THD Usefulness Thread = Son of Opti-MOS ?

You assume correctly, sir...
(Mark, I swear the first time I read the name Lavardin, my eyes reported seeing Lagavulin. Single malts are one of life's pleasures.)
Fascinating. I've been through it once at medium-concentration level, but will need to take more time and care with it.
Re: Tubes. I would assume that the same distortion mechanisms work in tubes, just at reduced percentages since plate dissipation is augmented by filaments, obviously.
Re: Transistors. Has anyone heard Lavardin gear? Does this concept actually work in practice, or does it simply look good on paper? I'm highly allergic to hand-waving, but this seems plausible. I don't own a good set of headphones, so I won't be able to try out this fellow's circuits as they stand. Does anyone out there have good headphones and the interest to reproduce what this guy does on his website? Subject to me getting a chance to re-read what this guy's saying, I'd like to propose a question. Since class A circuits (including the junctions inside) run at a more nearly constant temperature, is it possible that this distortion mechanism would be more prominent in AB and B circuits? Hence part of the class A sound. He may have addressed this issue. I'll be looking for it when I go back in.
Re: Current sources. They fascinate me. I use one under a differential as the first stage/phase splitter in my tube circuit. I can testify that the sound changes pretty noticably with I-source topologies, but I've not tried a cascode. Might have to give that a shot. On a related note, I believe the first stage of the Atma-Sphere amps uses a tube (6SN7) cascode as an I-source. They have a good reputation for sound. (I haven't heard one myself.) Could there be a connection? My current source topologies have all been solid state, as I wasn't willing to build another high voltage rail. Doesn't Horowitz & Hill have a bit on cascode current sources? (I'm at work and don't have all my stuff with me.)
Re: The parent of this thread: I can make my tube amp sound 'like transtors' with one of the I-source topologies I tried out. Just the one current source, mind you, and that from a portion of the circuit that some people don't even consider to be 'in the signal path.'
Mark, how'd you find this site, and are there any more like it? I'm always open to new possibilities as stuff like this sets my mind to fizzling with questions, and questions (sometimes) lead to answers.

Grey
 
"I swear the first time I read the name Lavardin, my eyes reported seeing Lagavulin" .... I nearly cried laughing ;-)

(see the very bottom of http://www.health.adelaide.edu.au/icu/ss_finnis.htm)
___________________________________________________

I have not had time to really dissect this paper yet, just posted because it seemed pertinent to the parent thread and I was interested to see what the higher-EE minds made of it - which is why I placed it "out in the open".

Found it on the http://www.audioasylum.com site, which I have only recently found and have not been overly impressed with ... people regressing to 4yo "sand-pit" level discussions with name calling and similar juvenile behaviour.

My suspicion is that this is a "cold amp" phenomenon, ie if you have a cool running amp then it would be much more susceptible, supported by the sonic signatures of A versus AB. Would also support the observation that SS amps sound "better" when they are at operating temp.

Note in his 2-stage example he has 20mA bias for both the pair and the output device. My 2S amp is closer to 20mA/3A.

The pulse signals he has used and graphed are going into cold devices ..... hell, after 20 mins there ain't nothing cold within my amp (or within about 1 metre actually!!).

Clearly haven't heard any of this gear.

not a convert just yet, but it was interesting me'thinks!

mark
 
Mark,
A quick brain tickler for you:
If we're talking thermal distortions of operating points, here, would it be better to use a small device closer to its limits (smaller mass, hence quicker settling times), or a larger device that is well over-spec'ed so that the thing hardly heats at all due to having larger junctions, etc.?
I vote for the larger device (subject to further thought--this just occurred to me) even though the hysteresis would be worse if it ever did start warming significantly.
In other words, there may be other ways to approach cures for this thermal distortion mechanism. (Can you say IRF9610 used in place of typical small signal device? Sure, I knew you could. Food for thought, there.)
Re: Ethanol. And you've got Penfold's Grange down there, you dog. Something I'd love to get acquainted with...excepting you could buy a complete high end system with one bottle around here...

Grey
 
Son of a Gun!

I wish I was more electronics literate. Even with lack of understanding of particular concepts, from what I can understand this is a serious contender in quest for quality. I hope it will not turn out to be another "ho-hum break through". Anybody going to give it a shot and build an actual working amp? Seems to me that existing cuircuits could be modified to accomodate a few extra parts , even if it means a couple of p-t-p conections. But, I could be totally wrong on this one. Anyway, I would buy some PCB's if one of you guys decide to make a prototype based on this design. I can handle assembly.
 
Thanks guys

... for your feedback.

I am the writer of the page on memory distortion... I appreciate your comments. I plan on making a PCB available for download for anyone who might want to experiment on it. I think the Internet is a great collaborating medium for that ; so when I run out of ideas, others can take over and add theirs to the melting pot.

By the way the 20mA bias is normal (it is 100mA in real life actually)... for a headphone amp. Yes, I do test with headphones...

Regarding my sources : yes, the CD63SE sounds grainy and compressed. Now, I do all my evaluation with my Harman Kardon DVD1, which is not perfect, but much better yet.

Cheers !
Pierre
 
Opti-MOS, finished.

I built the Opti-MOS Saturday. It wasn't difficult at all using the kit! :) Of course, I had to build my own power supply. That's OK though, I built a huge power supply. If it came in the kit, it wouldn't have been as good.

I spent all day Sunday listening to it and comparing it to my McIntosh on both my Legacy and Klipsch speakers.

This is my opinion, and only MY opinion. Not anyone else's. My opinion means nothing to anyone but me and those who trust me. Opinions are opinions, nothing more. I will accept no argument on this point. So, if you don't like my opinion, tough. :)

I love it. It is indistinguishable from my McIntosh, except , and this may be my imagination, but I think it has better control of the bass. The highs are perfect. I can detect distortion in the treble very easily, but I heard none. It blew the hell out of the Zen. I listened to my Zen for hours before it fried. The fact that the Zen is missing the upper quarter of audible frequencies kills it. It's a fun first circuit, but it's not a real amp.

I was able to get 100dB peaks out of my Klipsch and not hear any distortion or lack of bass. That's about as loud as I can stand. So, by every meassurement that's important to me, the Opti-MOS is perfect.

I highly recommend building this one. If you like clean perfect sound, It's the best I've heard yet. And at a total cost of $250. I'm a very happy man. If it is reliable for the next two or three months, I'm going to build three or four more.

Later,
pixie
 
Pierre,
Many thanks for posting the circuit idea. I have not yet had time to go back and give it a more thorough reading, but it is high on my list of things to do. I find it conceptually fascinating, and hope that it has relevance in the real world.
I had a posting above about device ratings vs. thermal (memory) distortions. I'd like to get your input on that one, since you've had longer to meditate on the idea.
Pixie,
Glad to hear that your Opti-MOS worked out okay for you. Just for fun, you might want to keep an eye out for surplus caps to add to the power supply. You might just find that the bass tightens further still...
Regarding your Zen, I'm curious as to how it came to be missing the upper quarter of the frequencies. I know that it had a difficult birth and, apparently, a short life span. Just wondering if some ambiguity in the schematic was responsible for one or both. In any event, it doesn't sound as though it had enough power for you--hope the Opti-MOS fits the bill.

Grey
 
The Zen is designed to roll off at about 15kHz. Look at passlabs.com. He gives the frequency graph. It's short life... I have no idea. Everything was fine except the MOSFETs. Maybe I had a bad one. Maybe it just couldn't handle the heat. I rechecked the schematic a dozen times. I got it right according to the web site. But, I can't speak for the accuracy of that. I'm assuming it's good since a few people here built one.

I currently have 40,000uF. 20,000 per rail. Do you think I need more? When I switch off the amp it continues to operate for about 10 seconds! It even has pretty good bass for 5 or 6 seconds. And that's playing it very loud.

I thought about opening my McIntosh again and looking to see how much reserve it has. I may be surprised.

pixie
 
Pixie,
As a fellow I once knew was fond of saying: If some is good, then more is better, and too much is just enough!
The proper way to talk about storage on rails is not in capacitance, but in Joules. If I recall correctly, the formula is J=1/2CV**2 (somebody check me on this, I'm at work, quoting from memory), where J=Joules of storage, C=capacitance in Farads (NOTE: *Not* microfarads!), and V= volts on the rail.
N.B.: Tube amps get a head start here, since the voltage is squared, and the rail in tube equipment typically is much higher than that of solid state equipment.
So what voltage are you running on your rails? I seem to recall that you said that the Opti-MOS circuit has variable output, presumably by applying higher rails.
Me? I'd look into the possibility of at least, say, 50,000 uF/rail. There's no magic number...it's just a question of money, and caps get expensive, fast. Search the surplus websites. Hit scrap yards, looking for junked electronic equipment that might still contain caps. Old computer equipment, particularly tape drives, can provide caps with suprisingly high voltage ratings. (Not to mention humongous transformers...)
Sound quality...don't expect a lot of difference in the treble by putting in more electrolytics, but the bass will firm up quite nicely. To do anything for the high frequencies, start bypassing with film caps. In a pinch, I've even thrown in a handful of old Sprague Orange Drop (Mylar, aka polyester) caps out of my junk box. A mere 5-10 uF will make a surprising difference with cymbals, etc. I know Slone/Self don't think much of this, but a little simple math will demonstrate what I'm talking about. Look up the ESR for a typical electrolytic, then apply that number to the capacitance...electrolytics cease to be useful at pretty low frequencies.
There is a law of diminishing returns, here. Infinite amounts of capacitance are *not* infinitely better than, say, 100,000 uF. When is enough, enough? That's between you, your speakers, your ears, and your wallet. Not to mention such practical considerations as spouses, and where to put the damned things so that they're not under foot.
I'll mess with your head another way: You can view power supply caps from another perspective. Think of them, not as a reservoir from which to draw power, but as a filter. You can take this in two ways. First, it's a filter to take out residual ripple from the incoming AC. Okay, everybody knows that. But here's the head-messer-upper: Think of the power supply caps as a shunt to ground for the audio frequencies that backlash through the rails. Pixie, this will particularly apply to you, since Slone designed that amp to be biased class B, and B throws ferocious amounts of hash back up into the rails. (Incidentally, this is a point in favor of class A & AB that Self/Slone kinda never get around to discussing...ahem.) Slone speaks of using decoupling caps to help isolate the front end. Good. But let's just get rid of the stuff while we're at it. A low impedence to ground is your best avenue. How do you accomplish this? With lotsa caps, of course! It's also helpful if the impedance doesn't vary too much from 20-20kHz (and beyond), so that's where the film caps come in. They reduce the impedance seen by the high frequency audio AC on the rails when it's looking for somewhere to go, since the electrolytics dropped out of the race a long time ago, frequency-wise.
Incidentally, the difference between 15kHz and 20kHz is somewhat less than 1/2 octave; each doubling (or halving) of frequency represents one octave.
Re: Running for 10 seconds after the power switch is hit. I once had an amp that would run for nearly a minute. That was, of course, with an ever-so-slightly augmented power supply...the depth and power of the bass was quite extraordinary.

Grey