Regulated power supply for class A ?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
The generally agreed view is that there is little point doing it not to mention that it's a very serious undertaking.

Then I disagree with the general view...

Due to the high continuous current the ripple in a class A amplifier really needs a regulator.

Already in 1978 I built the first capacitor multiplier.
The class A amplifier I had built had a faint hum with my +100dB sensitive speakers. The surprise was big when the capacitor multiplier was added...
A - at the time unexpected major enhancement in fidelity...
 
I'm a bit confused here.

Wouldn't you always want to use a regulated power supply? What is the disadvantage here beside more parts and the inefficiencies related to linear power supplies?

And not using a regulation would make your bias point subject to transformer load regulation.
 
Hi,
Fusion, thanks for your reply. I understand the advantages of less ripple and a regulated voltage, but what I am trying to assertain is the audable implacations e.g what sounds the best. By the way the front end will have it's own regulated supply, my question relates solely to the driver/output stages.
Thanks
Alan
 
The generally agreed view is that there is little point doing it not to mention that it's a very serious undertaking.

A high performance regulator for the front end, otoh, seems certainly like a good idea.

haha, I think its more the serious undertaking than general agreement, as i've only seen and heard much better results. people make up reasons why its not good because they dont want or cant afford to do it....

both can be good, but take a look at the wire power amp thread for the insanely clean transient response and low noise with DPS600

it has to be fast though, basically building a second set of class A amps to power your class A amps. possibly more important for Class AB than Class A though IMO
 
A class A amp driven by an smps? Sounds almost obscene :)

A linear regulator for a 100W class A amp at least to me does sound like a serious project. And in order not to limit transients it should probably be an order of magnitude more powerful than the amp.

Once i owned a Naim with a regulated supply. In complexity the regulator was similar to the amp. It did some things quite well but neither bass, nor dynamics were anywhere close to acceptable.
 
for class AB: I am a firm believer in BL**DY big caps. A 100% over-rated [regulated] power supply [with a cap multiplier] is a given

the reasoning: class AB draws per the signal, so the ps needs to be ready to 'dump' heavy current


for class A: a floating power supply, with cap multiplier, although I still put in bl**dy big caps.

the reasoning: class A draws steady current, but needs to have as little ripple as possible

cap multiplier is so easy, I have no idea as to the question.............especially if you are already using a series pass regulator.
 
If properly designed, a series regulator will be far less stressed than either a class A or class AB output stage - simply because you want to keep the voltage drop on it minimal, usually well below pass transistor SOA starts limiting available current below the simple P=VxI law. Not so easy with the output stage, especially for reactive loads!
In theory, with a high PSRR design, a regulator is superfluous. Adding one for the input stages is a far less involved affair aind simply put, it keeps a lot of unwanted stuff off the power rails before PSRR even comes in as a consideration. However - there are relatively simple topologies that never the less work well but have relatively low PSRR and a regulated power supply, or, keeping it simpler - a capacitance multiplier - is of great help.
Keep in mind that you can only use the available power supply voltage up to the bottom of the ripple waveform, so if your cap multiplier or regulator regulates to a voltage just a bit below, the losses are minimal but the amp is not subjected to ripple + harmonics from the power supply, and it clips cleanly. Also, it's easier to add a protection mechanism in the power supply, especially for a simple amp, where it will be further from any interaction with the actual audio signal.
 
Member
Joined 2003
Paid Member
Attached is a picture of the ripple eater (AKA cap multiplier) I used on my e-Amp. This only powers the front end, but you can see the very dramatic improvement you get - about 30-40dB reduction in ripple. Especially important is that the HF ripple on the supply is removed.

It is quite easy to apply this concept to power the whole amp. Because the series pass device is close to saturation, the power dissipation is very manageable.

I will be trying this on my next class A amp for sure (although I can say my current sx-Amp is very quiet even without a ripple eater).

The top trace is the PSU ripple before the ripple eater
The solid trace below it is the output side of the ripple eater.

Amplifer output was 150W into 8 Ohms (scope probe switched to 10x) and the driving signal into the load was 4 kHz.

I think a full linear reg is another story completey and in my view not a smart approach on a power amplifier - better to focus on improving PSRR, good layout and ripple eater (especially for the front end circuit). Just connecting your comp cap to the right node in a MC VFA can gain you 15-20dB of PSRR improvement (see Harry Dymonds paper and Doug Self for some examples and ideas).

:)
 

Attachments

  • Ripple eater.jpg
    Ripple eater.jpg
    532.8 KB · Views: 344
Last edited:
you know, if you actually looked at the topology its not really an SMPS. more a class A amp with differential feedback, powered by an SMPS, the noise is obscenely low and the transient performance quite remarkable. get with the program ;) its a way to get extremely good regulated performance at 2 x 600W without building an amp larger than the amp you want to power to power your amp (in the case of a linear regulated supply)

I knew it would cause a stir, I took some convincing myself before deciding to get a couple. the amps i'm powering arent pure class A, rather high bias AB (600ma @ +/-65vdc for the lateral outputs) so a somewhat different use-case to here, but it will be in Class A for most of my usage at normal listening levels.

a linear regulated front end with unregulated outputs and a fully linear regulated linear bench supply were compared to it, powering the amp in question. the difference in performance was significant across all measures with high end agilent and AP2. square wave, multitone and IMD were like nothing I had seen before, so I decided it was worth a try. the savings in space and expense were icing on the cake.

I will also be trying them on the outputs (the FE supply is too high voltage) of a babelfish JX, jfet circlotron (R125) and aleph JX, maybe the F6 (all partially finished projects) just for kicks, I expect it to do well with them too
 
Last edited:
I will look into the capacitor multiplier circuit.

Is there any issues of using a linear regular for the input stage and predriver and a capacitor multiple circuit for only the output stage? I suppose as long as the output stage collector voltage is higher than the predriver there won't be any issues of over driving the output stage.
 
for class A: a floating power supply, with cap multiplier, although I still put in bl**dy big caps.

the reasoning: class A draws steady current, but needs to have as little ripple as possible

Only Balanced Class A draws steady current! Ordinary Class A current changes according to the output waveform. Just that there is a high idle current and no cross-over so the current change is a lot less drastic than Class AB/B.
 
Only Balanced Class A draws steady current! Ordinary Class A current changes according to the output waveform. Just that there is a high idle current and no cross-over so the current change is a lot less drastic than Class AB/B.

This is just trying to be overly technical. In practical terms, the average current draw in Class A is constant and you will not see voltage sag.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.