Ditching Class A Amps due to Heat - Sort of a Poll

in the sense that it does what it’s designed to do, prevent notch distortion (the basic reason we want Class A) while reducing power consumption.

@axo1989 

Just so you know, any class D amp that uses a choke-filtered output (which is nearly all of them) are incapable of notch or crossover distortion.

I read Krell Solo 575 Mono Power Amplifier 575 Watt Amplifier features 'iBias'.  This is said 'to deliver the rich musicality of Class A amplifiers, the uncompromised dynamics of classic Krell amplifiers, and the efficiency and low power consumption of Class G and H amplifiers.'  Sorry.  I don't believe you can get something for nothing.  In their publicity Krell extol the virtues of Class A and pretend they are still building it. 

@clearthinker

Krell has used a sliding class A bias system for years. The way it works is at lower signal levels the bias is also reduced. As the amplifier power increases, the bias linearly increases with it. In this way it can be biased in the A region all the time (in this case meaning both output transistor banks are active throughout the entire signal waveform) at any power level the amp makes. Because music tends to have lots of transients that are short duration, even though the amp might be quite powerful the net result is it will run cooler and draw less power, significantly so.

If you were to run it at full power long enough for it to heat up, you would find it making the same heat and drawing the same power as any class A amp capable of the same power.

My point here is that its not a 'get something for nothing' proposition as you suggest. Seems to me there was a patent issued on this technology back in the 1990s.

The heat generated from a class A tube amp is significant and can be annoying.  I have a Black Ice Audio Fusion and a Cary tube amp.  I can only run them in the cooler late autumn through the winter seasons.  The rest of the year I run bridged Schiit Aegirs and Outlaw Audio 2200 monoblocks.  For class A, the electrical cost is not the issue; it's the heat.  If the bias current is not spot on, the amps will overheat and shut down.  The Black Ice uses four 6550s and the Cary uses eight KT88s.  Yes the 6550s are the functional equivalent of oven coils although KT88s are slightly cooler.  This may be sacrilege, but if I had to do it all over again, I would go with class D.  My Ampeg SVT tube amp runs nowhere as hot as these stereo amps.

@atmasphere thanks, that post was informative (both class D and Krell bias discussion). As I mentioned, I’d like to hear a Purifi class D design (either 1ET400 or 1ET7040 based).

I’m still happy to buy a Krell Evolution e-series though, I see it as the most technically advanced implementation of that design lineage. I can live with 2 watt standby. A (hypothetical) pair of 400 watt monos running 24/7 in old-school class A (no sliding bias) would be a madness though, as a main system left on all the time, consuming its rated power (multiplied by the efficiency factor) but achieving nothing when idle. That would be nothing for something.

@atmasphere 

You've been taken in like all the other punters.

The Krell sliding bias system and its predecessors are cons.

The orchestra is noodling and so is the bias on the Krell.

Suddenly the guy at the back whacks the six foot drum for all he's worth.

The guy in the amp says 'OMIGOD the guy in the orchestra has whacked that drum.  For *****sake get that bias up immediately.'  The other guy in the amp on the bias hears the msg and turns up the wick.  But the guy in the orchestra on the big drum has already gone home.

YOU CANNOT UP THE BIAS QUICK ENOUGH TO CATCH THE MOMENT.

I have read the patent and it does not answer this.  It cannot.  I asked a senior engineer working on my KRS200s how it solves this issue and he could not answer.

This started as a con and remains a con.  The bias cannot be changed quickly enough.  The only way to do it would be to buffer the signal for a second or so which would introduce insoluable clock problems.

YOU CANNOT UP THE BIAS QUICK ENOUGH TO CATCH THE MOMENT.

I have read the patent and it does not answer this. It cannot. I asked a senior engineer working on my KRS200s how it solves this issue and he could not answer.

@clearthink You can easily change the bias no worries.

I don’t think they want to publish the trade secret which is why its not in the patent. You are correct that you do have to be careful that the timing of everything works out. That means that you have to have the same bandwidth in both the actual audio circuit and the circuit driving the bias so that they are in fact properly time aligned (IOW no phase shift between the two at any audio frequency). You can’t have any timing constants to filter noise! It simply has to be as reactive as the audio circuit itself.

I think Krell did that- its not hype at all. The problem is that this type of circuit by definition has to introduce distortion of its own. That, ultimately, is why we never implemented it in our amps.