A U D I O    P A P E R

The TABOO of Amplifier Design

Jan 2005 by Steve Deckert


Some how I seem to frequently fine myself exploring the dark side of analogue circuit design. (Not usually on purpose either) I find it fascinating and challenging to twist the "rules" to the edge of operational stability just to see what happens.  The most recent distraction of this nature began in December 2004.  

It started when Paul (the instigator) bugged Dave to make him another amp out of spare parts laying around his shop.  Dave is heavy into his own guitar amp business now and together they rummaged through his shop to find an old original Zen amp chassis, a couple output transformers and even an old original power transformer.  This was perfect for Paul because his ongoing experiment in his basement was about one zen amp short of being one zen amp richer.  I don't know how many he has down there, but frequently he will hook up as many as 20 speakers at the same time.

They had all the parts in front of them to toss together another SE84C but decided to wire it in Pentode rather than in Triode.  As with all of Paul's new toys, it made it's way over to our studio for a listening test.  In the rush to get it done they decided to try leaving the negative feedback out or just forgot to add it I'm not sure... but the end result sounded like a giant tweeter.  

Not having a terribly good debut at my place, he then took it over to Dennis's house for the official "Dennis Test" and hoped for better luck over there. Desperately rolling tubes trying to get it listenable it continued to show up both here and at Dennis's house repeatedly over the next month.  While it had improved somewhat and did some things very interestingly good, it still remained unlistenable.

Needing an escape on night from the bazillion things I should be doing, and curious why Paul's new little amp just wasn't happening I snuck it over to the shop to see what was going on.  It's always fascinating to hear a different amplifier and puzzle over why it sounds the way it does.  This one was no exception, because we rarely hear something this far off for any length of time.  Basically the frequency balance was the only thing I could fault with it as everything else it was doing was good.  

I've always liked the sound of Pentodes but Triodes are far easier to get a good transparent sound out of.  The difference between them is linearity.  The Triode output device is usually very linear making it possible to run with no feedback. The Pentodes amplification efficiency increases with frequency so the midrange is louder than the bass and the treble is louder than the midrange.  To get it flat some manipulation must take place.  

The most common way to manipulate the frequency balance of the Pentode is with global negative (degenerative) feedback.  By applying a sample of the output from the amplifier back into the input stage out-of-phase you can achieve a flat response. Since the natural tenancy of the Pentode is to amplify high frequencies more than low frequencies, the out-of-phase signal that comes back from the output stage will be stronger at high frequencies thus creating more cancellation at the input stage at those high frequencies.

While this fixes the problem, when compared to a Triode with no feedback it usually lacks depth and transparency.  However, as you might imagine, since the Pentode is more efficient, has more power with a natural tenancy to become still more efficient as the frequencies increase it is also a subjectively "faster" device.  It is capable of adding a presence to the midrange and treble (because of this speed) that is less apparent in Triodes.  This natural ability of Pentodes would normally make the result more transparent than triodes but that transparency is often undone by the added parts in the circuit needed to manipulate the response so very few Pentode amps actually sound great when compared to Triodes.  Given both types expertly implemented, the Triodes are generally smoother, have more warmth, and tend to be more forgiving and musical.  However, the flip side is that the Pentode can be largely more honest, faster, and have more power.  Bass character on a great Pentode amp, especially push-pull is still a good reference or benchmark.  A 100 watt PP Pentode amp can make the bass of even the best solid state amps sound bad.

So, back to the story -  I tried to add feedback to Paul's little pentode amp but it screamed loudly every time I tried.  I noticed that the output transformers were wired backwards at the primaries making the grounded secondary out-of-phase so adding feedback simply created a perpetual oscillator.  Rather than fixing it right, I elected to simply unground the secondary of the output transformers and let them float.  This is where the Taboo started.

In the Zen Triodes I always float the transformers so they have no reference to ground.  Because of the extreme stability and simplicity of the circuit design in a Zen Triode floating the transformers adds a large degree of magic to the signature of these amps.  In particular the soundstage and imaging become more surreal.  Trying this in a Pentode amp or any amp with feedback is asking for trouble.  For one thing, whatever feedback scheme you design for the amp, ie., how many dB of feedback and what type of RC network you might arrange will completely change without the ground.  

After a few hours, I got a pretty decent signature out of the amp with what ended up being positive (regenerative) global feedback.  The following evening Paul showed up and I explained that Dennis and I couldn't take listening to his amp anymore so we gave it an enema and here it is!   As we listened a couple things became apparent.  It was fun to see the little Zen amp (visually identical) get so amazingly loud compared to it's triode counterpart.  It was also apparent that this new little amp had the spank and speed that when combined with the added power came together to add a level of excitement to most of the music we listened to.  Paul's comment was that it has some of the signature of all the amps we've made, the Torii being the closest.  But even the Torii didn't have quite the spank of this little amp?

A few weeks pass during which time we continue evaluating Paul's little amp.  It still wasn't perfect.  The Zen Triodes did several things better in the midrange.  But I have to admit that what Paul's amp did right was really starting to get me worked up.  He finally took it back home and in the days that followed I found I really missed listening to it.  So back to the shop to make another one.  Of course building a second from scratch allowed me to make some overall changes that I hoped would be improvements.  I got it done, and while almost exactly the same amp, it failed to sound like what I remembered Paul's amp sounding like.  It was however stone quiet now, and had adjustable feedback so you could tweak the tone.  

I worked on voicing the feedback network for hours at a time whenever I had a spare moment until I had probably 30 hours of time into it and had tried a couple dozen different feedback designs.  By this point I was becoming very happy with it and felt that it did many things better than Paul's.  We listened to it for the next week or so and it's potential became intoxicating.  I let it become an obsession for awhile and began to completely explore the range of possibilities it may have hidden inside.  

To keep the signal path clean, all the work was done in the feedback loop.  Around a hundred man hours later I found myself starting to spin my wheels.  The stability of the amp was perhaps the main factor since it had floating outputs with feedback.  Along the way I tried negative feedback schemes, positive feedback schemes, and even developed one that had both positive and negative feedback at the same time.  I was blown away that it actually worked ... but the searching continued.

The point of this rambling, if there is one, is that you don't just build what you think is going to be a great amp and expect it to blow away a Zen Triode.  I've been trying for 10 years.  This is just one story of that attempt, of which there have been hundreds.

The pinnacle of this long distraction came when I reached the point where nothing else could be done or get any better with the amp.  It was getting great comments both here at the shop and down at the studio and I was starting to salivate over the possibility of marketing it.  The final test came one night when I put it in the reference listening room at the shop.  I drove my corner horns with it, and used my vinyl rig as the source.  Thinking I would be able to just spin a few tracks here and there and decide if it's a winner or not I found myself there hours later still frozen to my listening chair in total amazement.  Without question it had pushed it's way up to a level that easily surpassed all of our other amplifiers.  It had imaging and presence that was outside my ability to describe with all the common adjectives.  

It remained in the reference room for a week and we all listened to it and were in agreement it was something very special.  Then one day I decided to take it out and play it out on the test bench so I could listen to the radio while I worked on other things.  I discovered that somewhere along the way it lost it's huge power advantage over the Zen that it originally had.  6 clean unclippable watts was now dirty after about 2 making it not any louder than a Zen Triode would have been.  I didn't notice this in the reference room because the corner horns were 94+ dB efficient.  With a saddened heart I put it aside and pondered what to do.  This lead to a few more sessions with the soldering iron trying to get back what was lost and at the same time hopefully add some stability to it.  For example, the amp had to have an earth ground to work.  The speaker wires could not be removed while the amp was playing during that few seconds after you unplug the power cord.  If they were, you would get shocked from a wild oscillation that occurred in the split second you removed the wires... stuff like that.

The end result - or should say where it ended - was wiring it back "by the book" with grounded outputs and conventional negative feedback.  Now stable, it sounds real nice, but nothing magic is happening that would make it any different from the norm.  I learned again that yes, it is possible to make amplifiers that blow away a Zen triode in all categories, but only in the otherworld where the rules are flipped upsidedown.  In the otherworld, you may not be able to replicate the result from system to system without unpleasant surprises.  There was no way, at least at that point, that I ccould see something like this even make it's way out of the starting gate as a finished product ready for sale.  It's almost like the Devil of Ohms Law had manifested itself disguised as an angle of audio nirvana.  

When things like this happen, I usually find out that a higher power is trying to give me answers and I'm not listening.

If you're a tube amp builder and want to experience the tease of a taboo circuit when it was in development - here it is.  May the force be with you.

Note: Schematic ERROR - Pins 2 and 9 are reversed on the output tubes.

Note that the output tube in the above schematic is shown wired incorrectly.  Use the pin out shown in the schematic below.

This is the circuit exactly as I listened to it for that week in the reference room.  The usable clean power of this variation is far below the 6 watts it can have, but also qualifies as one of the best sounding amplifiers I've ever heard.  Notice that all the voicing was done in the feedback circuit on cathode of the input stage.  The output stage cathodes are also carefully manipulated.  The network on the input stage evolved from over 30 different variations, some simple, some even more complex. Even though the performance of this amp is unpredictable, I will probably remake it for myself just for those evenings when I want something really outrageously good in my reference room that I can't sell.  

Start with this circuit, you may have to remove the .0047 cap to ground, but start with it in.  I found the SV83 output tube to be far more musical in the midrange than the EL84 in this circuit.  Not necessary true in the final version. Use the best 12AX7 you can find for the driver stage.  You'll also notice the low voltages used on the driver stage placing the operational points of the tube on the fringe of it's curves.  The choke is simply another output transformer with the secondary floating.

Believe it or not a production amplifier actually came out of this experimenting and I found ways to make it stable and sound good too.  We named it after this design experience... the Taboo.  


Decware is a trademark of High Fidelity Engineering Co.
Copyright 1996 1997 1998 1999 2000 2001 2002 2003 2004  2005 2006 2007 2008 by Steve Deckert