PSU for a chip amp/DC servos etc

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Simple question, but not found anything answering it so far...

I've got a split rail PSU which outputs, say, +/- 50V. It's past the range of input for a normal linear regulator, but I want to run some op-amps in the same box for DC servos, and probably speaker protection etc.

So, would a simple zener/emitter follower be good enough for running the op-amps? I can only envisage 10s of milliamps being drawn from these rails, after startup. Obviously though, I want a nice "clean" PSU for if I decide to put tone controls in, for instance.

So, in the end, I'd have +/- 50V, and a +/-15V derived from that. Any ideas welcome.

Ok, so with regards to the linear regulator, my concern would be the main supply rail dragging the LMxxx to it's output hard against the differential voltage limit. I've seen some LM317AHVTs that allow 60V input / 57V differential, but after finding some negative opinions on the LM series, I want something more flexible.

So, zener / transistor... I simulated this and some other basic circuits on LTSpice last night, and was pleasantly surprised to get 15V/26mV ripple @ 5A load from an op-amp feedback solution and a 2n2222/2n3055 darlington arrangement, running from a 30V supply with 5Vpp 50Hz "ripple".

As the PSU op-amp only needs to track up to it's output supply, this means I could most likely get away with operating the op-amps from a quick and dirty zener supply at ~30V.

This would allow input voltages as high as +/- 75V, but then if I'm drawing 500mA, that's (75-15)*0.5 = 30W to dissipate - per rail. I suppose that's the price you pay for not wanting SMPSU noise in the amp box though...
Kicking myself here - I suppose the easier solution is to have a second transformer / multi tapped secondaries for this sort of thing - as long as the op-amp inputs don't exceed supply rails of course. 30VA toroidals on farnell for ~£15+vat.

Edit: Yes, you're right Arty, I think that's one of the reasons people DIY - finding new and interesting solutions to these problems! /Edit
Last edited:
allso, you can divide one rail, to get a dual supply for opamps.
like.. power amp requires +/- 30 rails then one of these rails canbe split to somethigng close to +/- 15 opamp powersupply. if its a pre-amp, and output voltage is only a few volts then there is no need to go verry high on opamp powersupply rails. just to name a nother drity solution that is perfectly working and cost saving. since multi-secundary transformers have a higher price.
Well, this is what I came up with. Simulated, designed and built this PSU. It's got +/-45V main rails, and uses a zener regulator + op-amp to drive IRFP240/9240s to get the +/-15V for the DC servos.

Simulation predicted 20mV ripple; I've yet to measure it though. Voltages check out ok on both sides under a quick test but there need to be some mods before I'm 100% happy. Still, it works :)


  • AMP_Psu.jpg
    312.9 KB · Views: 87
Use a Zener and/or resistor to feed the 50Vdc to the 317 IN.
Add a Zener from OUT to IN to limit the dropping voltage across the 317.

Suppose the 50Vdc rises to 52Vdc at worst case conditions.
A 10V Zener will reduce this to 42Vdc. Add a 510r resistor in series. At 10mA (the recommended minimum for some 317) drops another 5.1V to leave a worst case 36.9Vdc at the input to the 317.
Add a 36V Zener across the 317 in case you somehow manage to short the OUT to Ground.

If the PSU drops to 48Vdc and the current increases to 30mA then the 317 IN voltage becomes 48 - 10.1V (Zener) - 15.3V (510r) = 22.6Vdc. That leaves 7.6V across the 317 and it dissipates ~300mW (no heatsink required). The Zener dissipates <300mW. The 510r resistor dissipates <460mW, a pair of 1k0//1k0 will run cooler if 30mA is a regular occurrence.
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.