Thursday, May 21, 2015

When is Class A "Real"

OK, after much heart wrenching deliberation, I'm planning to keep the Krell FPB 300.  I will get it fixed.

I'd really decided this on my own, as I was taking pictures of the Beauty it all came back to me.  But I've been reading blogs to shore up my decision.

And one of the things that ticks me off is how glibly people declare that their KSA-250 or whatever is a True Class A amp, and the Sustained Plateau Bias amps like the FPB 300 (mine), the 300c, and the 400cx are Not True Class A.

People will even gloat about their KSA-250 "250 Class A watts into 8 ohms, 500 Class A watts into 4 ohms, 1000 Class A watts into 2 ohms", etc.

But THEY are the ones who have it all wrong.  Actually the FPB 300 is far more of a True Class A amp than the KSA-250.

John Atkinson did the calculations long ago based on the bias current.  The KSA-250 puts out 25 watts of Class A power into 8 ohms.  So indeed it is a True Class A amplifier if you never use more than 25 watts and your impedance is no less than 8 ohms.  But that's not how people think about it, they're thinking 250/500/1000.

Furthermore, if the amp puts out 25 watts of Class A power into 8 ohms, and that's limited by the bias current and not the max voltage, then it necessarily puts out only 12.5 watts of Class A power into 4 ohms, 6.25 watts of Class A power into 2 ohms, and 3.125 watts of Class A power into 2 ohms.  And we're talking about a monstrous 180 pound amplifier.

Correctly speaking, the KSA-250 is a High Bias Class AB Amplifier.  The FPB 300 is a Sustained Plateau Bias Class A amplifier.  It is always and by necessity operating its output stage in Class A, even if the Class A Bias level is continually changing.

Operating the output stage in Class A (as compared to Class AB) has certain key advantages.  For one, it permits the amplifier to operate well without feedback on the output stage.  Class AB amps cannot do that.

Operating without feedback means that reactive loads cannot induce ringing or similar interactions with the amplifier.  I think this is key for me, as my electrostatic speakers are a very reactive load.

Sorry but at the end of the day, I see nothing especially interesting about large Class AB amplifiers, and notably the Krell ones.  If I were to get a large Class AB amplifier, I'd likely get a JC1.  John Curl knows that game as well as anyone, and even his lowest priced amplifiers are excellent designs for their cost.

Sadly I feel the same way about Dan D'Agostino's new Momentum amplifiers.  Just another large Class AB amplifier, as far as I can tell from published technical info (which may well be the limitation of the published info which may be limited for trademark reasons…and the 1W low idle power suggests a managed bias system of some kind…and also makes one wonder if Dan isn't seeking to reverse his earlier crimes).  It may be a fine amplifier…but I have no information that technically says it would be better on my 1+1's and one little factoid that suggests it might be worse…it appears to have "some" feedback.  The FPB 300 is claimed to have no loop feedback around the output stage.

If the Momentum isn't a Class A amp, I don't know of any reason to not prefer a pair of JC1's (which have much better and more published specs), except for the copper chassis (which I think is cool, but not so much the meter).  And neither of those amps qualifies like the FPB on the Zero Feedback score.

Given the most reactive load of all, something like the FPB with zero feedback is called for.  And in fact the FPB sounds terrific on my Acoustats.  All the edginess and harshness is gone, but the detail and layering increases.  The FPB is unique among all other amplifiers I have tried in this way.  Some people describe the FPB as sounding "dark."  What I feel is that other amplifiers on my 1+1's have glare, and the FPB uniquely removes that.  I'm not sure if any other amplifier would do as well, at any cost.  So I'd better keep it and repair it until I find something else as good…there might not be…though I don't have any reason to believe a 300c or 400cx wouldn't be still better.  Though on that score, I wouldn't think the improvement to be all that necessary from those even pricier models, since those don't especially have to do with the zero feedback output stage (which remains about the same, though the 400cx increases the available power…mainly by programming it in I've heard…it was just limited by the microprocessor).  And there is no way in which the FPB 300 is a slouch in it's voltage amplification given the low distortion, low noise, and bandwidth.  I know Martin Colloms rates the improvements in each Krell model as having been monumental--even more monumental than the first.  But I think this is hype.  The FPB was a new and better kind of amp for power sucking super capacitive Acoustat speakers.  After that the company made some improvements in it's input amplification and power programming, and these improvements would be nice to have, but no way do they compare with the initial advance.  And then the FBP series of amplifier were replaced with EVO, which appears entirely different (and I'm not sure if useful to me), and now those may not even be available as a new product.

But beyond the little world of FPB and it's c and cx descendants, I don't know of any other amplifiers that would serve my needs as well, despite costing many thousands of dollars more, and the shockingly high cost of the FPB amplifiers when it was introduced is now dwarfed by the charge for any audiophile blessed tin can.

Meanwhile, the Sustained Plateau Bias is not the only kind of variable bias system devised.  James Bongiorno patented the Sliding Bias Class A used in his Sumo Nine amplifier.  That amplifier has always sound intriguing to me, I've often lusted for one, but the fan makes it useless for me.  If only he had built it out with heatsinks into something like the fan-less FPB 300.

Early Threshold amplifiers also use a kind of trick Class A bias.  Back when I worked at a store selling them, it was often called "sliding bias" but that was not exactly correct.  James Bongiorno got the patent for sliding bias, and that was after the Threshold amps had been made.  I believe the Threshold amps clamp the transistors so that they always remain in the ON state.  Well that's better than turning on and off as most Class AB amps do, but it's not gaining the linearity advantage even of sliding or plateau bias, and I'd doubt it could be used without feedback either.

Sadly the Sliding Bias invented by Bongiorno may have hardly been used, precisely because it was patented.  Bongiorno himself had bad luck running audio companies, and others didn't want to use something that might be restricted or that they might have to pay royalties on.  So this may be another case of how patents slow progress rather than promote progress.

Sliding Bias might even work better than Plateau Bias, though the latter sounds more sexy.  Krell got it's own patent on Sustained Plateau Bias.

Now if your whole thing is electronic constancy, simplicity, etc., then Sustained Plateau Bias is going to sound like pure evil, and nothing like Class A normally does.  But all the neo- Class A designs still have the key benefit of making the output more linear before feedback, and/or with all the devices always in the on state and ready to respond to reacting loads.  And those are objectively measurable aspects of behavior.  The benefits, if any, of simplicity on sonic performance--it's possible it's there but it has absolutely no backing in objective audio science today, given similar outcomes in THD (better would be THD weighted for harmonic detectability) it wouldn't matter so much whether they derived from a simpler process or a more complex one.  As much as I'd love to have a Pass Amp (though I'd need at least a pair of XA 200.8's for my Acoustat 1+1's, go figure) nothing of the thinking behind their special circuit simplicity has not been confirmed by DBT.  DBT says you might as well just have another full-feedback Class AB amp.  That would apply to me too, but I wouldn't know if they tested Acoustats--I think there's a better chance of amplifier feedback audibility with the Acoustats than just about any other speaker.  But there's no reason to believe the simplicity thing applies more to Acoustats than other speakers, at least other speakers which are about equally transparent or reasonably close.

(LOVE Pass Amps anyway, love isn't science, but obviously a pair of XA200.8's is out of the question.  I'm wondering what kind of Class A amp to get to replace the HCA-1000A on the super tweeters.  Several of the First Watt amps would do, F5, J2.  Look at what they cost!  And now Pass is not freely giving out schematics on some models.  Certainly the old Nelson way was nicer.  However I can't complain as it appears the F5 build information still appears to be available.  I think I'd build an amp before paying $5000 for a 25 watt amp.  $5000 for a 1600 watt Class A amp (used FPB 400cx in 2 ohms--which is Acoustats at 20k) wouldn't be bad, if that's all you pay.  But no good having 1600 watt amp on the supertweeters.  But would I ever get around to building a Class A amp?  The HCA-1000A is an excellent amp already, surprisingly hard to beat (and DBT would say you can't) but I'd love to have a small solid state class A amp, a field which is actually fairly limited, First Watt, Sugden, Pioneer Series 20 M-22, Pass Labs Aleph, Early Krell, Mark Levinson ML2.  I think the best of the First Watt designs are head and shoulders above the rest, and with modern parts.  Arc welding and Acoustats may require a different strategy.)

I think amplifiers of the sliding bias/plateau bias types are needed for very inefficient but very resolving loudspeakers, and it's unfortunate the drive to simplicity--which I concede has benefits--has relegated such things and thinking of them to obscurity and worse.

One other idea I've had is the construction of the FPB and subsequent amps marked a clear break from the past (or it might have occurred slightly earlier).  The earliest Krell amplifiers have a slightly casual internal construction.  The later ones look mil spec.  Now the FPB 300 looks very professional inside..but also lacks ostentatious audiophile parts and fancy construction techniques.  It may be multilayered board and it is quality looking for sure but it's not a special engineered plastic like used in Mark Levinson's top gear.)

You expect the FPB 300 to behave like a monster but it behaves like a kitten.  There's no drama on turn on or anything else, unless drama is demanded at the input jacks.

An example of a rare if not unique Class A amplifier whose Class A power doubles down into lower impedances is the Mark Levinson ML2.  This is a 25W (into 8 ohms) which dissipates 400W at idle.  The 400W idle suggests 100W Class A.  Indeed the rated power is 50W at 4 ohms and 100W into 2 ohms.  It's possible it has 100W of Class A power into 8 ohms.  But remember it only has 25W (rated power anyway) into 8 ohms and weighs 65 pounds AND DISSIPATES 400 W !  Part of that dissipation is because the output stage has regulated power, very rare in audio amplifiers, but that set includes the FPB amplifiers.

If your amplifier is going to have 1000W of Class A power at 2 ohms it will be needing to dissipate between 3000W and 4000W at idle.

No comments:

Post a Comment