I stopped chasing CFA/VFA , or even PPM ... a while ago.
Ha , Ha "EFA" (you know what the F part of the acronym is ?).
A true "blameless" is an amp that has fewer semi's , under 30ppm , and
lower open loop gain.
The two (below1/2) were actually tested and came close to the simulator
predicted 20-30ppm. Not only can these be abused (shorted) , but I could
hear no real difference with many genres ,or home theatre use.
Absolutely clear, clean and very resilient.
What's next for an "EFA" ? A simple 10 semi input stage needs good
matching - BC846BPDW matched SOT "dualies"....
"credit card" input stages (below) , but I will keep >1W components through-hole.
PS - I bet those two designs would be indistinguishable if plugged
into the same output stage. Below 50ppm all descends into the "haze" of
reliability and other factors (noise/PSRR/ease of construction).
OS
Attachments
Yes, in particular when recalling that for famous designers always trusting their ears, the difference between "open loop gain" and "loop gain" is irrelevant.
The End of Simplicity as a isolated positive attribute?
For an MC stepup device someone at a client's organization bought a kit from a vendor who aims to allow self-described audiophiles to build their own equipment. The vendor declares that most of the components are obsolete, but they acquire them from eBay. They also add "I think simplicity is a huge part of why classic equipment sounds so good---and modern stuff can sound so bad." Of course no examples are given, let alone under what circumstances these evaluations were conjured.
The notion was that the MC stepup device could be reproduced on some spare area of the existing PCBA that now has a very good MM preamp. What could be easier? Well, I enumerated a number of things that would complicate matters, including excessive distortion, poor overload margin, gain variations producing balance problems unless trimmed in production, very high power supply sensitivity, and most of all, excessive intrinsic noise. And don't forget obsolete components, some of which are likely to be counterfeit. In addition the output coupling capacitor entails a first-order rolloff when loaded with 47k of 33Hz, which would bother some and not bother others. The latter is easily fixed, but indicative of a sloppy design IMAO.
I think it is time to slay the simplicity-is-better god. Ending its reign does not entail that more parts are necessarily better either. Good design will remain the province of good designers.
Having spent years simplifying for the sake of cost reduction I'm sick to death of it. The equipment sounded o.k. in spite of, rather than because of, the simplicity.
I'm reminded of the armchair savant who declared that in an IC op amp (and presumably in a comparable discrete circuit), every component added noise (today he'd probably haul out "sonic signature" as well), so clearly the fewer the better. Yes, if you reduce the parts count to zero you will be better off still.
The other frequently-cited red herring is that higher parts counts entail lower reliability. I think Pease wrote a trenchant column about this and demolished it with some well-chosen examples, like leaving out reverse-bias protection devices.
Why fewer semiconductors? Where does this notion arise?
For an MC stepup device someone at a client's organization bought a kit from a vendor who aims to allow self-described audiophiles to build their own equipment. The vendor declares that most of the components are obsolete, but they acquire them from eBay. They also add "I think simplicity is a huge part of why classic equipment sounds so good---and modern stuff can sound so bad." Of course no examples are given, let alone under what circumstances these evaluations were conjured.
The notion was that the MC stepup device could be reproduced on some spare area of the existing PCBA that now has a very good MM preamp. What could be easier? Well, I enumerated a number of things that would complicate matters, including excessive distortion, poor overload margin, gain variations producing balance problems unless trimmed in production, very high power supply sensitivity, and most of all, excessive intrinsic noise. And don't forget obsolete components, some of which are likely to be counterfeit. In addition the output coupling capacitor entails a first-order rolloff when loaded with 47k of 33Hz, which would bother some and not bother others. The latter is easily fixed, but indicative of a sloppy design IMAO.
I think it is time to slay the simplicity-is-better god. Ending its reign does not entail that more parts are necessarily better either. Good design will remain the province of good designers.
Having spent years simplifying for the sake of cost reduction I'm sick to death of it. The equipment sounded o.k. in spite of, rather than because of, the simplicity.
I'm reminded of the armchair savant who declared that in an IC op amp (and presumably in a comparable discrete circuit), every component added noise (today he'd probably haul out "sonic signature" as well), so clearly the fewer the better. Yes, if you reduce the parts count to zero you will be better off still.
The other frequently-cited red herring is that higher parts counts entail lower reliability. I think Pease wrote a trenchant column about this and demolished it with some well-chosen examples, like leaving out reverse-bias protection devices.
Brad,
Two tin cans and a piece of string, the best phone device ever produced, minimum distortion! Wireless 😀
Two tin cans and a piece of string, the best phone device ever produced, minimum distortion! Wireless 😀
And, finally, testable predictions for string theory!
EDIT: And not to forget the pivotal contributions to the development of the CAN bus.
EDIT: And not to forget the pivotal contributions to the development of the CAN bus.
Last edited:
Usually I trust other peoples ears more than my own. However, for my own personal hi end audio system, I am stuck with trusting my own ears mostly. I do, in any case, trust that what others and I hear is real.
Because I don't like people endlessly promoting their products, and relentlessly hunting for endorsements at any price, even disregarding any technical reasoning.
Not to be raw , but it is the performance to count ratio.
A higher count design (like Dadod's), has <5ppm as the goal.
The higher count version of my former second attachment has 16 , instead of 10 - semi's.
The extra 6 allow better/more choices of input devices , higher rail voltages
(cascodes).
Here , the extra's just enhance the base design. In this case , reliability just
may be enhanced (wider supply tolerance) , as well.
From the DIY point of view , it is much easier to advise on builder mistakes
with a simpler design. Many more new members will attempt <16 device
project and get better than OEM results !
Some of the 30 count input stages may approach sub ppm , but as long as a
10-16 can still be <20 (and last) , would we ever hear the difference ?
This is all about compromise , including cost/complexity/reliability.
You mentioned "classic equipment" , what's wrong with them ?
Some of these are still working after 40 years. WOW !
Some are quite spectacular ported to 21'st century semi's.
OS
I see some violent agreement here!
You both agree there is a sweet spot where performance is way better than needed and component count is optimal. Less and it gets worse (bean counters or gurus) more and it's added cost and complexity for gain you can't hear, or some useful gain somewhere else (lack of matching).
You both agree there is a sweet spot where performance is way better than needed and component count is optimal. Less and it gets worse (bean counters or gurus) more and it's added cost and complexity for gain you can't hear, or some useful gain somewhere else (lack of matching).
My own spin on classic equipment is that it should be judged on its merits not on its nostalgia nor simply its parsimony in parts. Which case there's plenty of good old designs that scale well with modern parts.
Audio design is an 'art' not just engineering. That is what is missing here with many well meaning designer critics.
We have been debating audio design for many decades. First, it was what sort of tube power amp sounded best? Triode or Pentode. Later, Ultralinear became a reasonable compromise.
Then, solid state (germanium) vs tubes. Tubes won. Then Solid state (silicon) vs tubes and solid was able to catch up. However, not ALL solid state (silicon) sounded really good, so many designers, including me, went out to find out why.
Slew rate became very important. So did lowering higher order distortion. Frequency response has remained essentially the same over the many decades, and maybe typical digital is worse in this regard.
Still, there is more. Any ideas?
We have been debating audio design for many decades. First, it was what sort of tube power amp sounded best? Triode or Pentode. Later, Ultralinear became a reasonable compromise.
Then, solid state (germanium) vs tubes. Tubes won. Then Solid state (silicon) vs tubes and solid was able to catch up. However, not ALL solid state (silicon) sounded really good, so many designers, including me, went out to find out why.
Slew rate became very important. So did lowering higher order distortion. Frequency response has remained essentially the same over the many decades, and maybe typical digital is worse in this regard.
Still, there is more. Any ideas?
Come on, it's not an art, it's a little bit of engineering and a whole lot of propaganda.
And about the easiest area of electronic engineering, the typical home turf for
beginners, where they easily find success in building something that works
and sounds OK. People who would never get a PLL to lock and who would
kill a laser diode in 1 usec. That does not mean that they don't master that
sooner or later, they are not dumb, just beginners.
Look into The Art Of Electronics, the book. How much is there on audio?
And they didn't even bother to talk about specialist areas like RF, radar,
control theory. You really should stop celebrating your tunnel vision.
And about the easiest area of electronic engineering, the typical home turf for
beginners, where they easily find success in building something that works
and sounds OK. People who would never get a PLL to lock and who would
kill a laser diode in 1 usec. That does not mean that they don't master that
sooner or later, they are not dumb, just beginners.
Look into The Art Of Electronics, the book. How much is there on audio?
And they didn't even bother to talk about specialist areas like RF, radar,
control theory. You really should stop celebrating your tunnel vision.
Gerhard,
This one I will argue with you.
To me the difference between engineering and just technician level building is in defining the problem.
If I ask a technician to build a 24 volt 1 amp power supply with less than 1% ripple I would expect any competent technician to be able to do that.
But what an engineer does is take the fuzzy spec of we need a power supply to power this ... And translate that into a final specification or design.
Now for audio many practitioners can build an amplifier of 100 watts to drive an 8 ohm loudspeaker with a signal to noise ratio of 95 dB and a distortion less than .05% at power levels from .01 to 100 watts.
But the art is knowing not just what parameters to specify, but the order of importance.
For example right now I am having problems getting loudspeaker cables that meet my project's needs.
It used to be loudspeaker wiring was in metal conduit. To save money folks want to eliminate conduit. Now I suspect some folks here have hung a long wire and used it for an antenna. Well it should come as no surprise that if you have an AM radio station on a frequency around 1 megahertz and are stringing 75M of loudspeaker cable at the fiberglass sunscreen level of a stadium there is a bit of signal ingress.
Now to really make things interesting many of the current crop of professional audio power amplifiers are not stable driving capacitive loads as would be found in shielded cable.
Now for entertainment value the architect's consultants are really only at the technician level and don't recognize the problem or issues. But they continue to design these cost saving systems!
The last Great Leap Forward was to change to digital audio distribution systems as the problems with power line hum go away with digital transmission. Now which do you think costs less; 1000' feet of single pair shielded cable or 1000' feet of optical fiber and the encode-decode electronics?
Why the hum problem, because quite simply they don't understand the issues involved.
Now I suspect you will consider these all engineering technical issues. But on this side of the puddle getting the big picture to see the issues is what is the "Art."
This one I will argue with you.
To me the difference between engineering and just technician level building is in defining the problem.
If I ask a technician to build a 24 volt 1 amp power supply with less than 1% ripple I would expect any competent technician to be able to do that.
But what an engineer does is take the fuzzy spec of we need a power supply to power this ... And translate that into a final specification or design.
Now for audio many practitioners can build an amplifier of 100 watts to drive an 8 ohm loudspeaker with a signal to noise ratio of 95 dB and a distortion less than .05% at power levels from .01 to 100 watts.
But the art is knowing not just what parameters to specify, but the order of importance.
For example right now I am having problems getting loudspeaker cables that meet my project's needs.
It used to be loudspeaker wiring was in metal conduit. To save money folks want to eliminate conduit. Now I suspect some folks here have hung a long wire and used it for an antenna. Well it should come as no surprise that if you have an AM radio station on a frequency around 1 megahertz and are stringing 75M of loudspeaker cable at the fiberglass sunscreen level of a stadium there is a bit of signal ingress.
Now to really make things interesting many of the current crop of professional audio power amplifiers are not stable driving capacitive loads as would be found in shielded cable.
Now for entertainment value the architect's consultants are really only at the technician level and don't recognize the problem or issues. But they continue to design these cost saving systems!
The last Great Leap Forward was to change to digital audio distribution systems as the problems with power line hum go away with digital transmission. Now which do you think costs less; 1000' feet of single pair shielded cable or 1000' feet of optical fiber and the encode-decode electronics?
Why the hum problem, because quite simply they don't understand the issues involved.
Now I suspect you will consider these all engineering technical issues. But on this side of the puddle getting the big picture to see the issues is what is the "Art."
Last edited:
Audio design is an 'art' ?
If and when the auto-placement and auto-route of a modern CAD can
figure out ground layout ( maybe some can) , then audio will no longer
be an "art".
Auto - route is nice , but does not group by currents/ground returns/loop area
of a trace.
Diptrace DOES have the rules automated
for digital circuits. Even a much more sophisticated MPU design would be
automated from beginning to end.
If you obeyed the electrical constraints of the design , you would end up with
a functioning end result. If you let Diptrace design the audio (layout) , you would have
a mediocre functioning circuit - analog is "picky".
OS
If and when the auto-placement and auto-route of a modern CAD can
figure out ground layout ( maybe some can) , then audio will no longer
be an "art".
Auto - route is nice , but does not group by currents/ground returns/loop area
of a trace.
Diptrace DOES have the rules automated
for digital circuits. Even a much more sophisticated MPU design would be
automated from beginning to end.
If you obeyed the electrical constraints of the design , you would end up with
a functioning end result. If you let Diptrace design the audio (layout) , you would have
a mediocre functioning circuit - analog is "picky".
OS
Gerhard, I do not agree with you. I think that AUDIO engineering is as much art as engineering. I learned this from long experience of making both successful and unsuccessful audio products over the last 50 years. It is the closed mind of the typical engineer that keep them from making great audio electronics or even great loudspeakers. You have to believe enough in what you are doing, and what you and your associates actually hear (yes, you have to trust your ears) in order to gain the 'feedback' as to what you designed was the right approach. I first found this out 45 years ago, when the Grateful Dead rejected a solid state circuit board that I helped design with Harris 911 op amps, operating at +/- 24V, and which measured pretty darn good with IM distortion (which is what I could measure to 0.005% at the time) and frequency response, and the slew-rate of +5/-2.5V/us. (pretty good for the time)
That is when I learned that there was more to audio design than just engineering, and I started developing the discrete circuits that I am now famous for. Can an IC take over, TODAY? Only for the rest of you who believe measurements first, and foremost.
For me, even my latest and best IC based circuits will no match my best discrete designs, and sometimes my best discrete designs do not always match the best tube designs that I have ever heard. Now, you don't have to believe me, but I do seem to be selling myself short advocating some tube equipment, over my own, so your argument does not seem to hold up.
That is when I learned that there was more to audio design than just engineering, and I started developing the discrete circuits that I am now famous for. Can an IC take over, TODAY? Only for the rest of you who believe measurements first, and foremost.
For me, even my latest and best IC based circuits will no match my best discrete designs, and sometimes my best discrete designs do not always match the best tube designs that I have ever heard. Now, you don't have to believe me, but I do seem to be selling myself short advocating some tube equipment, over my own, so your argument does not seem to hold up.
Last edited:
True for fashion audio, where the product is tied in with the story and the voodoo and where you have to sell a parity product at a major multiple. For the simple tasks of making a small signal bigger without any audible consequence, all it takes is engineering.
Which, in the bizarro universe of hucksterism means, "Don't use your ears, peek!" And I admit that's necessary when you're selling fashion.
Yes electronics are as much science and engineering as any other professional design field but to say that there is no art in this is to miss the point. It is not about peeking or just using your ears but the art of the compromises that every engineering field requires. What and where is up to the designer but just as one engine builder vs another can make an engine produce 500 hp at 5500 rpm one will sing and another will be a brute that just uses brute force. How you get to the end point is the art. not just the measured final performance. To argue that electronic components are all going to be equivalent and interchangeable with no sonic consequences seems silly, not all capacitors or resistors are created equal even if technically many may do the job. A speaker is the same, any first year engineering student should be able to create a dynamic loudspeaker, doesn't mean the end result will sound good just because it makes noise and moves air.
I have to give it to John on this one that indeed there is an art to this at a certain level beyond just being a competent electronics designer.
I have to give it to John on this one that indeed there is an art to this at a certain level beyond just being a competent electronics designer.
No one ever argued that. Basic engineering drives rational component choice. No voodoo needed.
- Status
- Not open for further replies.