Giovanni Stochino's ultra fast amp

I have built Giovanni Stochino's ultra fast amp as published in Electronics World and I am very impressed with its performance. My previous ss amp was mid-80s design from Sydney Uni which featured nested differentiating feedback loops. Again, I liked this amp. What do others think of this Stochino's design vs other ss amp designs?
Just for fun, you might want to take a look at the link for The Leach Amp under Amplifier-Projects. This is the current incarnation of one of the earliest low TIM (i.e. high slew rate) amps. I heard one of the older versions a long time ago--frankly don't remember being much impressed, but time may have played tricks on my memory and, at any rate, the circuit has been updated many times since then.
The circuit on the link you provided looks to be much simpler, which to me is a good sign.
One thing that occurs to me is that the Stochino circuit might benefit from some parts upgrades. Several of the small signal transistors used are, I believe, classed as 'general purpose.' Judicious substitution of MPSA18s/MPSA42s/MPSA92s might further improve the performance, but to do a good job would probably entail adjusting resistor values, etc., quite likely more trouble than it's worth.
Some of the lads on this site seem to get their jollies by running up software simulations of circuits and comparing them. Talk one of them into running a head-to-head between the Leach and Stochino amps. It wouldn't be the same as listening, which is my preferred way of tilting at windmills, but it might prove interesting.
I'll try to take another gander at it when I have more time.



2001-02-04 4:23 am
The circuit at has some errors in it.The IRF640s are drawn wrong.The 12V zener diodes need to be on the outputs not the drivers. Miller capacitance from drain to gate when driving an inductive load will hit the gates hard and decrease the life of the outputs.Active current sources are noisey and don't sound good.A resistor from a regulated supply ala Leach works better.An active current source as a load really pushes the open loop gain.It is better to trade gain for frequency response.The resistor between the emitters of the cross-coupled diff inputs is a nice touch and could be easily added to the Leach.The Leach can go faster if you go from a tripple to a double output stage and substitute MJE15030/31 30Mhz drivers and MJ1302/3281 30Mhz outputs.The Leach would benefit from a regulated voltage for the front end too.A .33 ohm source resistor doesn't do much for equalizing currents on an IRF9640 with an Rds on of .5 ohms, the Vgs need to be carefully matched in this design.
Stochino schematic

DJK wrote:

"The circuit at
has some errors in it.The IRF640s are drawn wrong. The
12V zener diodes need to be on the outputs not the

Actually the Zeners are correct, at least according to
the most recent EW article. Sorry if the MOSFETs are
drawn wrong - this was a tediously hand-drafted copy of
the original using Illustrator.

I'm afraid I still can't offer any comments on the
performance of this amp as I'm still working on my preamp.
I have completed the PCBs and built the chassis for the
power amps, but haven't got around to putting the
(substantial) power supply together yet.



2001-02-04 4:23 am
The error in your drawing is easily forgiven.I only pointed it out in case some inexperienced person tries to build it from this schematic.The placement of the gate protection zener diodes is a design issue as are most of my other remarks.I earn my paycheck analyzing design flaws and making suggestions on how to improve said designs.If I seem overly critical it is because it has become second nature to me.
I Built A Circuit Similar to this one and the Leach Amp Years Ago. It still Works Fine. I was wondering what the Concerns were with using a Constant Current Source For the Diff. Amp? I Thought that the CMRR is improved by Constant Current vs Just a resistor? Also What Advantage is the Resistor and capacitor from the Pos. Diff amp to the Neg. Diff amp for the Emitter sides. TIA
A constant current tail current source considerably improves the linearity and dynamic range of a differential amplifier;
along with a current mirror for the 'other side', it also
improves PSRR. It's quite a bit better than a simple resistor, but it may require other changes if it is added to an existing design.

Interestingly, Marshall Leach doesn't use these design
enhancements in his amplifier. I'm not clear why, though
he gives some reasons in the project web page.
Components in input stage

ppl asked about the capacitor and resistor coupling
the two halves of the imput stage. I'm not sure about the
resistor, but Stochino says of the capacitor in his April
1997 article:

"In this design, a 1nF capacitor has been added across R to
increase the dynamic transconductance and the available peak
current of the input stage....

...this results in a slight increase of speed for input
signals within the linear dynamic range of the amplifier and
in a further reduction of the already low residue of dynamic
intermodulation distortion"

Now you know...

From what I'm reading in various sources on amplifier design, current sources and current mirrors apparently reduce distortion considerably but are most effective on even order harmonics such that odd order harmonics tend to predominate in the residuals, which can be >very< low.

I've noted that Leach prefers not to use these techniques, and says he prefers the sound this way. He also gets very low distortion figures, but perhaps these are predominately even order which may indeed sound less offensive.

No one seems to take intermodulation distortion seriously, for some reason--perhaps not all of the well-known authors are equipped to measure it? Because of limitation in the available gain and feedback at high frequencies, distortion
rises as frequency increases, so perhaps that's where
designers ought to concentrate?

Myself, I'm going to play with Sziklai compounds in the output stage since that's where the bulk of the distortion is generated (as opposed to Darlingtons; Sziklai compounds or 'complementary feedback' pairs contain a 100% local feedback loop, right where its needed, though harder to design for stability).

Just speculating for speculation's sake...
Unfortunately, resistors do sound better than current sources. I wish it weren't so, as current sources (and current mirrors--nothing but a dual current source) provide an elegant solution to a lot of problems. When I built my tube amp, I set up the front end as a differential (6SN7) and used it as a phase splitter. To get good balance between the two output signals, you have to use a really large cathode resistor. So I said okay, no problem, I'll use a current source, which is in effect an infinite resistance. I got great balance between the outputs (less than .1 dB difference). Wonderful! So I'm sitting here looking at this current source and I says to myself,"Self, why not put another one of these under the driver stage?"
Tried it.
Sound stage collapsed. High frequencies got harsh. Went back to the resistor. Better. Tried a different current source. Still not as good as a resistor...
I'm on my fourth or fifth current source topology and have yet to find one that sounds as good as a common off-the-shelf NTE 2% metal film resistor. *And it ain't even close.*
Yes, I still have a current source under the front end, but the driver stage is running with a resistor, and I'm trying to figure out a practical way to get rid of the front end current source. I may resort to running a, say, -50V rail down under ground and bringing the cathode resistor up from that, but it's gonna be a lotta trouble.
In short...yes, dammit, I will (grudgingly) admit that they're useful little critters, but there ain't no such thing as a free lunch. You pay for it in sound quality. Everyone claims they want sound quality. But how many times have you ever seen someone admit that a circuit that they designed and/or built was anything less than a sonic revelation? I'll bet you can count the times on one hand and have fingers left over. Right? Now, we *know* that there aren't that many electronic geniuses out there, building perfect amps. So something's fishy, here. All I'm saying is, try it. Try to arrange a circuit such that you can drop in either a resistor or a current source (better yet, several different current sources, as they don't all sound the same).
Then listen.
Then report back.


I plan to try out current sources and perhaps mirrors in an existing circuit that uses resistors, just to see if I can
hear a difference. It won't exactly be drop-in/plug-in; I'll have solder in the parts, but it shouldn't be too cumbersome to play with.

My gut feeling is that some people like euphonic distortion, but I suspect the issue is more subtle and complex than that. Adding current sources to an existing
circuit changes several operating parameters of the entire
amplifier circuit. If I had a better understanding of circuit design, I might find it possible or necessary to reduce gain/feedback for stability, or simply to experiment
with reduced global feedback to see if that audibly affects the sound quality.

Life's an experiment, and hopefully some of the time it can be fun, too.

That's the spirit! I'll be waiting for a report.
A trick that I use in a case like this that may help is to treat the main circuit board as a mother board and tag little baby circuit boards (some as small as postage stamps) into the holes where the resistor would go. For a 1/2W resistor, I use .6" lead spacing. Knowing this, I make the little circuit board with a pair of holes that are also .6" spacing, then tag in a pair of leads that I clipped off of a resistor. Bend the leads at a right angle if it suits you, so that the board stands off the main circuit board at a right angle. (This trick also saves board space if you're ever in a really tight layout situation in the future...hint, hint.) Anyway, you can pop mini-boards in and out fairly easily this way.
Keep all mini-boards for future use. You never know when you'll need a current source in the future. It's kinda neat to be able to reach into your junk box and drop in a ready-made current source. (A little like a nice pre-tested subroutine when you're writing a computer program.) Since most current sources can be programmed simply by changing a resistor, you can adjust them on the spot and be ready for your next project.
Interestingly, I can make my tube amp sound "like solid state" just by changing the front end current source. Which one? The one with the LM317 and the resistor. You can guess what I think of LM317s sonically, but it's fascinating to hear. (Jeez, gimme a was late, I was tired, and I needed 14mA. It worked. I went to bed, got up the next day, listened, and immediately set about putting together another current source...)
Regarding euphonics:
Clearly, the sonic differences between different pieces of equipment lie in various distortions, whether solid state vs. solid state, solid state vs. tube, or tube vs. tube. Hopefully, we can all agree on that. The general view among many who listen to solid state is that somehow people who like tubes have been hoodwinked into believing that a signal with a lot of 2nd harmonic distortion is "truer" to the original than a signal that shows lotsa zeroes after the decimal point. The curious point is that addition of second harmonic to a 'clean' signal doesn't replicate the sound of tubes, nor does it sound quite like life. The Aphex Aural Exciter, used in studio work to fatten up a singer's voice, is nothing but a harmonic distortion generator. It doesn't sound like tubes. (If you want to sound 'like tubes' you use tubes. It's easier that way.) Clearly, there's more to this than meets the eye. Personally, I think that the key to making progress on the distortion front for both tube and solid state is to develop strategies to measure transient distortions. Mark (mefinnis) located a website and posted it in another thread where someone in France (I think) had an interesting idea along these lines. I won't comment further here. Go look up Mark's thread.
In the meantime, I find it amusing to note that solid state manufacturers fall all over themselves to tell you that their designs sound 'like tubes,' but you have to search long and hard to find a tube manufacturer that would brag that his equipment sounds 'like solid state.'
I like 'em both, each for what they do well.

Yep, that's the one. Whether it works out in the real world, I dasn't know, but it sure is interesting. I'd like to see more like it, if anyone knows where more fresh ideas can be found. I think we can clearly ignore sites that claim that circuits assembled under a full moon sound better than ones assembled under a new moon. We need ideas that at least look semi-logical.
That said, I have to confess that cables do have direction, although I've never heard a theory that I thought would hold water. (Diode effect between crystals in the wire, quasi-ferromagnetic properties being imparted be the direction that the wire passed through the rollers when it was drawn, etc.) Perhaps they were assembled under a full moon...who knows. The effect is subtle; you need a non-trivial system to hear it. Folks with party systems in dorm rooms are free to connect cables in any direction that suits, as it won't matter under those conditions. While I'm on the subject of cables, I'd like to note that there's an awful lot of B.S. to be found on the subject of various winding strategies and so forth, and that the prices they charge are scandalous. I have here, Chez Bear, a speaker cable made by a Reputable Company that sounds just awful. It's on loan from a buddy (who doesn't like it, either). They've got a lot of fancy theory to tell you why their cable sounds wonderful...except it doesn't. I told my buddy to take it back. He left it here. 'Nuff said. (Egad, the cost of these things! Glad I didn't pay for them...)
Oh, incidentally, it ain't hard to figure out why a gain device might be inferior to a plain ol' resistor. Consider which more closely approximates the ideal for its kind: resistor or gain device. Good resistors (by which I mean your ordinary metal film critter) are pretty linear, nearly perfect in the ideal sense. Yes, there's going to be some stray capacitance & inductance, some thermal effects, etc. but damned close to the ideal. Semi's and tubes? Fuggetaboutit! Nowhere near. Distortions run rampant. Arguments that semi's do the current thing well miss the point--resistors do it better. This is easiest to see in a single-ended current source/transistor pair (think Zen amp, here), where the current source may be delivering steady current, but people ignore the fact that the voltage is bouncing all over the place in order to keep up with the gain device. Clearly the poor thing can't be expected to go from DC to light, due to capacitance and whatnot, i.e. it's imperfect. A resistor (think SOZ) will be able to bounce faster than the semi, albeit at an efficiency penalty. Somebody check me on this, but I believe that the Zen goes to about 15-20kHz, while the SOZ would, in a pinch, do decent duty as a Ham radio transmitter. Three guesses as to why, and the first two don't count...