I did not meant to make another general: make my ADA8000 not fry itself. I could just not find any discussion on using +/-18V rails, as these would move some heat away from the regulators, and the caps which are in direct proximity, as these both are the parts that fail. I am quite aware of the other methods discussed to get the unit run cooler: Resistors, relocating regulators to a heatsink/chassis, fans, other transformers (and this being the optimum solution). Simply swapping regulators would in my opinion seem like the easiest way to keep the psu from dying, as they cost next to nothing and you have some lying around anyways, no? Yes your op amps will die sooner, but do they die within a relevant time frame?
JohnRoberts said:
In my judgement a bad idea to ever run parts at maximum rating.
Andy Peters said:
Since ± 18 V is right at the op-amp process limit, you need to verify that the regulator outputs are really 18.0 V and not higher.
Sure, could be brought out of the danger zone with a series diode. TL07x are working in quite a few places at 17v+ reliably. Does not make it the best choice, but seems to work fine.
Andy Peters said:
You have to make sure that your worst-case (low line) regulator input voltage doesn't go below the specified drop-out. For an 18 V regulator, you want a 21 V input (and similar for the negative rail).
That's quite obvious, but measurements should be taken. Anyways, it does not seem to be a problem here.
JohnRoberts said:
figure 12 in TI data sheet
Thanks, was only looking at the ST data sheet, as these are STs, but seems to be quite linear.
JohnRoberts said:
#1 use half wave rectification- this will double the PS ripple voltage but drop the dissipation in the regulator some (make sure extra ripple doesn't cause regulator to drop out).
#3 use smaller reservoir caps in PS. This will increase ripple voltage but drop average voltage and dissipation in regulators (again make sure ripple voltage is tolerated by regulator low voltagte drop-out). .
This is quite an interesting idea and I took some measurements on mine and decided to give half wave rectification a try for the 5V rail.
So: Desolder both middle diodes of the 5V rail rectifier (D9, D10) and add a 220µF capacitor in the now free holes for "safety". Scope shows minimum of 8,5V + ripple at 231V mains voltage. Seems pretty good, as it leaves enough voltage for -10% drop in mains voltage, the additional 220uF capacitor could be even smaller. Current consumption on the 5V rails should be quite constant as well. And of course: they run a bit cooler, but not something to really rave about for IC1 and IC2, but IC7 dropped its temperature significantly.
Andy Peters said:
And, as noted, you're simply moving the power dissipation from the regulators to the op-amps.
Yes, as noted above it will likely shorten the op amps lifespan, but how long is it and how much does it shorten it? Subjectively they are not near dying from over temperature.
Ok, the biggest problem I see here is frying your input OPs with your transistor preamp when running it with +/-18V as they only tolerate +/-15V on the inputs, regardless of the supply voltage being higher. This goes for the TL074 at least. So not a good idea.
I don't see any problems with the output stages regulators, even the OPs run cooler than these in the input stage. But: The regulators themselves are not even that hot... so possibly a waste of time to swap them - will think about it.
-Arne