µA7812 latch-up

[benb]

Despite the necrothreadism, I'm glad I read this. I've been playing with designing and making synthesizers, and everyone's into Eurorack which runs on + and -12V (and some modules need +5V).  A common power supply (to run only a few modules) is a single-seondary 12V to 16V AC output wallwart into two half-wave rectifiers and big filter caps, and then the usual 7812 and 7912 regulators for the final voltages. I'd like to have a 7812 that does this kind of latchup (I presume it's caused by the OUTPUT voltage being pulled negative) so I could tell if a Shottky rectifier fixes it. I'll have to make up a circuit that does this to test the regulators I have.

The devil's in the details, and there are so many details...

 

[JohnRoberts]

The devil's in the details, and there are so many details...

Why bother with shottky ?  A popular mod back a few decades ago when experiencing latch up was to put a silicon diode in series with the ground lead...  This gives you another 1/2V or so of output, but somewhat reduces regulation performance.

I expect modern regulators don't latch up...  I haven't encountered any this century and even longer ago. I've used truck loads of 78xx and 79xx regualtors over the decades, but lately more like LDO 3.3V regulators .  8)

JR

 

[gyraf]

A reversed 1N5819 schottky diode across output very often helps overcome this problem.

But I still avoid the  "TS" type positive regulators..

Jakob E.

 

[RuudNL]

Although (as mentioned earlier in this thread) it may not work in all cases, I always could solve this problem with an anti-parallel 1N4007 across the output.

 

[Harpo]

Although (as mentioned earlier in this thread) it may not work in all cases, I always could solve this problem with an anti-parallel 1N4007 across the output.

Same here.
The datasheet application information for ua7800 series already says so...  "In many cases, a regularor powers a load that is not connected to ground but, instead, is connected to a voltage source of opposite polarity (e.g., operational amplifiers, level-shifting circuits, etc.). In these cases, a clamp diode should be connected  to the regulator output as shown. This protects the regulator from polarity reversals during startup and short-circuit operation."

 

[saint gillis]

Here's a case where that reverse diode didn't help.
I didn't try the diode in series with the ground pin of the 78XX though...
https://groupdiy.com/index.php?topic=72272

 

[Harpo]

Here's a case where that reverse diode didn't help.
I didn't try the diode in series with the ground pin of the 78XX though...
https://groupdiy.com/index.php?topic=72272

Schematic of your Hartge amp (pg.3) doesn't show anti-latchup-diodes, only diodes across the regulators i/o-pins.
From regulators app.note you'd want a diode between 0V/+15V and between -15V/0V.

 

[saint gillis]

Yes but I added it, and it didn't change the latch up problem

 

[Monte McGuire]

I've experienced that sort of latch-up problem with modern regulators that use MOS output pass transistors (the ADP7142 and ADP7182). The problem happens when one of the regulators starts up before the other and reverse biases the unstarted regulator's output, turning on the MOS output device's parasitic body diode, thus preventing the regulator from starting.

The answer is in the datasheet, but it might not be obvious. The datasheet correctly states that the positive regulator's output cannot be pulled to less than -0.3V, and the negative regulator's output cannot be pulled up to 0.3V or higher. So, the solution was to add schottky diodes to the regulator outputs to prevent them from getting reverse biased. The diodes have a sufficiently low voltage drop when conducting the rated regulator output current, so it seems to be a fail-safe solution. Since adding those diodes (Toshiba CCS15F40,L3F), I have had zero startup problems with these regulators.