Pros/Cons on Split supply vs. Virtual Ground

[Archwn]

Hi everyone,

I have seen so many design on headphone amplifiers using virtual ground or ground channel driver.

OTOH I see none of them so far in normal speaker amplifiers. The laters seem to use split supply instead of virtual ground.

If the first one is good, why we don't see it in our speaker amplifiers? And if the later is good, why we still see people using virtual ground/ driver in their headphone amps?

Or if they are both have pros and cons?

I'm seeking for the explanation of what would be the difference between these implimentations and what will be pros/cons comparing these together.

Thanks in advance for your answer.

AK

 

[KSTR]

Split supply is a passive bridge circuit (easily seen when you draw it in its completeness, without using GND or supply symbol shortcuts). "Virtual ground" is an active bridge circuit which is not driven by the signal. This is not very economical for high power designs and also has some drawbacks (you only add the distortion of the bridge if you don't drive it). There are some clever design which drive the "virtual ground" with the error signal from the main amplifier to lower the effective distortion, though. But then the "virtual grounds" of different channels cannot be connected.

- Klaus

 

[Nico Ras]

Split supply is a passive bridge circuit (easily seen when you draw it in its completeness, without using GND or supply symbol shortcuts). "Virtual ground" is an active bridge circuit which is not driven by the signal. This is not very economical for high power designs and also has some drawbacks (you only add the distortion of the bridge if you don't drive it). There are some clever design which drive the "virtual ground" with the error signal from the main amplifier to lower the effective distortion, though. But then the "virtual grounds" of different channels cannot be connected.

- Klaus

The first power amp I have seen using it was QUAD 306. I have used it in many of my designs.

 

[Nordic]

Relating to your question... simply those units are mostly used with portable sources with floating grounds too...

Single sided is realy usefull from a simplicity point of view though, but I agree maybe not for a big poweramp... I am busy makeing an fx pedal tonigt, the same DPDT switch switches on the unit, and status light as well as swithing the signal from bypass to FX...

 

[leadbelly]

I have seen so many design on headphone amplifiers using virtual ground or ground channel driver.

OTOH I see none of them so far in normal speaker amplifiers. The laters seem to use split supply instead of virtual ground.

If the first one is good, why we don't see it in our speaker amplifiers? And if the later is good, why we still see people using virtual ground/ driver in their headphone amps?

For a headphone amp powered by a single battery, the virtual ground circuit is a simple way to derive split rails and thereby avoid a big coupling cap on the output. In short, it is the easy/elegant solution.

For a poweramp, the virtual ground circuit gets more difficult to implement since the currents involved are so much larger. The split supply is the easy/elegant solution in this case.

 

[Nico Ras]

Re: Re: Pros/Cons on Split supply vs. Virtual Ground

For a headphone amp powered by a single battery, the virtual ground circuit is a simple way to derive split rails and thereby avoid a big coupling cap on the output. In short, it is the easy/elegant solution.

For a poweramp, the virtual ground circuit gets more difficult to implement since the currents involved are so much larger. The split supply is the easy/elegant solution in this case.

The current through the loudspeaker returns through the supply capacitors not the center tap of the transformer.

Not difficult to implement a virtual earth at all, DC current is a few milli amps or about as high as the NFB current in the amp.

And no, a split supply is no more difficult than a single supply, one uses two transformer secondaries while the other two capacitors.

 

[leadbelly]

Re: Re: Re: Pros/Cons on Split supply vs. Virtual Ground

The current through the loudspeaker returns through the supply capacitors not the center tap of the transformer.

Not difficult to implement a virtual earth at all, DC current is a few milli amps or about as high as the NFB current in the amp.

And no, a split supply is no more difficult than a single supply, one uses two transformer secondaries while the other two capacitors.

You seem to be suggesting that a simple passive virtual ground is adequate for a power amp, which goes against EVERYTHING I have read on the subject.

You might want to look at http://en.wikipedia.org/wiki/Virtual_ground

 

[KSTR]

"Ground" is a virtual concept anyway (electrons don't know about it), so much discussion about it is more or less semantic.

In that sense, between split supply vs. single supply there is no difference in principle, both are passive bridges, for the single supply the bridge function degenerates close to and at DC, owing to the capacitors.

I repeat my recommendation of explicitely drawing the circuit topologies and then have a look where exactly the currents flow in circles, additionaly discriminating between three time domains: DC, AF and HF. For audio, AF (Audio Frequency) is the main concern, obvoiusly.

- Klaus

 

[Nico Ras]

Re: Re: Re: Re: Pros/Cons on Split supply vs. Virtual Ground

You seem to be suggesting that a simple passive virtual ground is adequate for a power amp, which goes against EVERYTHING I have read on the subject.

You might want to look at http://en.wikipedia.org/wiki/Virtual_ground

Been around before wikipedia and op-amps.

As Klaus points out draw the current flow then it will become clear.

If you analyze the circuit you will see that your ordinary amplifier has a virtual ground most of the time since the rectifier diodes only conduct for a small percentage of the time.

While the diodes are switched off your whole amplifier relies on the potentials across the power supply caps being the signal ground reference.

 

[KSTR]

Now I found that drawing again I made a while ago to show the mentioned current flow. As Nico points out, when the diodes are off, it's only the caps that are at work, closing the loop. And when the diodes conduct, the currents are rather uncorrelated to the signal.

With a little imagination and using some black box circuit equivalents for cap arrangements etc, one can also redraw a cap output amp and will end up with exactly the same schematic, with the only difference -- wrt to the output stage -- that the center tap is always floating (while with a split supply it's only floating most of the time).

When we speak of virtual ground in the context of power amp (speakers or headphones) then we have an active return point, some sort of output stage. This changes the situation a bit in the details, but not in the basic idea that the load currents always go through the supply caps in some way or another (exept for constant current class-A bridge circuits).

- Klaus

 

[Nico Ras]

Hi Klaus,

very nice explanation. I trust that our friend Archwn's curiosity has been satisfied.

 

[Andre Visser]

Re: Re: Re: Pros/Cons on Split supply vs. Virtual Ground

The current through the loudspeaker returns through the supply capacitors not the center tap of the transformer.

Not difficult to implement a virtual earth at all, DC current is a few milli amps or about as high as the NFB current in the amp.

And no, a split supply is no more difficult than a single supply, one uses two transformer secondaries while the other two capacitors.

Sorry I can't see your point, if you create a virtual ground it must be able to handle the full load current, how else can it be a stable "0V" for the speaker?

 

[Nico Ras]

Re: Re: Re: Re: Pros/Cons on Split supply vs. Virtual Ground

Sorry I can't see your point, if you create a virtual ground it must be able to handle the full load current, how else can it be a stable "0V" for the speaker?

Don't understand your statement. What do you mean by full load? Only half the current actually flows through the load. One would assume 0V is half way between the power rails because the geometric sum of all the voltages, signals and noises at the summing point is zero, hence the term 0 volts and not +1 or -5V

 

[Nico Ras]

Re: Re: Re: Re: Pros/Cons on Split supply vs. Virtual Ground

Sorry I can't see your point, if you create a virtual ground it must be able to handle the full load current, how else can it be a stable "0V" for the speaker?

Rethinking at what you say, you are now touching the subject of how large must the supply capacitor be to do the job properly.

This could lead to a interesting discussion when the rules of thumbs are displayed and some thumbs are longer or fatter than others.

 

[Andre Visser]

The way I see it is that on a split supply, you connect the speaker negative to 0V and speaker positive to amp output which can vary beween +V and -V. So all the current flowing through the speaker also flow to 0V.

If you want to create a virtual 0V it must be able to supply or sink the total current flowing through the speaker or else the "0V" will not be stable.

 

[Nico Ras]

The way I see it is that on a split supply, you connect the speaker negative to 0V and speaker positive to amp output which can vary beween +V and -V. So all the current flowing through the speaker also flow to 0V.

If you want to create a virtual 0V it must be able to supply or sink the total current flowing through the speaker or else the "0V" will not be stable.

What if we were to call the supply 0V; +V/2 and +V, you would then couple the speaker negative to +V/2.


Imagine the two caps are two batteries for most of the time.

 

[Andre Visser]

OK, so you rely mostly on the amplifier output, through the speaker to keep the supply caps virtual GND point charged at +V/2. Clever he , if you think about it this way it show the importance of good supply caps.

Thanks

André

 

[fredex]

My take on this. The virtual earth in this case is just a half way point between + and -. If you have two caps in series across a single voltage source and you parallel equal value resistors across each cap the voltage at the cap junctions will be set half way. The load is connected from this junction to the output of the amp. The amp's output voltage (no signal in) must also be set to halfway and so it is referenced to the point between the caps. The two resistors can be high in value as the load is not in series with them. Load current flows through speaker caps and output devices.

 

[Courtec]

Dear all,

QSC uses this type of amplifier with a virtual ground. It is also copied by chinees amplifier manufacturers.

(see the attachment.)

This amp is uncapable of delivering DC on the output, that's a big plus. Unless you have a short in one of the capacitors. then everything will blow!
When i compare it with brands like Crown, Peavey, Crest, it is a bit weak in the bass.

I am uncertan about the ground connection with this type of amp. Is is wise to install the power transistors to the heatsink without isolating the collector? and can you connect 2 channels ground to the same chassis? I think it will result in strange currents between the amps.

also, i analized the AC current running and it seems to me that the voltage swing is the same as the total voltage multiplied by 2.
Because the return current must run over the negative (or positive, depends where you start) voltage connection to the rectifier.

Or will the voltage over the capacitors always be the half of the total? When i look at the specs, this might be the case.

Interesting stuff, any replies /remarks/suggestions welcome!

Koert

 

[jcx]

that QSC is an example of a floating supply "common emitter" output - viewing the output stage as a single compound sziklai/cfp "transistor"

separate transformer sec are required for each channel

the channel grounds may be connected together - with the same or less concerns that exist for any amp with common power output ground connection

the QSC schematic actually supports a proper use of "star" ground referenced to the output gnd connection by having diff amp input stage

if the floating sec were center tapped then the ct could connected to the output/cap midpoint and you'd be able to put out DC all day long - this would also assure more equal V across the output/ps caps with C value tolerance variations