measure input and output current for the regulator to see if it was overloaded and hitting current limit or something. When working, it draws and supplies 150mA, and when not working, it draws and supplies 0mA.
This is interesting since the 78L05 in a TO92 package (what's used here) is rated only for 100mA.

I thought I had found the problem, and replaced it with a LF50CV regulator which can supply 500mA, to find I still have the exact same issue! (during this change I also changed the input capacitor to 0.1uF ceramic, and gave it an 10uF tantalum output capacitor)

I have heard that LDOs are inherently more 'unstable' than standard linear regulators but this seems like something else is going on here, especially considering when the +5vA rail is down, it is always negative 0.77 volts and not just 0v.

I have tried removing the UVC3120 from its socket and can repeat the problem, so I assume at least it is not causing the issue.

Any ideas before I just shotgun all the silicon that runs from that +5vA rail? I'm thinking maybe something is partially shorted or has some weird latchup(down? ) problem with regard to the -5v Rail, causing the +5vA rail to be dragged down below ground?

[stj]

Re: Please help me understand\fix this discrete linear power supply

that -v on the regulator output rings a bell.

i think you need a bypass diode.
example on page4

[budm]

Re: Please help me understand\fix this discrete linear power supply

That is the protection diode incase the input terminal is shorted to Ground it provides the output cap discard path.. it is recommended for many regulator ICs such as 78xx, 79xx series and other as well.

[stj]

Re: Please help me understand\fix this discrete linear power supply

yes, but it also stops the regulation getting messed up if there is a voltage on the output from another source when the regulator starts up.

i have seen a lot of circuits using +/- 78 & 79 regulator combinations that had similar issues wihout the diodes.

[Agent24]

Re: Please help me understand\fix this discrete linear power supply

Interesting idea but it doesn't seem to make sense or surely they would have caught the problem in R&D. It seems more likely that something else has gone faulty.

I do notice that they do have a diode across the regulator's output to ground though (reverse biased). It tests OK with diode check. I will remove and check it, maybe it's open circuit.

[stj]

Re: Please help me understand\fix this discrete linear power supply

maybe the internal difference between different makes of regulator matters.

i'v seen similar issues with old (fast) 74'series digital logic that you had to use TI chips because other makes couldnt run as fast.

[CapLeaker]

Re: Please help me understand\fix this discrete linear power supply

o.k. what does the input voltage to the 5v regulator do in that weird state?

[ben7]

Re: Please help me understand\fix this discrete linear power supply

Hm, it probably goes negative because there is a negative voltage rail that is feeding the loads as well, the ~0.7V appears to me like a single diode drop.

I do remember having an issue once with a 7805 regulator not regulating the output, it would droop (and oscillate) with even a minimal load (milliamps).
I seem to remember using different capacitors made it stable again.

It's possible that inductance, etc... could be causing it to be unstable.

I know, a switching vreg can be stable with a few inches of wire length to the power source, but then add a few more inches and it goes unstable ... (inductance causing this here..)

Also, since you replaced the 78L05/7805 regulator, and it still does the same thing, why not just try replacing the negative rail regulator, just for the hell of it?

Yes it might be possible that something is shorting out, and the -V rail regulator has a higher overcurrent threshold, so the +V rail reg ends up tripping on overcurrent first.

Also maybe try getting oscilloscope waveforms of the power being turned on? Might reveal some "funny business" that could be happening.

-Ben

[Agent24]

Re: Please help me understand\fix this discrete linear power supply

maybe the internal difference between different makes of regulator matters.

I did think of that; the original 78L05 is a National part, the replacement 78L05 an ST, but both had the same behavior so it doesn't seem very likely. Also if it were that critical, surely the LF50CV would be likely to show different symptoms?

o.k. what does the input voltage to the 5v regulator do in that weird state?

It stays at +7v the whole time as far as I can tell with my DMM.

Hm, it probably goes negative because there is a negative voltage rail that is feeding the loads as well, the ~0.7V appears to me like a single diode drop.

I do remember having an issue once with a 7805 regulator not regulating the output, it would droop (and oscillate) with even a minimal load (milliamps).
I seem to remember using different capacitors made it stable again.

It's possible that inductance, etc... could be causing it to be unstable.

I know, a switching vreg can be stable with a few inches of wire length to the power source, but then add a few more inches and it goes unstable ... (inductance causing this here..)

The original design had no decoupling capacitors close to the regulator at all, I added some to see if it made a difference and it didn't. But as far as I know, unstable regulators oscillate, right? This one just seems to shut down completely.

Also, since you replaced the 78L05/7805 regulator, and it still does the same thing, why not just try replacing the negative rail regulator, just for the hell of it?

Good idea, I will try that at some point. Not sure I have any 7905s at the moment though.

Yes it might be possible that something is shorting out, and the -V rail regulator has a higher overcurrent threshold, so the +V rail reg ends up tripping on overcurrent first.

I was wondering though, I thought these regulators just current limited into a short at low voltage? But when the fault condition is present, the +5vA regulator draws no current. I don't think the datasheet describes such a condition (or I didn't see it). I might try seeing if I can replicate the problem on a breadboard...

I also did check the -5v rail, it stays stable at -5.04v even while the +5vA rail is -0.77v.

Also maybe try getting oscilloscope waveforms of the power being turned on? Might reveal some "funny business" that could be happening.

Would love to check the power on waveforms but I have nothing to capture it with.

[CapLeaker]

Re: Please help me understand\fix this discrete linear power supply

and what happens if you use one of these small cheap buck converters instead the linear one?

[Agent24]

Re: Please help me understand\fix this discrete linear power supply

Decided to have some more fun with this today.

Started pulling all the parts that sit between the 5vA and -5v rails. I found that removing the TL074 Op-Amp, IC427 makes the problem vanish - the 5vA rail comes up fine first time, every time.

So it seems that on a cold power up, IC427 is going into some kind of latch-up condition, shorting out the 5vA LDO which shuts down - if I turn the unit off and quickly back on, everything is fine.

Naturally I replaced IC427 with a new TL074, and naturally, the problem is still there!
I also tried what ben7 suggested and replaced the 7905, but still nothing changed.

Does anyone know what could cause the Op-Amp to go into this apparent latch-up state? Or is it even latch-up at all? Or something else?

EDIT: As a little experiment, I took the old Op-Amp and bent all the input and output leads out and inserted it into the socket I installed with only the supply pins connected. The 5vA supply did not start - so unless the floating inputs cause a problem, perhaps it is not any issue with the inputs coming up before the supply or anything like that. Anyway, I then removed the IC with power applied and saw the 5vA rail come up as soon as I had done so.