The function of the 0 v is to provide a common reference for the input and output signals and the supplies. You input the signal with reference to ground, and you take out the output signal with reference to ground. You give the amp a supply with reference to ground. It's not just in AB, it's in ANY electronic circuit. Some circuits (like transformers or fully differential circuits) will let you change the reference point between input and output.
The important thing is that ground is a single point, conceptually and physically. If you run a ground wire and connect input ground to one side, output ground to another side, supplies in the middle, the currents in the ground wire will cause voltages between the various points. That means that if you measure the input ground it will already have a signal w.r.t. the output ground even with no signal. Depending on the specific circuit details, that ground-wire voltage can become added to the input signal and be amplified, so your output is no longer the defined gain x input. Especially non-sinusoidal (assuming sinus input signal) supply currents can cause quite high distortion components in the output signal. That's why most people would try to use a star ground point to avond these problems.
Jensen Transformers had a good text on this, titled somethig with "Grounds bla bla and getting it right for a change". I'll see if I can find it.
Jan Didden,
The important thing is that ground is a single point, conceptually and physically. If you run a ground wire and connect input ground to one side, output ground to another side, supplies in the middle, the currents in the ground wire will cause voltages between the various points. That means that if you measure the input ground it will already have a signal w.r.t. the output ground even with no signal. Depending on the specific circuit details, that ground-wire voltage can become added to the input signal and be amplified, so your output is no longer the defined gain x input. Especially non-sinusoidal (assuming sinus input signal) supply currents can cause quite high distortion components in the output signal. That's why most people would try to use a star ground point to avond these problems.
Jensen Transformers had a good text on this, titled somethig with "Grounds bla bla and getting it right for a change". I'll see if I can find it.
Jan Didden,
I forgot. Most seasoned engineers content that an amp in reality has 5 inputs: the input, ground (see my previous post), the two supplies and the output.
Unwanted signals can enter and end up as output signals at any of these points. Therefore, ideally supplies should be zero volts AC: no ripple, no noise, no signal-dependent components. This is no sinecure: If you have 1V of ripple, and your amp has 60dB supply rejection, you have 1 mV to the input referred and multiplied by a gain of, say, 20, gives 20mV at the output. With an output signal of 20V, that's just 60dB down, so whatever the clever design of that amp, your S-N & THD will never be better than -60dB which is 0.1%.
Of course this never shows up in a simulation, where normally you use batteries or DC sources for supplies which *are* ideal. So your 0.001% simulation THD+N is pretty meaningless unless you know how to built it to get that performance out.
Jan Didden
Unwanted signals can enter and end up as output signals at any of these points. Therefore, ideally supplies should be zero volts AC: no ripple, no noise, no signal-dependent components. This is no sinecure: If you have 1V of ripple, and your amp has 60dB supply rejection, you have 1 mV to the input referred and multiplied by a gain of, say, 20, gives 20mV at the output. With an output signal of 20V, that's just 60dB down, so whatever the clever design of that amp, your S-N & THD will never be better than -60dB which is 0.1%.
Of course this never shows up in a simulation, where normally you use batteries or DC sources for supplies which *are* ideal. So your 0.001% simulation THD+N is pretty meaningless unless you know how to built it to get that performance out.
Jan Didden
This seems to relate to something I have been puzzling over in considering power supplies for my sonic impact tripath amps; these things are ludicrously power-supply sensitive, and I am currently running one with a SMPS power supply to run oris horns from 160hz up, and a linear supply for the bass amp.
I recently read a review of a commercial tripath based amp that used a pair of 540VA toroids; but this wasn't a dual-mono supply, the transformers were split between powering the positive and negative rails. The theory was that powering each line with its own transformer avoided variances in the 0V line, which slow the recovery time of the power supply. This is because the 0V line has to re-centre itself after a transient due to it being pulled off the 0V reference, slowing overall recovery of the PS. With two transformers both with their own 0V line, you always have one line locked to 0V no matter which line has the transient - hence no recovery time.
I don't know enough to tell if this is b-s or not; and if not, how easy would it be to apply to a 13.4V tripath supply. Are there any DIY PS designs using this approach?
I recently read a review of a commercial tripath based amp that used a pair of 540VA toroids; but this wasn't a dual-mono supply, the transformers were split between powering the positive and negative rails. The theory was that powering each line with its own transformer avoided variances in the 0V line, which slow the recovery time of the power supply. This is because the 0V line has to re-centre itself after a transient due to it being pulled off the 0V reference, slowing overall recovery of the PS. With two transformers both with their own 0V line, you always have one line locked to 0V no matter which line has the transient - hence no recovery time.
I don't know enough to tell if this is b-s or not; and if not, how easy would it be to apply to a 13.4V tripath supply. Are there any DIY PS designs using this approach?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.