Hello guys

This is my first post on this forum

I got an idea of making a dual power supply from a single power supply without using a centre tapped transformer (my c.t trans should be arriving today but while I was waiting thought i'd experiment)

Basically the design revolves around using a potential divider to reference a ground halfway between the + voltage and the actual ground of the single power supply.
If 2 resistors (for example 1k and 1k) are used , then when a current is pulled across one of them then there is a voltage drop and the 'ground' shifts which is impractical.
Now what if we put a pair of transistors instead of the resistors and an op amp which references the voltage divider ground to the new 'ground' and corrects the transistors so that the new ground stays at the voltage divider ground.
I've attached a link to the design that I made in circuitlab.
Well... it works when I try pulling current by putting a low resistor (200Ω) across the new 'ground' and say the Positive rail then there is a very slight shift (a few tens of milivolts) in ground corresponding to half of the voltage droop across the positive and negative rail of the power supply which is completely logical.
The problem is if I try to use this supply with an op amp to drive a speaker or something (at about 1khz ), then the ground goes haywire and just can't keep up which is frustrating because the design looks pretty legit to me (i'm still an amateur) and it is as if the op amp can't keep up which doesnt make sense to me . I mean i'd understand if it was at a few hundred khz or at Mhz , but at a few khz it should work fine no?
Any advice/comment/critisimn or improvment that I can make on the design?
I'd well appreciate it

Thanks

https://www.circuitlab.com/circuit/7qnzpt/single-to-dual-power-supply/
The is also a png of it in the attachments

 

I would use a better ic for the split design like a TDA2030, this will hold the ground stable at higher currents.

http://www.seekic.com/circuit_diagr...ngle_dual_power_supply_converter_circuit.html

 

damn and here I was thinking that i'd be patenting a new IC with my revolutionary design D
Thanks , I'll check it out the TDA2030
But any advice on why my actuall design isn't working as it should when driving an ac load?

 

Which op-amp is that? I'm wondering if it's fast enough for audio. It probably is. How about those transistors?

I also wonder if you are maybe getting some oscillation on the approach to steady state. Can you challenge your circuit with a step voltage change and see how fast it recovers?

 

I'm using an LM358N
and the transistors are
BC546B NPN
BC556A PNP

I was also thinking that the circuit might be oscillating when approaching steady state but I dismissed it because there are no capacitors in the circuit and the junction capacitance of signal transistors comes into play in the hundreds of khz range from what I read.
How would I go about challenging it with a step voltage change?


My opinion is that it is just poorly designed, I was thinking of replacing the pnp with an identical npn and using the LM358 as a dual with 1 op amp for each transistor . Also I was thinking of experimenting with mosfets instead of Bipolars seeing as they do not take input current so the op amp draws less which keeps the voltage divider at a steadier point.

What do you think?
Thanks for the reply

 

damn and here I was thinking that i'd be patenting a new IC with my revolutionary design D
Thanks , I'll check it out the TDA2030
But any advice on why my actuall design isn't working as it should when driving an ac load?

Your resistor values are wrong.
Try the changes shown in the attachment.
Be aware that the circuit can only handle about ±100mA, and this needs to be AC, to avoid overheating of the transistors.
The diodes guarantee class AB operation as temperature and battery voltage change. It would be best if D1/Q1 and D2/Q2 were thermally coupled, but I think R2 and R4 will provide adequate negative feedback to ensure thermal stability.

 

The LM358N is pretty slow for an audio application.

 

Thank youuuu Ron H
I'll build and test the improvement first thing tomorrow morning.

''The LM358N is pretty slow for an audio application.''

Really?
I've got another 10 diffrent op amps so i'll try them out and post result tomorrow.
I thought even general pupose op amps could handle much higher than audio frequencies

Thanks for the reply

 

Take a look at the data sheet. At higher audio frequencies above 5k or so, it's falling off quite a bit. Would be fine for low-fi.

I'm not certain this is the cause of your problem, I'm just sayin'.

 

The LM358N is pretty slow for an audio application.

In simulation, the output impedance at 20kHz with a TL081 was about 0.1Ω. With the LM358, it was about 1.3Ω.
The circuit has a little crossover distortion with either op amp, so it might still have some irritating harmonics. The best fix for that is an op amp with higher gain-bandwidth product.

 

Why not add some large capacitors from each rail to the pseudo-ground to decouple any higher frequency AC currents?

 

Why not add some large capacitors from each rail to the pseudo-ground to decouple any higher frequency AC currents?

Hmm not sure how that works, in fact, I happened to come across this suggestion in another forum earlier on and someone explained that it wouldn't work .I'll try it tomorrow anyway along with the rest of the suggestions in this thread.

 

Take a look at the data sheet. At higher audio frequencies above 5k or so, it's falling off quite a bit. Would be fine for low-fi.

I'm not certain this is the cause of your problem, I'm just sayin'.

I was testing it with a 1khz sine wave but nevertheless thanks for the chart , useful info

 

That chart is with gain of 100, with lower gain it should not be as tragic (but I din´t actuall check the datasheet, so it may be because of slew rate and not beacuse of gain badwidth product).

 

Your resistor values are wrong.
Try the changes shown in the attachment.
Be aware that the circuit can only handle about ±100mA, and this needs to be AC, to avoid overheating of the transistors.
The diodes guarantee class AB operation as temperature and battery voltage change. It would be best if D1/Q1 and D2/Q2 were thermally coupled, but I think R2 and R4 will provide adequate negative feedback to ensure thermal stability.

So i'm looking at the design and i'd appreciate if you took a few minutes to explain some details.

The diodes put the transistors in AB push and pull , I can understand that . - - Why are the bias resistors so low (1k) . Won't that turn on both transistors on hard and dissipate alot of power even when not operational?
- What are the R2 and R4 10 ohm resistors there for?
- What is the purpose for the 100nf cap across the op amp supply (To provide current bursts at higher frequencies to increase high frequency response? )

I notice you have replaced the op amp with a JFET op amp (Forgot those exist and that I have one )
Also I notice that there is no feedback on the Op amp from the output stage.

Thanks for your improvement and thanks in advance for the reply

 

Also , If I double up both transistors in parallel with identical ones will I be increasing by current handling capabilities by 2?

 

Also , If I double up both transistors in parallel with identical ones will I be increasing by current handling capabilities by 2?

That should work if you include a 10Ω emitter resistor with each transistor.
If you actually need to handle that much current, a standard 9V battery won't last long.

 

That should work if you include a 10Ω emitter resistor with each transistor.
If you actually need to handle that much current, a standard 9V battery won't last long.

Yes thanks for the info , I'm using an adjustable power supply with an LM317 so i't s all cool