Building the ultimate DIY power amp

[till]

And those parts they really need precision are manufactured over here...

[LRRockBox]

There seems to have grown a lot of confusion about what it is exactly I aim to do, here, so here it goes.
Since this is my first attempt to build an amp, I would like to use an existing design. Possibly some * minor* mods.
I would like to build a decent sounding amp I can listen to. I generally like the sound of US amps. I think the Germans are way too analytical, and the Italians sound thin. These are of course generalizations, and open to misinterpretation if that's what you want to do. There are VERY good Italian, German and Scandinavian amps, but I'm just trying to give an indication of the kind of sound I'm looking for.
I'm not in a position yet to start building an amp from scratch. I'd like to construct something I can live with, an take it from there. Who know's what tangent I'll go off to in a few years. but now I need a good starting point, one that will have me able to listen to some music in a few months.

Peter R.

[geewhizbang]

The topic has diverged from somebody's earlier post about a Honda Civic being modified to Ferrari performance.

I have ridden my bicycle behind a bright red Ferrari at a stoplight once, and just the incredible SOUND alone is worth the cost of admission, if you have that kind of money. There is no way that you could modify a Civic to have that sort of sound. It would still be front-wheel drive, and it would still look like a breadbox on wheels.

A Porsche can't compare, even if it does go around a curve better than the Ferrari. The Ferrari looks and sounds like it is going faster.

*****

On the topic of amps, I really miss my old PS-Audio amp. If you opened up the amp blocks, you could see a very modest number of components. I had two 450 watt or so blocks (I was running them bridged). I saw one of these blocks go for $2000 on ebay recently. Like I said, I had two of them. I got it for about $200 used in 1985 or so.

Someone stole mine. So I'm designing my new system around a less demanding power requirement this time and it sounds better than the old one in many ways.

While I can't use my current system outdoors and get high sound levels, but it sounds really fabulous, detailed and clean indoors.

I am going to build a gainclone and see what it does. I used to frown on switching power amps, but given that Linn and other high end companies have no trouble asking very high prices for their implementations using the same modules, I will try to get over that inclination.

I can't afford the AKSA kit right now, but it looks like a good amp.

[Damon Hill]

Originally posted by LRRockBox

Since this is my first attempt to build an amp, I would like to use an existing design. Possibly some * minor* mods.
I would like to build a decent sounding amp I can listen to. I generally like the sound of US amps. I think the Germans are way too analytical, and the Italians sound thin. These are of course generalizations, and open to misinterpretation if that's what you want to do. There are VERY good Italian, German and Scandinavian amps, but I'm just trying to give an indication of the kind of sound I'm looking for.
I'm not in a position yet to start building an amp from scratch. I'd like to construct something I can live with, an take it from there. Who know's what tangent I'll go off to in a few years. but now I need a good starting point, one that will have me able to listen to some music in a few months.

I don't know of any source that provides a complete kit
to build a high quality amplifier of around 100-200 watts.
Most of these sources will provide at a minimum a PCB;
Seal can provide most of the major components. You'll
have to build it essentially from scratch no matter what,
unless you buy a used Hafler, for example, and replace
the electronics.

Consider a chassis design that would allow you to try
several different circuit designs; that means plastic
cased output transistors, which most any of these
designs listed below will work with.

some amplifier project sources, in no particular order:

http://users.ece.gatech.edu/~mleach/lowtim/

http://www.aksaonline.com/welcome.htm

http://www.sealelectronics.com/

http://sound.westhost.com/index.html

http://www.aussieamplifiers.com/

I'm certain there are other good sources, too.
You'll have to do some research and decide which
of these designs you like, and which might provide a
basis for a more powerful amplfier in a later project.
I think the Leach amplifier could be modified to as much
as twice the output power with three or four output
pairs and higher rail voltages. I have gotten many years of service out of the Leach amplfiers I have built and certainly
can recommend this design.

No matter what your choice, you'll find plenty of help
from the nuts at diyAudio.com!

[traderbam]

Nelson wrote: I have yet to see actual evidence that any audio manufacturer has actually had custom transistors made for them...While the quality of the parts is important, what you do with them is of far greater interest and consequence.

How quaint. No Farrari for you, then.

[Jennice]

LRRockBox,

After reading about Civic's, Mark Levison's etc. for pages on end, I'll give my try to get back to more hands-on basics...

I made my first, simple power-amp before I attended the engineering academy. It was basically a design from a fairly basic book on this topic. It still plays today, some 10 years after, in daily use. It's old-fashioned, simple, but everyone in the family likes it, and it sounds less sharp and analytical (probably just destortion or low bandwidth ) than most others, but it's pleasant to listen to. It's not dual-mono either (was intended to, but didn't end up like that). However, for it's wattage, it has a perversely over-dimensioned PSU and capacitor array, and it seems to just jump alive shockingly fast. VERY dynamic! It has single output transistors (complementary push-pull pair, no magic there), but without any (!) resistors to limit current. The Bias voltage-multiplier transistor is mounted directly on the back of one of the power devices to respond fast enough to ensure thermal stability.

My next project was the Pass A40 (but with 1.5 A bias!).
Slight instability problems (probably because there were too long leads to/from the power devices). Actually, I don't know if I really preferred the sound of it compared to the first one I did, even thougfh there was no doubt which one was most expensive to build and run (electricity isn't free here...). However, it didn't seem to mind what load I hooked on it. 8ohms (nominal) or 4 ohms, or 2 ohms... Very rugged thing!

The last (most recent one) I've built also has a relatively big PSU, and has about 4 * 80W in 8 ohms. It can be bridged to 2 ch, and is still very stable and powerful in 4 ohm. The test ran with power resistors as load, which boiled water (didn't have >100 W resistors on the shelf ) Explodingly dynamic IMHO, but a very normal design throughout. My experiences show that it's not the rated wattage you need, but available current, since a transient will reveal that your speaker can go to a much lower resistance value than the nominal 8 ohm or whatever it's rated.

Hence...*drum roll* (phew it took long to get there... sorry )

1) Find a fair and proven design (nothing revolutionary under the sun, as we've learned over the last 10 pages of postings)

2) Make a over-rated power-supply with all the storage capacitors you can fit (within reason, but 60.000 uF per channel is not unrealistic).

3) Don't be cheap on the output transistors. They'll take a lot of beating, so use ones with good power/thermal ratings, and use plenty of 'em.

4) Put some fused in the right places, but don't over-do it with safety-belt circuits that sense and adjust everywhere. They don't help the sound get better - on the contrary.

5) you don't need more watts than your speaker can handle. You "just" need the watts regardless of the conditions, and that's where the power supply comes into play.

6) KISS! (Keep it simple, stupid) Don't try to cram too much into one casing. It'll just interfere with neighbouring circuits. Mono blocks help ease this.

7) Pairing devices is good.. but if you have to put up priorities, go for better devices and less pairing. That's what the feedback will help with. *I will now expect to be killed by linearity freaks, but I'd prefere that to happen elsewhere*


Just my 2 cents...
Now go and have fun. You'll probably be surprised to see how far you'll come with modest budget.

Jennice

[carlosfm]

Originally posted by traderbam
Ferrari make their own parts because they need precision.

Sure.
That's why they use Momo wheels.
And Magnetti Marelli or Bosch electronic fuel injection.
No Weber carbs anynore.

[K-amps]

Originally posted by Nelson Pass


I have yet to see actual evidence that any audio manufacturer
has actually had custom transistors made for them. They
simply seem to have re-labeled them. None of the fabrication
houses are going to tool up for the volume of a high-end
company, and Harmon, who could conceivably command such
a service, apparently chooses not to.

While the quality of the parts is important, what you do with
them is of far greater interest and consequence.

Nelson,

I have read (more than once) on two separate interviews where Dan D'Agostino made statements about how they had custom parts ordered from Motorola. In the same interview he also mentioned that Motorola was not entertaining the idea of the Part he describes 20A 250v and fT of 30Mhz custom made for Krell so they said they would not do it for less than 1 million pieces... Dan agreed and they went ahead with the order.

Historically IMHO Krell and others have probably used re-labeled MJ15003/4 or MJ15022/23/24/25 variants in their amps.

But here is the deal, none of these devices does the above parameters. The 15003/5 will do 20A but not with a VCE of 250v. The 15024/25 will do 250v but at 16A not 20A, moreover none of these devices is 30Mhz capable, not even the newer MJ21193/4/5/6 parts that MOT/ON probably used to design the parts that Krell wanted. (The newer MJL4281 and 4302 do have an fT of 35mHz but they just came out now as opposed to when Krell got their devices in 1996.)

If I am to agree with your assessment, then Krell did come up with a pretty elaborate story that at least on the face of it ties out.




K-

[Nelson Pass]

Originally posted by K-amps
I have read (more than once) on two separate interviews where Dan D'Agostino made statements about how they had custom parts ordered from Motorola. In the same interview he also mentioned that Motorola was not entertaining the idea of the Part he describes 20A 250v and fT of 30Mhz custom made for Krell so they said they would not do it for less than 1 million pieces... Dan agreed and they went ahead with the order.

Historically IMHO Krell and others have probably used re-labeled MJ15003/4 or MJ15022/23/24/25 variants in their amps.

But here is the deal, none of these devices does the above parameters. The 15003/5 will do 20A but not with a VCE of 250v. The 15024/25 will do 250v but at 16A not 20A, moreover none of these devices is 30Mhz capable, not even the newer MJ21193/4/5/6 parts that MOT/ON probably used to design the parts that Krell wanted. (The newer MJL4281 and 4302 do have an fT of 35mHz but they just came out now as opposed to when Krell got their devices in 1996.)

If I am to agree with your assessment, then Krell did come up with a pretty elaborate story that at least on the face of it ties out.

It seems to me that your analysis says it all. It is not appropriate
for me to divulge the information I have, but suffice it to say
that the ex-president of Krell Digital is now the president of
Pass Labs (me - I just cash checks), and one of our best
engineers used to be one of their best.

[K-amps]

Originally posted by Nelson Pass


It seems to me that your analysis says it all. It is not appropriate
for me to divulge the information I have, but suffice it to say
that the ex-president of Krell Digital is now the president of
Pass Labs (me - I just cash checks), and one of our best
engineers used to be one of their best.

I will leave it that.

K-