How to derive split supply from a single rail PSU?

['LIVEWIRE?']

The PSU in question being one I built many years ago from a PW design.
It's a variable voltage regulated unit, using a 741 with series pass transistors, and can supply up to ca. 25v @ 2A Max. Needing a +9/0/-9v DC supply I thought 'simples!' Just connect two equal value electrolytic capacitors across the o/p and 'Bob's your Uncle'. However, what I'm getting is about 2.5v on the plus side and 16v on the -ve side, relative to chassis. The supply is 'floating' with respect to chassis, enanling to be used with equipment of either polarity, and the pass transistors, etc., are in the positive side, so why don't i get an equal positive & negative voltage with two Capacitors across it's o/p?

[KeithsTV]

Can you post the circuit or which PW was it from?

Keith

[llama]

I would expect the two equal electrolytics to do very little in the steady state. And they would have to be very precisely matched for the centre point to rise to mid-supply during switch-on. But I would think the major problem is difference in current drawn from the two rails by the load.

First, I would replace the pair of electrolytics by a pair of equal resistors - say 1k - and re-test. You might get lucky. If the mid point is still unbalanced I'd put another 741 or similar straddling the rails. Strap the inverting input to the output and take the non-inverting input to the centre-point of 2x 100k resistors which go from negative to positive rail. Decouple that if you're feeling keen. The new mid-point at the output of the new 741 should be solid enough.

Hope that is clear!
Graham

First, I would replace the pair of electrolytics by a pair of equal resistors

I would leave the caps in place and add a 9V zener diode across the capacitor with the higher voltage across it. To make it 'better' a couple of 9V zeners one across each cap would be better, don't turn up the PSU beyond the sum of the two zeners!

[orbanp1]

Originally Posted by llama

...
First, I would replace the pair of electrolytics by a pair of equal resistors - say 1k - and re-test. You might get lucky. If the mid point is still unbalanced I'd put another 741 or similar straddling the rails. Strap the inverting input to the output and take the non-inverting input to the centre-point of 2x 100k resistors which go from negative to positive rail. Decouple that if you're feeling keen. The new mid-point at the output of the new 741 should be solid enough.

A 741 can only sink or source about 10mA, depending on the manufacturer.
You better add a complementary emitter follower pair to the opamp output, just like a B-class audio amp output. There is no need for Vbe bias, and feed back to the opamp negative input the joint emitter output point.

Peter

[mark-jacobs]

You can also look at something like a DC Single Supply to Dual Supply Converter http://www.ebay.com/itm/DC-Single-Su...item20adcde9e5 .

[richrussell]

Yes, these days, I'd just use a dedicated single to dual supply converter ic too. There's no need to mess about with finely balanced opamps when you can just stick in an off the shelf module that'll do the job and have a level of overload protection built in.

Unless of course you want to do it with opamps or discrete components for your own satisfaction of course.

[short wave]

Have a look at this.... (may be worth reducing the 1Ks to a lowere value)

http://www.instructables.com/id/Spli...e-rail-supply/

regards S-W

['LIVEWIRE?']

Originally Posted by KeithsTV

Can you post the circuit or which PW was it from?

I think it was from the November 1972 issue of PW, Keith, but I can't find my copy. The circuit is based around a 741 op amp, with zener diode reference voltage set at 5v. The series pass transistor is a darlington pair using a BC108 or similar and a 2N3055. With an unstabilised i/p of around 28v, I can set the o/p to anywhere between 5 and 24v.

Originally Posted by short wave

Have a look at this.... (may be worth reducing the 1Ks to a lower value)

That's certainly easy enough to build!!

[Kala_12]

Hi
Is this any use?
Stuart.

['LIVEWIRE?']

I'd forgotten about that book, Stuart. I have a copy somewhere. All sorts of useful circuits in there. Incidentally the problem which prompted my original post turned out to be a fault in the CD player I was testing - A SONY which is normally powered from a 15vac supply derived from the associated receiver. The +ve DC rail is below 2v, whilst the -ve is at or above 12v!

[short wave]

Found a practical circuit for splitting a single rail PSU , I remembered I had seen is some where ! (Farnell LF1 , LFP1 , LFM4).
Looking top left .The “UA78MGUIC” is used as a 20 volt regulator , programmed by the R100 / 101 / 103 network. The series network also sets the base of VT 100 at just above “half volts” allowing for 0.6 volt drop (V, b-e) thus providing the half volts point.
As a further point you could “Darlington-a-rise” it if further increased current is required . The 2N3053 is limited to about 1 amp but dissipation values must be considered also.
Regards S-W

[cmjones01]

Beware that the Farnell circuit can only source current into the '0V' connection, so if you want to draw more current from the '+10V' rail than from the '-10V' rail, you'll find that the '0V' point starts to wander up towards the '+10V' rail and so the rails will be out of balance.

Chris