PSU startup problem and how to fix it

There are lots of designs for synth power supplies around. Many of them are linear supplies in which there is a transformer — sometimes in the form of what’s affectionately known as a wall wart, sometimes built into the supply — that reduces the wall voltage to 12 VAC, which then is rectified with some diodes and filtered with some big capacitors. Then a couple of voltage regulators convert that to steady and accurate +12 V and -12 V outputs.

(Or ±15 V, same idea.)

But there’s a known quirk of positive linear voltage regulators, and not all power supplies take it into account. Both the MFOS Wall Wart Supply and the Frequency Central FC Power , for instance. With a supply like one of these, sometimes when you switch it on, the positive regulator will fail to start up, so you get nothing on the +12 V output. There’s a workaround: If you switch the PSU off and back on again, often both regulators will start up and you’ll have both outputs. That’s a nuisance, though.

The issue

The part of the circuit around the positive regulator looks something like this:

The LM7812 is the regulator. Unregulated voltage, around +15 V (probably higher, but let’s say 15), goes in on pin 1 and regulated 12 V is produced on pin 3. The resistor marked “LOAD” represents the modules being powered. The two 1N4004 rectifier diodes are there to protect the regulator against capacitor discharge if the input or output is shorted. Actually, according to Texas Instruments Application Report AN-182 , the diode from output (pin 3) to pin 2 is to protect against discharge from a capacitance between pin 2 and ground. But there is no capacitor there in this circuit. Granted, there will always be parasitic capacitance, but would that be enough to endanger the regulator? I wouldn’t think so, and so I’d think that diode is superfluous here, but I could be wrong — it’s not doing any harm, anyway.

The negative regulator is connected similarly, so we have this:

The trouble occurs if the negative regulator gets going first. Then the output of the positive regulator, which in the absence of the load would be floating, gets pulled to a negative voltage. According to AN-182, if the output pin is pulled low enough below ground, the positive regulator will not start up.

Note that’s just the positive regulator. Negative regulators have different circuitry, and will start up even if their output is pulled positive.

The fix

AN-182 goes on to say, “Clamping the output to ground with a germanium or Schottky diode usually solves this problem.” The point is that the diode from the positive regulator output to ground will prevent the pin from being pulled negative by more than the diode’s forward voltage, but that’s about 700 mV for a 1N4004 rectifier, and that’s too large. Germanium and Schottky diodes tend to have lower forward voltage, usually around 200 to 300 mV.

So a better circuit for the regulators is this. The 1N4004 from +12 V to ground has been replaced by a 1N5817 Schottky:

Again, this issue doesn’t arise with the negative regulator, so the rectifier diode is fine — and as noted above, no diode there at all may also be fine, but either a 1N4004 or a 1N5817 does no harm. If you’re building a new PSU, use either; if you’re retrofitting an old one, leave whatever is there in place.

On the FC Power board, the diode in question is this one:

The assembly instructions say to use a 1N4004 rectifier diode there, but replacing it with a 1N5817 Schottky will eliminate the possibility of startup problems. Likewise with the MFOS Wall Wart supply, the diode to replace is this one: