I'm trying to picture what the power supply "sees" in terms of needing to source and sink current for the output stage of a class A amplifier.... trying to think about power supply design in terms of it's impedance at audio frequencies.
With a class A/B amplifier, the supply is asked to source and sink currents above the quiescent current level when the output stage amplifies an audio peak in class B... these varying current demands at audio frequencies makes me think that low impedance for the supply is essential for good sound in such a design. I get that.
But in class A, the output stage is always drawing the same current, and the current is either being dissipated in the output stage or is sent through the load (i.e., the speaker)... does the power supply have to sink or source audio-frequency currents in a class A amplifier? It doesn't seem so, since the current drawn is always the same....if that is the case, then there's no need to worry about supply source / sink impedance...? Could that be right? I just can't wrap my head around this.
With a class A/B amplifier, the supply is asked to source and sink currents above the quiescent current level when the output stage amplifies an audio peak in class B... these varying current demands at audio frequencies makes me think that low impedance for the supply is essential for good sound in such a design. I get that.
But in class A, the output stage is always drawing the same current, and the current is either being dissipated in the output stage or is sent through the load (i.e., the speaker)... does the power supply have to sink or source audio-frequency currents in a class A amplifier? It doesn't seem so, since the current drawn is always the same....if that is the case, then there's no need to worry about supply source / sink impedance...? Could that be right? I just can't wrap my head around this.
So, you're saying, at some point in the AC signal cycle a class A amplifier draws more current than it does at idle? Wouldn't that only be true with one of the sliding-bias arrangements, like Stasis, etc?
For example, a mini-Aleph, that's the particular amplifier I am visualizing for this exercise. Fixed bias, heavily into class A. Does the power supply in a mini-Aleph "see" audio-frequency changes in current sink / source when the output stage is amplifying a signal?
Implications of constant current being sourced / sunk from supply....
Sorry about my comment re: Stasis, I don't really know how Stasis works, my bad
, I made an assumption that it was a sliding bias arrangement. There ARE sliding bias designs, meant to keep the output stage in class A as power demand increases due to the drive signal. Stasis isn't one of these so Stasis isn't what I want to discuss. Neither are sliding bias designs.
As Mr. Pass has said, class A designs with constant current sources don't have varying demands on current from the power supply, likewise balanced push-pull Class A designs.
My question relates to considerations of power supply source / sink impedance for such constant-current designs. Obviously, the DC supply needs to filter out ripple from the AC line, and it is wise to use capacitors that will also attenuate higher frequencies that might get through the power line (but I assume there's at least a rudimentary RFI filter somewhere in the path of the incoming AC power, this is just good practice.)
So let's assume the power supply filter capacitors are quite adequate to remove ripple, and that RFI is minimal on the DC output. Let's further assume that there is enough "reservoir" capacitance to help stabilize the DC supply in case of minor short-term sags in the power line.
In a class AB design it's best to have DC supply sink & source impedance as low as practical through the audio band, especially at low frequencies where instantaneous sink / source current demand is quite substantial as the output stage drives the speakers to produce drum whacks, etc.
But here we have a constant current source / sink situation, I don't see the need to think about source / sink impedance of the DC supply at all, really. What I'm asking is- is that right? With a high-bias balanced class A push-pull amplifier, there's no audio frequency currents being sourced or sunk by the DC supply? All the audio current action is happening in the output stage itself? Yes?
I'm just trying to straighten out my thinking here.
Sorry about my comment re: Stasis, I don't really know how Stasis works, my bad
, I made an assumption that it was a sliding bias arrangement. There ARE sliding bias designs, meant to keep the output stage in class A as power demand increases due to the drive signal. Stasis isn't one of these so Stasis isn't what I want to discuss. Neither are sliding bias designs.As Mr. Pass has said, class A designs with constant current sources don't have varying demands on current from the power supply, likewise balanced push-pull Class A designs.
My question relates to considerations of power supply source / sink impedance for such constant-current designs. Obviously, the DC supply needs to filter out ripple from the AC line, and it is wise to use capacitors that will also attenuate higher frequencies that might get through the power line (but I assume there's at least a rudimentary RFI filter somewhere in the path of the incoming AC power, this is just good practice.)
So let's assume the power supply filter capacitors are quite adequate to remove ripple, and that RFI is minimal on the DC output. Let's further assume that there is enough "reservoir" capacitance to help stabilize the DC supply in case of minor short-term sags in the power line.
In a class AB design it's best to have DC supply sink & source impedance as low as practical through the audio band, especially at low frequencies where instantaneous sink / source current demand is quite substantial as the output stage drives the speakers to produce drum whacks, etc.
But here we have a constant current source / sink situation, I don't see the need to think about source / sink impedance of the DC supply at all, really. What I'm asking is- is that right? With a high-bias balanced class A push-pull amplifier, there's no audio frequency currents being sourced or sunk by the DC supply? All the audio current action is happening in the output stage itself? Yes?
I'm just trying to straighten out my thinking here.
In push-pull the supply current does normally vary, even with classA. To have a static supply current demand you need an SE topology (active pass element loaded by current source) and your load needs to be in parallel with the active pass element (not, as normally shown, in parallel with the current source).
So, you are disputing what Nelson says...?
" and balanced push-pull Class A
output stages also tend to have constant draw."
you need to learn to sort the wheat from the chaff. And deliberately omitting the offending short sentence is not being clever.
Even N.Pass can make mistakes where a much more detailed sentence would have made clear what he meant to say.
The shortened sentence is misleading.
Balanced Push Pull ClassA stages have substantially constant current draw from the PSU, only when the output current is less than the ClassA current limit.
Thus even his "balanced ClassA" claim can be wrong because he has not defined the limits.
Last edited:
The full sentence is "Amplifiers biased by constant current sources will have
constant draw (obviously), and balanced push-pull Class A
output stages also tend to have constant draw."
I didn't shorten the quote to mislead, I thought we were not talking about a single-ended design but a push-pull balanced design, so I didn't see the relevance to balanced push-pull operation of an observation about a single- ended design. Two different things, no? Push pull and single ended -
I'm not trying to argue with anyone here, I am just trying to understand the physics. I do not know the answer, and am trying to find out.
Wouldn't a class-A push-pull balanced amplifier normally be operated within the class A regime? I mean, what would be the purpose of building such a thing, if not to use it in class A? If the amplifier is expected to be pushed into class B, how is it a class A amplifier?
Are you saying that it's impossible to make a class A push pull design that remains in class A? If that is true. then I see that the DC rails in a class-A push-pull circuit DO see audio currents sourced and sunk, and my question is answered: for such an amplifier it is important to have low source and sink impedance in the DC supply.
OK, so let's consider the other case - Single-ended class A amplifiers biased by constant current sources. Pass says "Amplifiers biased by constant current sources will have
constant draw (obviously)" - and, please I am not truncating the sentence to mislead anyone, I am just trying now to understand a single-ended amp of this type, and so now I am not asking about, or referring to, a balanced push-pull design.
So, if "Amplifiers biased by constant current sources will have constant draw (obviously)" that is a statement which strongly implies that there are no audio-frequency currents being sourced or sunk by the DC supply. Is that correct? If so, it would imply that source and sink impedance of the DC supply at audio frequencies is not relevant to output audio quality. Right? Yes? No? I'm trying to understand this, not make any kind of claim, not trying to mislead anyone, not trying to spark an argument about this.
I actually have a "Mini Aleph" board around here someplace, and I think my bench power supply can provide it with ± 18 V, the supply is rated 5 amps, I think the Mini-Aleph draws around 2 amps. I can put 0.5 ohm 5 watt resistors in series with the DC rails supplying the amp board, drive the amp to full power into an 8 ohm load, and measure the voltage drop across the resistors with my 'scope. Any audio current being sourced or drained from the rails will show up on the 'scope. I'll post the results.
Last edited:
This appears still to be the case. If a negative rail is used the current demands on that rail vary.
Whether or not there is constant load on the PSU doesn't mean the amplifier is not operating in class A.
Read Rod Elliott's very clear article on Class A amplifiers
Yeah I know that steady current draw doesn't define the amplifier's class. What I am trying to determine are the optimal source and sink impedance characteristics of the power supply for various Nelson Pass DIY type amps. Elliott's article, while very clear, is also rather basic and doesn't touch at all on power supply impedance, so I'm not sure why you referred me to that article. Maybe you didn't understand what I am trying to learn here.
I have a Class A amp here with a Quasi output stage. It is biased so that it will deliver its rated power into an 8 Ohm load in Class A. I have a current meter in line with either the positive or negative supply rail. As I increase the voltage into the load (sine wave) the DC current on the measured rail does not change. If I push it into clipping the measured DC current begins to rise. If I drive a lower load the measured DC current rises as it exceeds the Class A bias available.
Sent from my x600 using Tapatalk
Sent from my x600 using Tapatalk
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.