ExtremA class A amp

[Piotr]

Many thanks for your input Jon and John.

And sorry to Jon for spelling your name wrong in my post earlier.


/Peter

[Johnloudb]

All the electronics experts are on this thread so I ask this here. I just completed a preamp, a variation on Pass' "Brideof Son of Zen" preamp. I'm taking noise measurements, with a DMM. It's has a resolution of 0.1mV (10 -16,000Hz). Anyway, what's bothering me is the noise level from the positive/neg. rails to ground (55mV - 65mV). From the pos. rail to the neg. rail it has < 0.1mV of noise. Mine has dual regulation and lots of filter capacitance/bypass caps. It this normal? :??

I've built preamps before but neglected to take this measurement. The hum and noise at the output is <0.1 mV for almost all volume control positions. So, the circuit has pretty good noise rejection. The way I see it noise has to go somewhere and it does - to ground.

Any thoughts?

Edit: OK, maybe I'm just being neurotic.

[Johnloudb]

OK, I took some more measurements with my 100MHz HP o'scope and the supply noise has only 2mV, at 20MHz+. So, I detected no reliable, measureable noise at any frequency. I'm happy with these results. I don't trust my DMM.

[Dennis H]

Jon, is this the leaded solder you recommend for PC boards? I googled a few of the Kester part numbers and it appears to be available online.

“331” Water-Soluble Core 1 lb.
Part # Alloy Diameter Core Size
24-6337-6422 Sn63Pb37 .015 66
24-6337-6401 Sn63Pb37 .020 66
24-6337-6417 Sn63Pb37 .025 66
24-6337-6403 Sn63Pb37 .031 66
24-6337-6411 Sn63Pb37 .062 66

What's your favorite diameter for PCB work?

[Jonasz]

Seems like a nice solution Jonasz!
Maybe a get-together in the future if you'r back in the land of the wikings? Would be fun to compare and perhaps listen in a "LTS aproved before/after test".

That would be very interesting indeed! I have family in Stockholm were I can stay so a get-together isn't impossible. Don't know about the LTS-thing though...

[Piotr]

I bought a small package of Kester '275' which I tried on some non sensitive parts yesterday. It was easier than I thought to use it. Not like the best lead solders but good enough and easy enough to use. It melts at a higher temperature, it also floats a little slower but still fill the hole and have good wetting properties.

'275' is 96.5Sn/3Ag/0.5Cu. I used a 0.8mm wire with 2.2% flux. Will probably buy a spole with 0.5mm or 0.6mm for smaller parts/holes.

/Peter

[Dennis H]

Jon says:

If you go with a "conventional" rosin core solder, I can't recommend Cardas Quad Eutectic highly enough- I've used a number of other silver bearing solders in the past, and the Quad Eutectic is formulated similar to what we use for chip die attach- the combination of lead, tin, copper and silver melts and hardens at just one temperature- it's practically impossible to get a cold solder joint, it flows great, and the flux cleans well but has relatively low residue for a rosin flux. I've been using it exclusively for speaker crossovers and point to point wiring the last five years.

Looks like he may have to switch to a new favorite solder for PCBs. I just stopped by the Cardas site and it sounds like they're offering a water-soluble core now. I don't know if Sonic Craft is stocking it yet but they have the best price on the rosin core stuff.

"QUAD EUTECTIC ROLL SOLDER
Ultra pure, tin/lead/silver/ copper, Quadeutectic .032 solder, activated rosin core or organic water base flux. Available in 1 lb. or 100 gr. rolls."

http://www.cardas.com/content.php?area=oem&pagestring2=Solder%20and%20Flux&pagestring=Solder%20and%20Flux+(closeup)&content_id=19&part_id=70

[Piotr]

O boy is it hard to go lead-free..

The problem is oxide formation on the tip.. it blackens in no time and makes the tip almost useless in no time.

I've collected information about lead free soldering and will try it some more but if it contineus like this I will go back to standard solder.

Anyway.. the powersupply boards are almost fully populated so the project is moving forward.

Any tips on lead free soldering and expecially oxide removal from the tips are greatly apreciated.


/Peter

[ThomasW]

I've collected information about lead free soldering and will try it some more but if it contineus like this I will go back to standard solder...... Any tips on lead free soldering and expecially oxide removal from the tips are greatly apreciated.

I'm with you I just stocked up buying 2 more 1 lbs rolls of Cardas Quad eutectic....(sourced from SonicCraft)

The best tip is forget about going lead free for DIY.

For the latest prototype projects Jon's building for Infineon, he's using the Quad-Eutetic since it flows so well for SMT devices

More and more testing is showing that for long term reliability Rohos isn't a good idea.

[Dennis H]

Is Jon using the new quad eutectic with the water-soluble flux for the PCBs?

[ThomasW]

Is Jon using the new quad eutectic with the water-soluble flux for the PCBs?

He didn't say. In the past he's only used the water-soluble solder for PCB work. Last conversation he just muttered something about the crappy lead-free solder caused him to go into his stash of Cardas to build a prototype board...

[Piotr]

About long term reliability.. I wonder if the problem is the LF solder itself or the fact that it is harder for untrained people to make a good solder with LF?

I've read something about LF solders done right cope with thermal and mechanical stress better than standard solder.

/Peter

[ThomasW]

About long term reliability.. I wonder if the problem is the LF solder itself or the fact that it is harder for untrained people to make a good solder with LF?

I've been told it becomes brittle and cracks. This is of course an issue for the military, aircraft mfgrs, and others.

[Piotr]

By using a salmiac stone and not using a damp sponge I have managed to maintain the wetability (sp?) of the tips. Overall I have to say that I have managed to get the leadfree thing to work pretty good now.


BTW, could someone have a look at the ExtremA circuit and tell me which of the resistors that are tied to VDRV+ and VDRV- that carrying dc and which are carrying ac?

http://www.hardwareanalysis.com/content/image/11997/

The thing is I have standard metalfilms at home but I'm thinking on using these only for non critical positions and use non magnetic precision types for feedback resistors and other that carry ac/music signal.

I think I have a pretty good grip but I'm not sure.


/Peter

[Victor]

Piotr,

Have you managed to finish the ExtremeA project? Please let us know!
Victor

[Piotr]

Piotr,

Have you managed to finish the ExtremeA project? Please let us know!
Victor

Hi!

Ohps, time flies.. hehe! No, I have not assembled the ExtremA's yet.


/Peter

[Victor]

Piotr,

I know, exactly what you mean...I also always am between 2 or 3 projects at any given time. At this moment I am looking at heat sinks from Fischer Electronics (SK-533 ) for my extremA amp.

I have 4 sets of PCBs on hand, so I'll be building 2 stereo amps. I have 2 old Bryston 4B ST boxes with all metal parts, an on/off switch and input/ouput connectors, - so it is a good start. THe old Bryston electronics is now a pile of circuit boards. It served me well over the years, but relative to ExtermA, Bryston is hopelessly outdated.

I am looking at Spring 2009 to fire them up.
Victor

[cotdt]

That's a very expensive and time-consuming amp to build, and it will run very very HOT. It might be hard just to find the Toshiba transistors. In my experience the Hypex UcD amps come pretty close already to the performance of high-end discrete Class A MOSFET amps, albeit a bit too polite in the treble. A good solid state or tube amp produce more realistic treble, but that's about it.

[Victor]

That's a very expensive and time-consuming amp to build, and it will run very very HOT. It might be hard just to find the Toshiba transistors. In my experience the Hypex UcD amps come pretty close already to the performance of high-end discrete Class A MOSFET amps, albeit a bit too polite in the treble. A good solid state or tube amp produce more realistic treble, but that's about it.

Well, - I totally disagree.

Is it expensive? It will cost approximately $400-500 per channel to build. Considering that you get an amp that outputs near 24-bits resolution, this cost is a pocket change.

The output transistors are available from several places, here http://www.ampslab.com/trans_2sc2922.htm for instance . So this is not a problem.

You cannot possibly or seriously compare this amp to UcD modules. You are talking apples and oranges here. UcD will favourably compare with most amps on the market and it is definitely the most efficient design yet to emerge that has acceptable THD+N specs. However, for those few of us that have speaker systems sporting performance in 16-bits territory, it makes more sense to have an amp that less noisy then UcD.

As for your comment that “…good solid state or tube amp produce more realistic treble…”, well again this is not so at all. You cannot objectively substabtiate this comment, period.

Show me another Power Amp that can run with modern 24-bit D/A converters!

Other then ExtremA I do not know of any commercially available or DIY designs. There is Cordell amp that can do this, but other then a prototype I saw no PCBs anywhere. I can think of perhaps 2 or 3 others, but again no available PCBs.

In my on practice I had to modify my Bryston amps to the level by far way beyond the current SST line to get 20-bits resolution out of it, - still the ExtremA is better and more elegant in every way.

Finally Class A amplifier does not have to be hot or excessively heavy; - it depends on how you build it. I intend to bias it to Class A up to 50 Watts, so the weight will be about 50-60 pounds.

Victor

[Dennis H]

....

[Amphiprion]

Is it expensive? It will cost approximately $400-500 per channel to build. Considering that you get an amp that outputs near 24-bits resolution, this cost is a pocket change.

You are saying this is a power amplifier with THD+N of -144dB? I've got a bridge I'd like to sell you dude.

[Victor]

You are saying this is a power amplifier with THD+N of -144dB? I've got a bridge I'd like to sell you dude.

THank you, - but let's play nice, - shall we? Here is the direct quote:

"...What can be read from this figure is that at 40W the true THD is 0.00008% (-122dB). A figure you will be hard pressed to find with any commercial amplifier. The nature of class-A distortion and its harmonic distribution allows us to extrapolate down to lower signal levels. At 1W the distortion would be around 0.000002% (-154dB) something which can’t be measured directly..."

See the write up at http://www.hardwareanalysis.com/cont...diy-amplifier/

I hope this will suffice.

[Amphiprion]

Sorry about my previous tone, it's just that I've spent my fair amount of time in front of an AP 2700 and claims of that kind of distortion set off some blinkers. Measuring it is difficult enough, which is why I wasn't surprised when I read:

The nature of class-A distortion and its harmonic distribution allows us to extrapolate down to lower signal levels. At 1W the distortion would be around 0.000002% (-154dB) something which can’t be measured directly.

The -122dB I might be able to buy, but does the extrapolation hold when we're talking about distortion levels this freakishly low?

[cotdt]

Victor, that amp when completed will cost at least $2000. The cost of the amp modules is only a small portion of the total cost. Once you start building it, you will realize this.

Yes it will have much lower background noise than the UcD modules, but the noise of the UcD modules are inaudible for speakers. Only if you put your ear right next to high efficiency speakers do you hear background noise from the UcD amps. In terms of sound, UcD sounds quite similar to the really high end Class A amps (though different sounding from euphonic designs like Pass Labs).

And the distortion of an amp is a nonissue. Even really cheap solid state amps have less than 0.01% harmonic distortion, which is far less than audible. There are much more important parameters that affect the sound of the amplifier, such as slew rate, output impedance, amount of global negative feedback, etc.

[Victor]

...The -122dB I might be able to buy, but does the extrapolation hold when we're talking about distortion levels this freakishly low?

Mark,

Clearly -144 dB is only of theoretical importance. Nobody here is suggesting that 24-bit precision can be reliably measured, let along taken advantage of in a real set-up. In order to do 24-bit measurements one needs nothing less than a Faraday Box.

I know that you know that if yours or mine signal chain can somehow output a 16 bit signal in the context of our HT Rooms, - we are sitting pretty!

The only reason to build this amp is if you want to have one of the very best, state of the art, available power amp driving those speakers you spend years to build. Otherwise the UcD-based deal from diycable.com will do nicely or one of those Emotiva amps will fit the bill just as well for a little less money and by far less effort.

Does the extrapolation hold? Mathematically speaking, - I would have to say yes, due to the inherent periodicity of the distortion mechanism evident in Class A power amps. Hmm...I hope I am right about this...I think I am.

Anyway, I pretty much gave up on my efforts to make my Bryston ST into an even 18 bit design to be on par with the rest of my signal chain. And I need at least 4 channels for my front speakers. The other DIY power amps that I have are current feedback types good to nearly 18 bits, but I only have 2 channels of those. The rest of my DIY amps doing an honest 16 bits.

I was hoping to get the Bryston basic design to where I want it to be and then get more of them used for my HT and modify them. Now, with ExtremeA in the picture I will finally have enough channels of truly high resolution amplification.

Victor

[Victor]

Victor, that amp when completed will cost at least $2000. The cost of the amp modules is only a small portion of the total cost. Once you start building it, you will realize this.

Even if you count the metal work the total cost will not exceed $700 per channel. In my case I already have the box, the caps and transformers, so my cost is less.

Yes it will have much lower background noise than the UcD modules, but the noise of the UcD modules are inaudible for speakers. Only if you put your ear right next to high efficiency speakers do you hear background noise from the UcD amps. In terms of sound, UcD sounds quite similar to the really high end Class A amps (though different sounding from euphonic designs like Pass Labs).

I know that UcD amps are quite enough for nearly all applications. This is not an issue here. I also know from experience as well as theory that no discernable sound can be attributed to UcD modules. For all intends and purposes, I think, that UcD design is transparent enough. Still this is not an issue here.

I am building ExtremA because objectively speaking the ExtremA design is a better amp on all grounds for a given power bandwidth. I have no illusions regarding my, or anyone else’s, ability to discern ExtremA from UcD in operation, - I know I cannot. In fact nobody can.

And the distortion of an amp is a nonissue. Even really cheap solid state amps have less than 0.01% harmonic distortion, which is far less than audible. There are much more important parameters that affect the sound of the amplifier, such as slew rate, output impedance, amount of global negative feedback, etc.

Indeed even a chip-based amp will do well these days. Nearly any chip amp will give you better slew-rate, better damping factor or as you put it output impedance, will have flatter pass-band and even better THD numbers them most high-priced commercial amps.

Does all this mean that IC-based amps are a way to go?

If you compare IC-amps or UcD-based amps to most power amps you can buy in the store, then the answer is yes. However, ExtremA is clearly better on all those counts within the guidelines that actually matter. For instance, I really do not care for 500 kHz capable amp. I used to though. I also do not care for damping factor better than 100 dB, - I only use active crossover these days. Global feedback is a none-issue from engineering perspective, since we are assuming properly designed amps here.

My point is that well-designed power amps do not have a discernable sound, -so we build them or buy them for other reasons. An engineer in me says that my full-range 3-way dipoles deserve the best, - so I go with ExtremA.

Victor

[Piotr]

Victor, that amp when completed will cost at least $2000. The cost of the amp modules is only a small portion of the total cost. Once you start building it, you will realize this.

I payed half of that for my pair with powersupply. Sinks and chassis of old stuff I had allready.

Yes it will have much lower background noise than the UcD modules, but the noise of the UcD modules are inaudible for speakers. Only if you put your ear right next to high efficiency speakers do you hear background noise from the UcD amps.

Well that depends on the voltage sensitivity of your speakers. Common numbers are all the way fromm 80dB to 100dB for 2.83V.

In terms of sound, UcD sounds quite similar to the really high end Class A amps (though different sounding from euphonic designs like Pass Labs).

Why comparing a transmission link to another one? Compare the output with the input instead, that's what interesting for hifi use.

And the distortion of an amp is a nonissue. Even really cheap solid state amps have less than 0.01% harmonic distortion, which is far less than audible.

At 1kHz and 1W into 8ohm resistive eh?

Read up on intermodulation (IMD) my friend! :-) Almost NO amps are transparent and those that are measures low on HD, not only at 1k but thru all the audio spectra.

There are much more important parameters that affect the sound of the amplifier, such as slew rate,

Please.. what do you think inadequate slewrate results in? Hint.. harmonic and intermodulation distortion on high frequencies and/or high levels/transients.

output impedance,

Output impedance is not a quality parameter but is something that needs to be balanced against the load.

amount of global negative feedback, etc.

And what about that?


/Peter

[cotdt]

Piotr, thanks for confirming that the final cost of the amp (properly built) costs around $2000. You already had the tools, chassis and heatsinks, and it still costed you $1000. Chassis and heatsinks cost about $300 each, you need two of them for monoblock operation, and custom labelled/engraved panels are another $150 each. I like to do things the right way, use overspec'ed parts, and make the casing aesthetically pleasing. I built a similar amp as the ExtremA as an integrated, with a 4-channel remote control volume pot that ended up adding another $400 by itself. And at this cost, most people will go for expensive boutique components whether they make a difference or not, adding further to the cost.

You make all these claims about IMD, etc., which amongst the better amps still seem quite low to me. Anyway, I never cared for measurements beyond the ones used to design speaker crossovers, it's all about the final sound. And I bet most people won't be able the tell the difference in a blind test between a UcD and a discrete Class A amp that measures reasonably well.

[Jonasz]

Isn't $2000 pretty cheap for an amp with ExtremA's performance? That last percentage of performance is always the most expensive.

[mazurek]

I'm interested to hear when you finish the amp. I'm a believer in aspiring to a minimum level of quality electronics, especially after having heard and measured some. Also, taking time to understand some projects, even expensive ones, can offer more bang for the buck than some university classes.

[Piotr]

cotdt,

I dont understand what you are wining about. You say that these amps are expensive and time consuming to build. So what you want, you want us to ditch them and walk your way?

You want us to go the McDonbalds and buy a burger? After all that is cheap and quick.

If you want gold plated 1" thick alu-front with your high-school nick laser engraved on it and nude vishays in your amps that is up to you. I spend my money on audioperformance.. jeez!

You talk about doing the right thing and using overspeced parts.. well that's nothing but stupid in my book. Using overspeced parts is bad engineering, but hey, that would be up to you but I still don't get what you are complaining about.

I built a similar amp as the ExtremA as an integrated

Yea, I'm sure you did..

You make all these claims about IMD, etc.,

Claims..!?

which amongst the better amps still seem quite low to me.

Well I guesse you are entidled of an opinion even though you do not have understanding of the subject.

Anyway, I never cared for measurements beyond the ones used to design speaker crossovers, it's all about the final sound.

Yes final sound. And audio gear has a direct relationship between sound and its intrinsic physicall behaviour.

And I bet most people won't be able the tell the difference in a blind test between a UcD and a discrete Class A amp that measures reasonably well.

You may bet all you want. A well executed class A/B amp outperform all known switching amps in the highest octaves. It's very very hard to find audio gear that is transparent and even though it is possible to find gear that has identical colorations it's not very likely in the end. This should be read as most audio gear can and will be picked out in blind levelmatched tests.



/Peter

[Piotr]

Isn't $2000 pretty cheap for an amp with ExtremA's performance? That last percentage of performance is always the most expensive.

Yes and to the best of my knowledge ExtremA is the most linear amp in existence.

BTW, did you build your FET amps?


/Peter

[Jonasz]

Yes and to the best of my knowledge ExtremA is the most linear amp in existence.

BTW, did you build your FET amps?


/Peter

Nope, not yet... ops:

Got speakers to finish also... ops:

I've been veeery lazy... ops: ops: :evil:

[hartono]

Yes and to the best of my knowledge ExtremA is the most linear amp in existence.

BTW, did you build your FET amps?


/Peter

most linear........errr

how about Halcro
or if you prefer DIY :

http://home.tiscali.nl/audio/

[wettou]

I would like to build a Class A amplifier what do I need and waht could be the cost of such a project.

200W at 8Ω doubling to 400W at 4Ω and 800W at 2Ω in pure Class A

Similar to pass labs 200.5

[Victor]

most linear........errr

how about Halcro
or if you prefer DIY :
http://home.tiscali.nl/audio/

Indeed the amp you referenced is one of the best around. The problem is that the amp is essentially a vaporware, - that is to say you cannot get the PCBs. And making your own boards although possible introduces the variables that I am not willing to live with, because the end result is far from guaranteed.

The amp is otherwise is an excellent work and it shows that in order to achieve 0.0001% THD one needs to employ error correction schemes and common feedback, nested differential feedback, etc. Those are complicated circuit tricks to get right, but if you can get them right, then you will get an exemplary THD.

The ExtremA amp, on the other hand, is real, i.e. the boards are available, and its performance is only fractionally worst then the amp you referenced. In fact at full power Extreme A is only 3 times worst then the amp you referenced. We are truly arguing semantics here.

One more issue to look at, - the ExtremA design is fully balanced, while the other amp is not. For my purposes fully differential topology is important because I intend to have a fully balanced signal processing chain from Blue Ray player all the way to speakers.

So, in the end Peter is correct, - ExtremA amp is most linear amp in existence. I am now in the process of building mine having finalized the heatsinks and I also will be getting a custom power transformer with all the secondaries in one package.

regards,
Victor

[Victor]

I would like to build a Class A amplifier what do I need and waht could be the cost of such a project.

200W at 8Ω doubling to 400W at 4Ω and 800W at 2Ω in pure Class A

Similar to pass labs 200.5

Class A 200 W into 8 Ohm is a monster! Continuous output of 800 Watt into 2 Ohm while maintaining Class A is more than a simple monster, - it is a Godzilla with jet engines on steroids! Do not do it!Let along I do not think it is actually possible in the context of commonly used circuit breakers.

200 Watts into 8 Ohm is possible, but anyway it will take 2 or 3 very strong man to lift it. You are talking about 400 watts stand by dissipation, leading to a 1 Kilowatt capable transformer and the heatsinks size of desktop computer. The heat management will be very difficult and it may negate whatever benefits the Class A will bring you.

In my opinion, 100 Watts into 8 Ohm is an absolute maximum that the Class A amp should deliver. In fact I am thinking of making my ExtremA amp 75 Watt capable in Class A operation.

Victor

[wettou]

Class A 200 W into 8 Ohm is a monster! Continuous output of 800 Watt into 2 Ohm while maintaining Class A is more than a simple monster, - it is a Godzilla with jet engines on steroids! Do not do it!Let along I do not think it is actually possible in the context of commonly used circuit breakers.

200 Watts into 8 Ohm is possible, but anyway it will take 2 or 3 very strong man to lift it. You are talking about 400 watts stand by dissipation, leading to a 1 Kilowatt capable transformer and the heatsinks size of desktop computer. The heat management will be very difficult and it may negate whatever benefits the Class A will bring you.

In my opinion, 100 Watts into 8 Ohm is an absolute maximum that the Class A amp should deliver. In fact I am thinking of making my ExtremA amp 75 Watt capable in Class A operation.

Victor

Pass Labs has such a beast it is just $17,000 each?

[BobEllis]

Actually the XA200.5 leaves Clas A at 400W into 8R.

To get to 200W/8R you'll need about 70V rails. With a push-pull output, you'll want to have 10A bias to stay in class A at 800W/2R. You'll have 1400W idle power per channel. Each channel will need twice the heat sink available in the large Pass chassis. You'll need a 5-8KVA transformer per channel. If the X1000.5 is 220 pounds, you're probably looking at 350 pounds 19"w x 15"h x 50"d monoblocks. With that heat, the AC will run constantly, even in winter. You'll need a dedicated 20A circuit for each channel....

[Piotr]

most linear........errr

how about Halcro
or if you prefer DIY :
http://home.tiscali.nl/audio/

From Hardware analysis on ExtremA:

Next up is the residual THD at 40W/1kHz. The 2nd harmonic is due to the fact we have not used Hfe matched transistors throughout. What can be read from this figure is that at 40W the true THD is 0.00008% (-122dB). A figure you will be hard pressed to find with any commercial amplifier. The nature of class-A distortion and its harmonic distribution allows us to extrapolate down to lower signal levels. At 1W the distortion would be around 0.000002% (-154dB) something which can’t be measured directly.

From tiscali.nl:

Although the average THD at 1W is slightly over 1ppm, we are quite happy with the overall results. Our amp beat the 0.0001% (1ppm) distortion barrier, at all frequencies and (almost) all output levels! The distortions are essentially 2nd harmonic, although a 3rd harmonic component is also contributing to the THD. All other harmonics can be neglected in a quantitative analysis.

The ExtremA seems to have a distortion level of about 1/100 of the tiscali amp at 1W.

You may want to check the THD+N graphs of Halcro DM58:

http://www.stereophile.com/solidpoweramps/683/index7.html

And compare to a similar graph of ExtremA:

http://www.hardwareanalysis.com/content/image/12004/

And come to the conclusion that up to its power limits the ExtremA measures better. One of the Halcro amps was also found to be non-transparent in blind controlled listening tests in Sweden. My goal is for an amp that is audibly transparent.


According to all documents I have found on the internet ExtremA is the low distortion champ.

Anything else I can help you with?



/Peter

[lgreen]

ExtremA

Hello there. I am new here but I did see you are discussing ExtremA.

I have built one and can report that the ExtremA sounds very good; until it clips. Mine has some wierd oscillation at clipping.

My Extrema

Has anyone here built one? If so please email me.

[villastrangiato]

I'm a little skeptical about the need for a class A output bias in this design. I learned a long time ago that ultra wide bandwidth, low switching noise, extremely low intermodulation distortion, and high speed are not necessarily desirable attributes for an audio amp. After owning one of the first truly high speed Kenwood 150 wpc amps from 1980 to 1990 - I can say with certainty that what little benefit you get from these rare and expensive Sanken output devices and low capacitance feedback - you pay with instability and a tendency for self destruction when a driver stage component goes bad after enduring years of high levels of heat. The amp I had was an A,AB design that always ran hot and after ten years, rapidly consumed itself. Although it was very quiet (undetectable noise floor >100db) and exhibited very flat response, I can honestly say I never noticed any real difference in sound between that unit and other "slower" amps I've since used. So I have real doubts about the usefulness of an amp that has a 500khz 3db bandwidth, 0001% intermod and THD, a rise time less than a microsecond, and a slew rate of 200 V/microsecond. After living with one of these Kenwood/Accuphase type amps for many years, I'd rather have an amp with better current capability, headroom, and damping factor that stays clean with low impedance loads. To me, these are more "real world" attributes that make a difference I can hear. I guess what I'm trying to say is - everything in moderation. That would include speed, negative feedback, linearity inducing techniques...etc....If I were building a Hypex amp, I'd be more inclined to agree with Cotdt and build it with an output biasing configuration that is more consistent with the overall Class D approach to efficiency and performance. Without delving too deeply in the feedback circuit's bandwidth limitations, the posted Hypex data indicates a fairly narrow power bandwidth - so I'm not sure whether or not the "super duper" output configuration is made useless by limitations further up the chain - the specs suggest that might be the case.

[wettou]

So let me ask this

What would be best using a Pass Labs 100W or Classe 200W?