20W Blameless Class A

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hello.

I was reading Mr. Self's book Self on Audio second edition just last night and saw an interesting circuit. It was a very sophisticated A-Class design with a Novel current control. It's located on page 305.

Has anyone built this amp? It seems quite efficient - at least, in terms of 'A' Class amps! Here is a quick description;

  • 24VDC Dual rail power supply.
  • Current mirror in the two long arms of diff amp.
  • Beta boosted VAS
  • 1.6A Iq
  • Low cost/high value components.
Here's a picture:
 

Attachments

  • Blameless 20W Class A Amp.jpg
    Blameless 20W Class A Amp.jpg
    96.8 KB · Views: 1,583
Member
Joined 2010
Paid Member
There's nothing wrong with such a design, the controller works just as claimed though Silicon Chip's 20W class A design is already physically worked out for the application and uses more modern parts with lower distortion, current and heat if that is what you seek.

The real problem is buying a transformer, around 150VA, of the precise voltage to give rails no higher than 24V but still allow the amp to deliver 20W (18V AC is too high IMO and regulation is a further waste)

I built a Stripboard version a few years ago and was quite happy until I heard SC's version which had the noise issues of constant high current solved, where I never got it quite right.

There is little point building high performance designs if hum & noise exceeds the capability of the design and Self's Trimodal design is as good as those parts get.You will find more refinements in later editions of his Manual covering lower noise input transistors among other developments.
IMHO, any good class A is down to critical power layout and thermal arrangements.
 
Hi there Ian, Thankyou for your comments on the power supply and also the SC 20 watter. I have a couple of other irons in the fire at the moment but I will try to order a back copy of SC to read up on the mod's. Do you have any idea which issue the article is in?I have fond memories of Coffs Harbour. Back in 2000 I was part of a bike ride from Tweed Heads down to Sydney, we were clapped out by the time we got to the land of the big banana so camped at the show grounds and I really enjoyed the warm water and cold beer!Nice to meet you Ian.
 
Member
Joined 2010
Paid Member
Likewise, farmerjack
Try 2007, August IIRC, in SC but you can see much of the detail on the their site archives.
I should say the mods were to Self's design in his subsequent Manual editions which increase the features to add a nifty class A/B or Winter/Summer switch for example.
(You may appreciate that in Devonport, Tas. - also a lovely part of the planet - Brits would feel as at home there as in Devonshire, I think. I could swear I was in "Tassie" when I drove the UK tourist routes there some years ago. You're even on the Tamar River! :D )

SC's is a straightforward CFP version which is not an issue in class A, just more efficient and linear with only 1,2A quiescent on 22V rails. Neither would be called 4 ohm designs, perhaps 6 Ohms as many small HT speakers tend to be.
 
Last edited:
Class A amp

Hello again Ian.


I wanted to ask if you might give me a little guidance on my above project?


I have read a chapter of Mr. Self's book; Self on Audio - but still don't undererstand the circuit completely.


The schematic I saw in the book (the one I previously posted) had a different output configuration to the two pole cfp configuration that I wish to use:- can you look at this picture now and tell me if I have actually drawn a '2 pole CFP'?


I still haven't seen the Silicon Chip schematic and that issue is proving difficult to get at the local library - I may have to lash out and by the article online!


As you can see from my schematic, I want to use the the original circuit but change the drivers for 669/649 pairs and the outputs for MJ15003/15004 pairs. (because I have them in my transistor box).


I have two of the Alltronics heatsinks H0545 (0.4 deg. Cels/W) and the local sheet-metal shop folded two 40mm by 40mm angles 300mm long and from 6mm aluminium. The angles simply bolt onto the machined face of the heatsinks and provide a platform to mount the TO-3 Output devices.


I recall what you said about 18V being a little high, BUT I actually have a 0-18, 0-18 transformer of 300VA and would like to use it on a good Class A beastie that also uses the above sinks.


If you can find time to look at the attached picture and make some constructive criticisms I would really appreciate it. Although I have been dabbling in electronics for years, I don't actually know all that much about amplifiers.


If you know of a better circuit, then perhaps you might be able to point me in the right direction [schematic].


Cheers Ian, sincerely, Phil Elliott aka farmerjack61
 

Attachments

  • My Blameless 20W A-Class.jpg
    My Blameless 20W A-Class.jpg
    106.1 KB · Views: 1,163
There are a few changes I recommend particularly for a beginner.
1.) delete the CFP output stage and replace it with a dual EF output stage. These are in the book.
2.) delete the shared CCS from the input LTP and the VAS. Dedicate a pair of CCS for the LTP and VAS.
3.) add a protection resistor to the collector of the EF VAS driver. It has 1k0 in the emitter, add ~ 1k0 between collector and audio ground.
4.) add degeneration to the VAS (emitter resistor).
5.) add provision for base stoppers (~2r2) for the output devices.
6.) add emitter resistors to the output devices (~0r22 to maintain the bias sensors).
7.) change the input LTP pair for devices that have much more gain.

D.Self's book shows an explanation of how tolerances around the input stage affect output offset and how currents/voltages change. This in effect is telling you that selected pairs of everything in and around the LTP should be used.

folded two 40mm by 40mm angles 300mm long and from 6mm aluminium.
Make sure the mating faces are flat.
 
Last edited:
Hello.

I was reading Mr. Self's book Self on Audio second edition just last night and saw an interesting circuit. It was a very sophisticated A-Class design with a Novel current control.

Quadrature feedback. Hmmmm. Didn't know Self went there...
I have his book , why didn't I notice this before?

I can think a dozen ways to implement that same function with
vastly fewer parts. One or two transistors at most, and emitter
drop of the same can double for his mystery block of a voltage
reference... He over complicates this to the point of being silly.
I'll attach one example, how simple this all could have been...

Self made a mistake to put such a huge cap across his sliding
bias. When you shunt regulate in real time with a quadrature
feedback of output currents, slowing this down is detrimental.
Should not have cap'd any greater than to make his quadrature
control loop stable.

You only want big cap here (and thermal compensation) when
you drive blind, with no feel of the actual output stage currents.

---

OK, I've labeled incorrectly which MPSA is for NPN and PNP.
NPN should be the 06. PNP should be 56. Just switch them.
You know what I mean...
 

Attachments

  • Selfless.gif
    Selfless.gif
    13.5 KB · Views: 980
Last edited:
www.hifisonix.com
Joined 2003
Paid Member
Yes I also think this current control scheme is a bit complex. It's fairly easy to do it with a single transistor and its electrically and thermally stable.

I also agree that you need a decent supply for a class a amp. Means large filter caps and well filtered supply rails to the front end.
 
There are a few changes I recommend particularly for a beginner.
1.) delete the CFP output stage and replace it with a dual EF output stage. These are in the book.
2.) delete the shared CCS from the input LTP and the VAS. Dedicate a pair of CCS for the LTP and VAS.
3.) add a protection resistor to the collector of the EF VAS driver. It has 1k0 in the emitter, add ~ 1k0 between collector and audio ground.
4.) add degeneration to the VAS (emitter resistor).
5.) add provision for base stoppers (~2r2) for the output devices.
6.) add emitter resistors to the output devices (~0r22 to maintain the bias sensors).
7.) change the input LTP pair for devices that have much more gain.

D.Self's book shows an explanation of how tolerances around the input stage affect output offset and how currents/voltages change. This in effect is telling you that selected pairs of everything in and around the LTP should be used.

Make sure the mating faces are flat.

1) CFP is just fine, providing you have real time overwatch of collector
current. This one qualifies, once you get rid the cap on the sliding bias...

2) Yeah. I woulda just used bootstraps and shunt regulated them down
to the correct (not necessarily constant) current, its a lot fewer parts for
same or superior result...

3 & 4 ) If you had said a base resistor on VAS, I might have agreed.

5) Pointless waste of 2R2 stops nothing. Base itself is 26ohms min.
Ferrite bead on base leg if you worried. Ferrite can easily provide 1K
equivalent resistance against high frequencies.

6) Why? Superior bias sensing is already present in the collector.

7) Maybe the only thing we agree on completely.

Flat mating surfaces) OK, maybe that makes two things....
Sometimes TO92 pairs jam well together in TO5 heatsink.
 
Last edited:
Member
Joined 2010
Paid Member
Appreciate you comments, kenpeter
Hi Phil,
Lots of things are said in comment on Self's designs, but this is a work in progress to the Trimodal. I've built it in that form for others and it works just as described but I would also agree in warnings against heatsink brackets and their flatness if bent. The dissipation of the output transistors is constantly high and with 24V rails, the sinks will go over 60 deg.C in the centre. That hurts, I assure you.

The parts on the design are generics as befits a model but a final design would have selected inputs and output transistors at least. Your OPs will work OK, much as the design types but the inputs would likely be improved with matched low noise KSC1845/A992 or similar non-obsolete parts.

At the very least, space the OP transistors widely, >100mm, as the narrow sink needs every help to get the heat out to the ends to meet its rating. Surely, it's no issue to stand the Txs up and bolt or screw direct to those sinks?

It might be better not to attempt high quality class A unless you are confident of the right way, to design the board and layout as it is going to be expensive whether the result is as fantastic as it darn well should be or just so-so and something you will eventually bin or try to improve.

Yes its a bit lightweight for a reply but others covered specifics and this is New Year, man...recovery time---:h_ache:
 
Maybe the only thing we agree on completely.

What are you on about Ken?

I didn't understand one word of that post.

Did you say that you knew a half a dozen ways to implement this circuit? Could you please draw them up from front to back or input to output for me, I should only take a few minutes for a man of your calibre.

This would be helpful because I am sure you are capable of and confident in the things you say. This would make my job easier.

I have never got a bum-steer from Scotland and Mr.T has stuck with me through other projects that are now up and running like athletes. We attacked each problem in the way that we are doing right now, and it is very thorough and proper. Helpful too!

BTW the second last amp that I built was tickled by the 'T' and other members of this forum, and it is pure and pristine. The output resistors go between the rails and the emitters of the output devices and their collectors went to the speaker output. I am very happy with this configuration. The amp that was produced sounds by far and away, better than my Sony STR-DB2000 which cost $600 Australian. That Sony amp has had no use since I finished the new beast.

I am not stuck on this design, and if someone has a fully tested circuit that would suit my transformer and heatsinks then I am more than capable of laying out and etching a pcb and sticking the bits into my hand drilled holes - no gerber files required.

Are we just urinating headfirst into an onshore gale?

Does anyone have a better tested circuit? One that doesn't reqire wasteful voltage regulators? One that has lots of low down grunt and tinker-bell highs?

The point of mating surfaces is well taken and I read a good article on it at Rod Elliott's website a while back. My plan is to fit the angle to my sinks with 6mm bolts (then dissassemble) and take them down to the local engine builder so that he can give the bits a quick skim under the milling machine. A thin film of silicon grease will then be used to mate the two surfaces.

Further to this I have already assembled the heatsinks using my trusty G-Clamps and applied the heat-gun, to which the sink responded very well. ie. the heat passed through the joint with relative ease. So once I do the job on it, it will respond even better and lower the resistance by about 0.1 degree/watt - calculated using the mass v surface area of the materials involved.

The issue of heatsinks for me is a connundrum because none of the hobby stores stock the sort of stuff I am looking for here in Tassie. The best sink I can buy off the shelf are the ones that I have right now, and the aluminium angle will increase the heat dissapation if I do the job with care, using my trade skills.

On the topic of matched pairs of transistors in the differential amp:
what is meant by the term "matched pairs" are we simply talking about transistors of the same batch and similar beta or is there some other MOJO going on?

Thankyou for your input but try and keep it simple and to the point in the future. There is no benefit to me if you just shoot another member down. I am an amatuer amplifier hobbyist and not an electronics design engineer as every other man seems to be, on this site.

Lets keep it real.

Yours sincerely, Phil Elliott
 
If you build true to Self's schematic , with big cap on bias spreader, yeah.
You will have to match output transistors and all that wasted effort.

But you get rid that cap, so quadrature feedback is realtime: The rule that
"R16 + R17 add to a constant voltage" will force even slightly mismatched
output devices to behave as-if much better matched...

The one thing that probably needs matching the most is the current mirror.
Thats what assures the LTP transistors see similar currents, and thus have
similar VBE. Matching the actual input transistors is secondary even to this...
 
The feedback loop that enforces "R16 + R17 add to a constant voltage"
(or optionally some other combination of resistors and diodes that define
another class) is at 90 degree effect to the feedback loop that enforces
"output will be X times input voltage".

One loop shapes only currents, the other loop shapes only voltage. These
completely independant yet cooperative loops are said (by me anyways)
to be in "Quadrature".

That may or may not be the correct term, but its the closest I know of.
I test this thing called a "Quadtrature Modulator" at work, with inputs
names I and Q. It goes in cell phones. 90 degree phase relationship is
very special, and allows various stOOpid pet tricks to unique effect.

Wether we call it Hawksford, or Circlophone, or SRJLH, matters not.
All the same exact things, if completlely diffrerent. Once you get your
head wrapped around the concept that currents will be shaped to a
class by a different loop than the one that determines output voltage.

As opposed to being at the whim of device curves that might or might
not be ideal compliments (especially over a temperature variance).
We watch actual currents and force them to follow an actual rule.

I see through this "Novel Scheme", because I abuse the very same
trick constantly in my own design. It is very potent once mastered.
Maybe the very best trick to know, outside the classic box.
 
Last edited:
Hello.

I was reading Mr. Self's book Self on Audio second edition just last night and saw an interesting circuit. It was a very sophisticated A-Class design with a Novel current control. It's located on page 305.

Has anyone built this amp? It seems quite efficient - at least, in terms of 'A' Class amps! Here is a quick description;

  • 24VDC Dual rail power supply.
  • Current mirror in the two long arms of diff amp.
  • Beta boosted VAS
  • 1.6A Iq
  • Low cost/high value components.
Here's a picture:

That's a nice book to read isn't it - so much in there, you can read it twice and still there's things to discover in it.

I haven't built this amp. In fact I've not seen any interest in it on this forum - the 'road less travelled' provides some interest and scope for fun.
Do you have access to a high current bench supply ? For experiments etc. a current-limited bench supply is a god-send and saves lots in fuses and blown parts. If not, it might be worth building the power supply first so you have a solid platform. There's one design that I remember seeing on the internet that has built-in over current protection... http://www.pmeweb.co.uk/audio/ ...

Your recent experience with the JLH amp looks great by the way :D
 
Last edited:
General chat

Hello Bigun,

I have a current limited 0-40V & 0-3A single rail supply that uses a switchmode regulator and also a Heavy Duty 3A dual rail variable supply of 0-18V capacity - this one has no current limiting.

One day I would like to build a combination of the two but I have been a little pre-occupied with A-Class amps and the desire to build one that sounds like all of the reports that I have read. - I'm not quite there yet!

I am a wayfaring pilgrim from way back mate, and the road less travelled is my favourite track to peace and quiet. I hope you make time to get out into that wonderful wilderness up there (at least I hope there is some left)!

My uncle went to visit your country about 30 years ago and told me about the big fish the Moose and Bears. Hunter could easily become prey in your outback - worst things we find over here are the snakes and spiders unless you're a swimmer and then the White-Pointers come into play.

Talk soon Bigun.
 
Great to meet you Mlloyd. I'm a muso too, but only in my own lounge room (open E tuned guitar and a harmonica in a frame). Sometimes if I have a few beers it starts to sound good! I am going to google the Pass 20 watter now.

The thing that started me on this trail was Mr. Selfs design, as it involves no voltage regulation on the rails and although others may find the current control complicated it seems less complicated than the three other designs that are available to me.

Have a great New Year and go out and listen to a good band for me, We mainly have Kareoke and Guitar Hero over here and believe me, it's not a good sound. What will happen when all the muso's die and all we have is a bunch of ******* jumping around with virtual air guitars?
 
big brother is watching

Krikeys, I've been censored for saying ******, and yet it is not a particularly offensive term in my country, even the PM is a ****** at times, and she's a woman. When put correctly into context the word ****** can actually fill the void when one gropes for the correct terminology.

I am a little disappointed in the censor but I can just type six blanks ______ to describe the situation person place or thing next time, even if it is a fictional character.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.