Actual DIY's master (amplifier)

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

I made a few audio projects so far (amps, etc.) and I believe I'm ready for the next step.

I'm wondering what would be the ultimate amplifier to build (>100W, solid state). I'm seeking for the "highest-fi". Best THD, best S/N, best everything. I know there're always compromises but I still want to know.

At this point I'm considering elektor's crescendo, but I would like to have as many opinions as possible before making a move.


P.s.: Don't forget about schematics ?
You want the very best? Excellent!
What you go for sort of depends on your experience and your access to test equipment. If you can get hold of a spectrum analyzer, network analyser, scope and function generator, each with bandwidth of >10MHz, then a high-speed feedback design is a good approach. If not, then a low or no feedback design may be more fruitful because it is easier to get them stable and to get a clean sound out of them.
You won't find schematics for the top brands like Krell and Mark Levinson on the web although Nelson Pass is generously open with his designs. The Naim amp schematics can be found at:

I don't have an spectrum analyzer at home, but I've a 100MHz scope and an 11MHz function generator.

The AKSA is a kit and is not necessarly what I'm looking for. However, they're selling heatsinks, hardware that I'm having a hard time to find (any other suggestions for heatsinks?)

The Naim power amp design looks interesting but I haven't been able to find speficications. Power, THD, etc.? However it is well documented and I would like to hear other comments on it.

Keep feeding me with suggestions, this is the only I'll be able to make the best choice for my needs.

Hi Gabster,
You'll be able to get quite a way with a scope and FG. I would recommend not building a Naim since it is a high feedback design and needs considerable "tuning" to sound great and you don't have the equipment.

Actually, I just had alook at Neils's site and the schematic he has there isn't quite right. C9 should be 22uF not 220pF! D5 shouldn't be there. C6 is 39pF. Q1/Q2 should be BC549. Q9/Q10 should be MJE243/253. Q11/Q12 are Naim proprietary but are essentially extremely fast transistors (not chosen for linearity). The base networks of Q9/Q10 need to be tailored to the output stage.

You won't find the AKSA schematic on the web site but you can deduce it from the PCB layout graphic. It has simple bipolar PP topology with local feedback around each output transistor. A sort of complementary version of the Naim circuit but with lower feedback and a bootstrap thing on the gain stage (I still don't approve Hugh!). This circuit would be a good place to start and can be fine-tuned.

Most circuits can be made to produce >100W by beefing up their component specs. You need >42V supplies for bipolar outputs and >50V for FET.

I have network analyzers, spectrum analyzers, and probably stuff you haven't seen. You are right it is easier to build no-feedback designs, but you are never going to convince me "then a high-speed feedback design is a good approach." We could go on for years about this, but let's not!

If that is what you prefer, then fine. Just remember how much harder they are to build, and that doesn't automatically mean better, too.

not using the bootstrap circuit?

traderbaum says:

..... A sort of complementary version of the Naim circuit but with lower feedback and a bootstrap thing on the gain stage (I still don't approve Hugh!).....

I hear what you're saying :) What I've read on this kind of design in the past is that you can delete the bootstrap circuitry if you simply supply the driver circuit with a higher rail voltage than the output stage.

I suppose the question is - what magic has Hugh built in to the AKSA that doing this would upset?

Please let us not make things more difficult here.
Neil McBride's schematics may be not "the orginal thing" (one has to read the text also...), but in the 250/135 clone (following the numbering)
1. C9 definitly IS a 220pF styroflex (22uF is WAY off)
2. D5 definitly MUST be there !!! (back to the start... :) ). Please read basic writings, JLH for example.
3. C6 is 39pF indeed. But I think there will not be much of a difference.
4. Q1/Q2 are NOT BC549s (at least not in the orginal). The changed them several times, I think, I use BC239Cs and BC182s.
5. Q9/Q10 should be MJE243/253. You are RIGHT here !
6. The base networks of Q9/Q10 really needs to be tailored to the output stage. But the details shown are not so far off.
7. The AKSA schematic has not very much to do with the Naim schematic, at least they are both SS amplifiers...

OK, the Naim circuits need some tuning, but I my experience no DIY amp sounds perfect from start. My "Naim-clones" sound very good :) after all.
You are somebody with experience in making more complex devices, so you can try them. Some equipment you will need anyway ALWAYS, like a scope, a sinus generator, multimeter, etc. You know that already.

But beware: ALL amps have a destinctive "sound", more or less, you may like it or not. That is one of the many reasons we all are constantly talking here... (Did you hear a Naim-Audio amp at your home or some other place yet ?)
The AKSA could be a tempting thing, many people are raving about it, and the actual cost of the kit does not exceed the cost of the parts and the "cost" of the time selecting them (NO construction of a good amp without selection of parts, sorry).
ESP's project 3a is a very good amp indeed, but somebody took mine away from me already... It is so cheap and simple, you can use for a secondary room if you do not like it.

There is no solution, except to try out and listen. Forget the specs, when you consider the sound quality you want to reach.

But if you are not satisfied with your existing amps, did you consider a better CD / record player and better speakers too ? Sound quality is about synergy too (bad CD-player + good amp = worse sound, you know).


ps: I heard the elektor crescendo twice, I have actually forgotten how it sounded, but I remember having blown them away with a much cheaper lo-poered little "british" amp. Not my kind of thing.
Thanks for correcting me in such a forthright manner (I think). When I wrote D5 I meant D8 - the LED. Q1/Q2 BC239 yes ok. The last time I took a NAP110 apart C9 was a 22uF tantalum. The design must have changed. Odd.

You have the confidence of an expert. At the risk of making things more difficult here ;) , may I ask you a couple of clarifying questions?
1) What do the base networks of Q9/Q10 do?
2) What output transistors have you used and what diode for D5 and why?

IMO the NAP and AKSA are very similar in topology - much moreso than say the Leach or Adelph. We are both, of course, entitled to have different opinions.

Gabster, since you stated you're seeking the "highest-fi" then by all means find the amplifier with the best specs possible. Look for a design with very high feedback, as generally they will provide the best specs. Think Japanese amplifiers in the 70s and 80s. I would suggest you look at the Halcro amplifier at and do a US patent search for the schematics. It's not too difficult to come up with the missing values if you know anything about amplifier design (and just goes to prove that despite all the hype it's obviously not difficult to be issued with a patent these day ... but I never said that!).

HOWEVER! If you are building an amplifier to listen to, forget the specs as they are simply meaningless and sometimes contra-indicative as to which amplifier sounds better. As convenient as specs are, they are generally a poor representation of how the amplifier will reproduce music. To prove my point, look at the number of designs out there. If there was an ideal design based on the best specs there would be only one typology used in high end amps.

Unfortunately once you start circulating at this level, the “best” amplifier is a subjective decision based on your own personal taste. The difficulty is that in DIY there’s often little opportunity to listen to a design before you build it. That’s the price you pay. However it’s not impossible in some cases, for example the Pass designs.

Good luck with your search, and happy listening.


C9 was a 22uF tantalum

There are normally two (mostly tantalum) caps near the front end: The input cap, usually 10uF, and the DC blocking cap, usually 47uF, in the feedback loop to input ground.
The 220p cap goes from the base of the npn driver to ground, you cannot use a 22uF cap there, I think. The amp will work also if you leave this cap out. So you probably mean the cap cross the bias transistor ?

What do the base networks of Q9/Q10 do?
They are a kind of frequency/time compensation of the output stages, that is also why they are different on both halves. It is a part of Naims secrets to this designs. The compensation circuit seem to refer to the special TO3 output transistor Naim is using in their bigger amps. In my smaller clone I left it out (like in the NAP90 or in the Nait) without any problem.

What output transistors have you used and what diode for D5 and why ?
I have used different transistors, always sad about not having the original. I used f.e. for the bigger mono versions the PT77, made by ST for the Cyrus amps, or the Sanken 2SC2921. For the smaller ones I tried the PT7 or the BD911 with success, both made by ST.
I am trying a "El-cheapo" version with the BD241C now, because I have lots of them. I do not yet know if they will work. But it is more a switching transistor anyway...
But my "quest" for the right transistor is far from beeing finished.

The diode is used to linearize the semi-complimentary output pair. I always use a 1N4148. I will try to find the text by JLH for you.
BTW, the LED in Neil McBride's naim schematics is seen as an alternative to the two diodes.

I do not like to discuss the differences between the topology of the AKSA and the Naim amps here, because it is not the topic of this thread. But we could start another one...
Looking at the schematic shows several differences, it seems only similar "from the outside".
Both have indeed in common that they are more or less optimized "oldfashioned" circuits (sorry, Hugh), based on proven techniques.

best wishes
Alright folks,

This is getting very interesting. You're giving me really good suggestions and I'm realizing that I'll have to build them all!

At this moment I think I will pass on the Naim design...maybe too tricky. The same way, I don't think I'll get into cloning a Halcro amplifier.

The specs of the Elliot P3A are very interesting and it's sure that I'll build one in a near future.

All the Pass Aleph amps are very interesting (simple designs but seem to require huge heatsinks). Is there web sites out there that present all the variations as well as schematics(Aleph4,5,60,etc.). This is probably the amp I'll be considering, but once again I will have to make a choice. Any comments on these amps would be appreciated.

Finally could someone else comment on the crescendo, it's a complicated design, does it worth it?

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