Dual polarity power supply with a single secondary transformer?

[tuhkam]

Hi,
I was wondering, what would be the possible problems with this design?

My transformer only has a single secondary winding of 42V. I'm intending to connect it to a filter board, that has AC-GND-AC on the input side. I tested it without connecting the filterboard GND to EARTH (blue line), it gave a split voltage of +/-30V on the filterboard output.
It should power two LM3886 amplifier boards as the end result.

[DF96]

No, unless you can guarantee that under all conditions (including faults) the circuit draws exactly the same current from + and - rails. Even then you would need equalisation resistors across the caps.

[tuhkam]

Thanks for the reply!
I'm afraid it raises even some more questions.
The filter section is actually a bit more advanced, than on the schematic. It is one of those kits that has a multilayer board and comes without a proper shematic (only layout picture). There are two resistors on it, but it's hard to tell if they are for stabilization or for leds (or both?). It came together with the LM3886 boards, so there is a chance it will work for them, but i'm not sure it will work with a single secondary (no center tap) tranformer. See below pictures!




I could be getting it the wrong way, but I was thinking that since the two LM3886 boards (each with one IC) should be running in phase, they could never draw the same current from + and - at the same time. As audio is AC it should be on + or - side, not both at the same time.
Would a center tapped or dual secondary transformer be way better for the application, why so?

[kubeek]

You have to either get a center-tapped secondary, or don´t use the cap-midpoint at all - i.e. use bridged amps, and make another reference ground. I am not sure if you still need the equalising resitors, but I am sure they wouldn´t hurt

[AndrewT]

if you must have a dual polarity supply then you can use half wave rectifying for the two polarities.
Ripple will be roughly doubled.
Almost certainly need to fit rCLC filtering to get decent DC from the supply.
Cheaper to buy a dual or centre tapped secondary.

[macboy]

Originally Posted by AndrewT

if you must have a dual polarity supply then you can use half wave rectifying for the two polarities.
Ripple will be roughly doubled.
Almost certainly need to fit rCLC filtering to get decent DC from the supply.
Cheaper to buy a dual or centre tapped secondary.

Consider what the rail voltages would be if you did that with a transformer having a 42 VAC secondary... around +/- 60 VDC. That's far too high for any chip amp I have ever seen.

[macboy]

Originally Posted by tuhkam

Thanks for the reply!
I'm afraid it raises even some more questions.
The filter section is actually a bit more advanced, than on the schematic. It is one of those kits that has a multilayer board and comes without a proper shematic (only layout picture). There are two resistors on it, but it's hard to tell if they are for stabilization or for leds (or both?). It came together with the LM3886 boards, so there is a chance it will work for them, but i'm not sure it will work with a single secondary (no center tap) tranformer. See below pictures!

I could be getting it the wrong way, but I was thinking that since the two LM3886 boards (each with one IC) should be running in phase, they could never draw the same current from + and - at the same time. As audio is AC it should be on + or - side, not both at the same time.
Would a center tapped or dual secondary transformer be way better for the application, why so?

A single supply is perfect for amps driving a bridged load, since no current needs to flow back into the ground (it goes rail to rail). You might want to look into going bridged, but you need to carefully consider the impedance of the speakers, rail voltages, etc.

You can also do what you had proposed in the first place. You will need a couple of large (at least 10 watt) resistors to use as ballast resistors. I would guess that your rails will come out to about +/- 28 to 30 VDC, so using a 100 ohm 10 W resistor from each rail to ground would probably work (note that they will get very hot). You will also need to make sure that the amps are AC coupled; this is very important in order to avoid loading one rail more than the other.

You are right about each amp only drawing from one rail or the other at any given instant. This is true. If you realise that bass is where >90% of the power is used, and bass is always in phase between channels, then you will also realize that both amps will tend to draw from the same rail at any given time. Now, you can mitigate the effects of this. Make one side inverting (preferably by inverting the signal then feeding to the amp rather than changing the amp configuration to inverting) so that it draws from the opposite rail as the other amp. Wire that channel's speaker terminals backwards (red: ground, and black: hot). And don't forget that you did this so that you don't assume that the black terminal is the ground one, though you should NEVER assume that anyway. This comes with a bonus feature: you can easily bridge the amp by wiring a speaker of appropriate impedance between the two hot terminals and not using the ground terminals.

[chrispenycate]

Run it single rail with an output capacitor? It's reasonably easy to define a nice stable centre point for the input reference; it's just the speaker current that's upsetting things.

Not elegant, perhaps, but once upon a time nearly all transistor amps were wired like that.

[tuhkam]

Thanks everybody!
I'm considering buying a 200VA 2*24V (2*4,16A) transformer to keep things more simple.

Would a 150VA 2*19V (2*3,94A) give a too low wattage, weaker bass?

The first one might result in about 35V DC and the second one at around 28V I imagine.

[Greg Erskine]

Hi tuhkam,

Quad had a power supply with a virtual ground. You need a couple of transisitors and 4 resistors in parallel with the caps.

http://quadrevisionspot.blogspot.com...ual-earth.html

regards