trouble with bipolar powersupply

[janoc]

11.5V from a 7812 off load is not fine, it really should be 12V.

For the life of me I cannot see how you're getting more than 30V input to either regulator if you're measuring from the centre tap of the transformer and yikes, mains on a breadboard?? 

There, fixed it for you. That is seriously scary stuff. Breadboards are definitelly not rated above 50V and many have metal plate backing - with the only thing insulating the contacts from it being paint and/or a thin layer of doublesided tape ...

[arjepsen]

Yeah, I'm aware that it's not the most safe setup - I am being reeeeeally carefull :-)

For the life of me I cannot see how you're getting more than 30V input to either regulator if you're measuring from the centre tap of the transformer and yikes, mains on a breadboard??

Power it all off and measure continuity from the centre tap to the ground pin of both regulators, measure continuity from centre tap to the shared connection of the main capacitors.

Are you sure you've got a connection between the secondaries, it looks like you've got the link between them in adjacent holes and both wires in one hole would be a tight fit.

Unloaded I get 25,6 volt AC between each side and CT. With nothing but the first pair of caps after the rectifier, I get 35,7 volt DC between each side and CT.
And yes, both wires for the CT are in the same hole.

Does your circuit have protection diodes from the output rails to
ground?
See Fig.16 from the attached Fairchild data sheet for a dual supply with output protection diodes.

Just tried with that and zeners. Still 11,6 on the 7812 and 17,5 on the 7912.
One question about the datasheet though. In firgure 16, the negative rail has bigger caps than the positive rail. Would that make any difference? (though of course I'm already using bigger caps than in that figure).

Another question: shouldn't it be enough load to put a 220k resistor from each output to ground?

[Zero999]

Why not get yourself some strip board?

Don't use it for mains. The safe way to do this to solder the neutral of the mains cable directly to the transformer's primary, connect phase conductor via a 100mA fuse in a suitable holder.

[tatus1969]

I wonder though - wouldn't a 2x12V trafo be to low on the output to supply the regulators with high enough voltage, when under load?

The transformer will deliver 2x12V AC at its rated (sinusodial) current. After the bridge rectifier, this will result in roughly +-17V DC (keep in mind that the capacitors charge to the PEAK voltage, whereas that 12VAC is an RMS value), minus one diode drop for each rail. Plus the current is not sinusodial but pulsed, but I would assume that you'll still get +-15V, which is enough for the 7812/7912.

As I explained, for such small transformers the secondary voltage can be way higher when there is no load, this is why you measure more than +-30V without load. You say you added zeners to both regulator inputs, so that should limit that voltage. Did you check their temperatures? Also, are your electrolytic capacitors new and fresh?

[tatus1969]

Another question: shouldn't it be enough load to put a 220k resistor from each output to ground?

220k is definitely not enough to provide a base load, the current would just be 12V/220k = 55 microamps. Try a lower resistor value, for a 0.25W leaded part -> 680R.

[tatus1969]

In firgure 16, the negative rail has bigger caps than the positive rail. Would that make any difference? (though of course I'm already using bigger caps than in that figure).

Should be okay as you do it.

+1 to lift off the primary from the breadboard and see if that makes a difference (plus avoiding possible real danger).

[arjepsen]

Man, this is getting weird.

First of all, yes the caps are new. Tried different ones - same result.
And I did check very carefully, to make sure the mains leads arent having a continuity to anything.

Anyways, I just put a couple of 470R 1W resistors from each output to ground.
Now the 7912 shows a nice -11.8  Happy happy.
However, now the 7812 only shows 0.7   
The resistor across the 7912 does get a bit warm, whereas the one across the 7812 stays cool.

EDIT:
Just changed the 7812 again, and it turns out that was a bad one. Now the 7912 is around 11,8 and the 7812 is around 11,6 under load.
However unloaded they are right back at -17,8 and +11,7

[tatus1969]

congrats, looks like you have won this battle! just one last question: what specific type are your output caps? many medieval linear regulators have problems with low ESR types and tend to overshoot. maybe that is what you are experiencing, you would be able to verify that with an oscilloscope. but maybe it is bad chinese replica that is causing your priblems. anyway, minimum load is a must, maybe you turn these into some nice indicator LEDs

[arjepsen]

Caps are panasonic fc's. They should be fine, right?

[tatus1969]

Caps are panasonic fc's. They should be fine, right?

hard to say. the 100uf 50v part has only 0.162 ohms. the 78/79 family was not designed for such low ESR caps back then, simply because they didn't exist. you can try adding 1 ohm in series with them and check if it makes a difference. modern LDOs mostly specify the ESR range that is allowed, the better one are "ceramic stable" meaning the ESR can be very low.