Class A power amplifiers

[Reid Malenfant]

This is a project that has literally been coming together for about 22 years Ever since i started working at a rather large electronics firm back in 1988 i have gradually been gathering components together to do just something like this

I'm pretty sure that anyone who has experienced what a real class A amplifier can do with a decent set of speakers is never likely to be satisfied with anything less & i guess i'm one of those people. I have built a good few solid state class A power amplifiers of different design, only one pair were from a design other than my own which is here:-

Apologies as it doesn't give the component values, but it's a 60W RMS 8ohm class A amp (inverting).



This was actually the first one i built & it's quite a decent design with fairly low overall feedback due to the local feedback used to cut each stages gain. From what i remember there is 20Db (approximately) of global feedback. In contrast a class B amp may well have over 20Db @ 20Khz which will result in vastly more in the midrange & low frequencies. Class B amps need this to attempt to eliminate the distortion caused by the output stage due to the output transistors switching on & off

In a class A amplifier the output transistors should never turn off & because they don't the distortion caused by the output stage is very much lower. The big problem though is that class A is so inefficient it's a joke. Theoretical maximum efficiency is 50% in a push pull output stage but with component losses 40% efficiency is nearer the mark or less. Here's the rub though, it's only 40% efficient at full output IE in the case of the above amplifier at 60W RMS output

Taking the above example again to produce 60W RMS in a push pull output stage the output transistors need to have a bias (quiescent current) of 1.936 Amps assuming an 8 ohm load. In reality it's best to assume that the speaker (rather than an 8 ohm test resistor) is connected to the output & the impedance of these varies. So we'll assume the impedance drops to 6 ohms & set the bias current to 3 Amps. With a +/- 40V supply we are now dissipating 240W per channel without playing any music If i remember correctly the above amplifier was actually good for 75W RMS 8 ohms before clipping & this was probably the reason that it was suggested to set the bias current to 4 Amps. So how does a couple of 320W musical heaters sound?

The good thing about class A amps is that as you increase the volume level they actually cool down a bit as some of the power dissipated accross the output stage is actually fed to the speaker


There are ways of making things more efficient.

As i'm not the richest person in the world i'm going to make these amplifiers as efficient as possible. One reason is because they are going to be quite powerful & the second is because there are going to be 6 of them.

As long as the output transistors never turn off the amplifier will remain in class A operation. Only one manufacturer that i know of actually designed & built a much more efficient class A amplifier, but they made a hash of it In typical Japanese style they used a shed load of feedback on the thing & killed the sonics in the process, enter the Technics SE-A1 @ 350W RMS 8 ohm per channel. It cost a fortune & weighed a ton & probably sounded unremarkable. Here is the circuit diagram (as far as i know)..



What you have there is effectively two amplifiers, one class B & one class A. They are effectively in parallel & how it works is the class B amp output is connected to the ground of the class A amps power supply (which is floating). Because the class A amps PSU is only +/- 5V the dissipation is low. It's the class A amps output that is connected to the speaker though

Here is a more simple block diagram:-



How the low voltage PSU tracks the output:-



Finally a japlish explanaition




I'm not sure where or who designed these but here is a similar take on the very same thing with a few circuits





Here is a single ended ouput version




Been there & done that.

Well i have also sucsessfully built a couple of these weird class A amps & like the Technics circuit i used two amplifiers per channel but unlike them i used a single ended output stage Only problem was that as i required each amp to have vastly different feedback (much lower in the class A section) there was a difference between the phase margins & phase shifts at high frequencies that i wasn't entirely happy with. Still i have a couple of 50W RMS 8 ohm single ended class A amps that sound rather good & are efficient. They'll also happily do 95W RMS 4 ohms again in class A. I just went to town on the things & built in stupid stuff like speaker sensing so it automatically selected the right bias current on power up & silly power meters etc

What i'm planning will just be as simple as possible with the exception of different bias settings for different loads. I have just about everything i need except for aluminium sheet & 6 dirty great big transformers for the class B amplifiers. I'm not going to list what i have here but I have all the output transistors for both amps (A & B) storage capacitors for the PSUs & all the small signal & medium power transistors as well as 36 x 0.5C Watt heatsinks (6 per amplifier).

These are drilled for 2 x TO3 transistors per heatsink:-



I'll be using 6 x MJ15024 & 6 x MJ15025 per channel as the main dissipating elements in the class B section, these are 250W rating per device. In the class A section i'll be using 3 x MJ11032 & 3 x MJ11033 per channel. These are 50 Amp 300W darlingtons so well over rated

Improving efficiency further still.

As i'm not a goldmine i tend to use what i have kicking about. As i happen to have a couple of boxes of BHC 10,000uf 40V screw terminal capacitors i thought i'd use them in the class B PSU. The astute or knowing among you will know you can't generate 250W RMS 8 ohms with a +/- 40V supply (unless bridged).

So the class B amp won't be class B at all, it'll be class H or if you like rail switching. These are normally used in high power PA amps to increase efficiency & that's exactly what i intend to do. The problem is they tend to introduce spikes in the output of the amplifier when rail switching & even though i'll have a class A amp on the output which should sort it out i'm not that sure it'd be the best thing So what i intend to do is use the MJ15024/5 as cascodes & the actual output transistors of the class B amp will be TIP35/6. The cascodes will not only seriously reduce the distortion from the class B amp but will shield the output transistors from the effects of rail switching & get rid of those glitches on the output Here is a simple block of what i intend to do. Notice that the class B (H) output & class A output stages have a gain of 1. I can use different amounts of feedback for each stage & only the class A output will have it's output fed back to the front end. This is an NDFL or nesting differential feedback loop & the same as what i used on my 50W single ended amps.




Like my other stalled projects it'll be a little while before i get cracking on this. I need a room to work in & i reckon a few months will see me there with a clear room to get on in. Meanwhile i thought i'd dump this lot on here just in case i missed something. I can't think of everything & all this cobblers is floating around up here in my head so i guess it'd be easy to mess up.

I guess that gives a bit of time for any holes to be picked at & solutions thought through Hope i haven't bored you all to death

[Puffin]

ZZzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzz

[Rare Bird]

2 x 180,000uF! they would cost a fortune

[anthonyTD]

interesting stuff mark!
Early transistor circuits had quite a few similarities to valves, even down to using output transformers.
The first circuit you provided reminds me of that.
Anthony,TD...

[Rare Bird]

interesting stuff mark!
Early transistor circuits had quite a few similarities to valves, even down to using output transformers.
The first circuit you provided reminds me of that.
Anthony,TD...

Quad '50' used output transformers.

[Reid Malenfant]

ZZzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzz

Well you are still alive

2 x 180,000uF! they would cost a fortune

Got them already Andre, 12 x 120,000uf caps & a box load of 15,000uf caps Don't forget they only need to be low voltage so not so expensive after all

The 120,000uf caps are 15V (well over rated & bloody huge i might add!) & the 15,000uf are just about the same at 10V Supply will be between 4V & 5V per rail +/- so not a problem & all here already. The 120,000uf caps are each 3" diameter & just under 6" in height They'll fit in though - just..

I guess i have actually spent about £800 so far getting all this stuff together, like i say though it's taken many years I have a feeling i'll be spending getting on for that again to complete this. Which is bugger all in reality

For those that have posted after i started to reply to Andre there will be no output transformers, horrible things

[Jonboy]

there will be no output transformers, horrible things

My mate has just got a solid stater as above, it sounds very good indeed and very valve like, thus musical

[anthonyTD]

Well you are still alive

Got them already Andre, 12 x 120,000uf caps & a box load of 15,000uf caps Don't forget they only need to be low voltage so not so expensive after all

The 120,000uf caps are 15V (well over rated) & the 15,000uf are just about the same at 10V Supply will be between 4V & 5V per rail +/- so not a problem & all here already.

I guess i have actually spent about £800 so far getting all this stuff together, like i say though it's taken many years I have a feeling i'll be spending getting on for that again to complete this. Which is bugger all in reality

For those that have posted after i started to reply to Andre there will be no output transformers, horrible things

I was refering to the circuit itself!
nothing wrong with output transformers when their designed correctly, also they provide excelent isolation for your expensive speakers when things go tits up further down the chain, unlike horrible directly coupled transistor output stages.

[Reid Malenfant]

My mate has just got a solid stater as above, it sounds very good indeed and very valve like, thus musical

Exactly Low feedback & class A = happy bunny. Doesn't matter if it's valves or solid state if each are done right

PS. pmsl at your avatar

[Rare Bird]

Got to admit i slightly prefer the sound of Quad '50's (Output Transformer) againt the '303' (Output Transistors)