Stupid question on LM7812

[mozwell]

Dear fellow MEF's
I have been using the LM7812 (etc) for about 30 years and never come across this problem.
The LM7812 IC2 refuses to start and has about -0.9V on its output, with all other regulators starting without a problem. (C7 and D3 form part of a tripler for a phantom 48V supply)


The load is op amp loads of 2x TL072 & 1x NE5532 with a two pairs of 0.1u bypass caps from pos to 0 & from 0 to neg.
I need the 2200u on the supply as i am running a 12AX7 heater from a separate LM7812 so i want to minimize supply ripple.
Often the LM7812 IC2 refuses to start & its output stubbornly stays at -0.9V this may happen 10 times in a row so its something i need to fix.
The LM7912IC3 regulator doesn't have this problem. The LM7812 IC1 for the 12AX7 heater (with no output bypass cap) and it starts fine every time.
It looks to me its the capacitance on the positive supply rail being the difference between the two LM7812 regulators, but the negative rail has the same capacitance as the positive rail...

I am assuming i may have one of several issues. The capacitance on the LM7812 output causes a brief short circuit as it starts to charge & the LM7812 goes into shutdown & stays there. Maybe the half wave supply causes the main filter caps to come up slowly causing the LM7812 to be confused & go into a protection mode.


My solutions are...
1. use a 220R on the output of each regulator & bypass it with a relay after a second or so. It starts every time.
2. place a 2k2 shunt resistor from the input of the LM7812 IC2 to its output. It starts every time.

questions.
A. has anyone else seen this
B. are my solutions reasonable
C. have i done something so stupid i cant even see my problem

Thanks in advance & have a merry christmas
Mozwell

[loudthud]

You need diodes reverse biased from +12 to ground and from -12 to ground. The load from +12 to -12 will cause one of those outputs to go past ground on startup due to the way the the AC is rectified.

[mozwell]

Doh...
I will try this tomorrow night
Thanks for your suggestion

[GainFreak]

Just try this:

[mozwell]

Loudthud
those diodes worked a treat
many thanks

have a great xmas & avoid covidiots

[loudthud]

Some people think these three terminal regulators are bullet proof, but they are far from it. If you short +12 to -12, one regulator usually wins and the other blows without the diodes. Another failure mode is a big cap on the output holds the output high while the input side collapses when power is turned off. Another diode will fix that. The thermal shutdown won't save you if the heatsink is too small, the regulator will eventually fail.