Before I go forward with this, I would like to point out a thing or two. I personally concentrate on audio quality, something that I seem to have an affinity for, and success at achieving. However, I am no 'expert' in cars, food, wine, or many other things that many here, including those who criticize me, appear to be more involved in than me. Some people here will pay almost 'anything' for a better bottle of wine, yet laugh at an audiophiles concern over what sort of volume control he selects.
Now, what is 'wrong' or different about IC's compared to my own designs that are made almost entirely of discrete devices?
First, a little history: Back 40 years ago, solid state audio quality, whether discrete or IC based was marginal to lousy. Tubes really DID sound better in virtually every case.
What to do? Improve topology to get lower inherent distortion? (my approach) Or what about adding lots of extra negative feedback to suppress the distortion? (IC approach)
Both approaches were attempted. First, the Levinson products stood out beyond mid-fi and gave tubes a good run for the money. This was because we used good parts, good topologies, and Class A, to make extra fast, low distortion products.
What about IC's? By about 35 years ago, with the introduction of the 5534, many of the BIG problems between the best discrete designs, either tube or solid state, diminished significantly, BUT not completely. So I kept going with discrete designs, rather than just use IC's. In truth, it would be a lot easier, cheaper, and more convenient to just use IC's at this point. If it works for you, go for it!
I have found, however, when I finally got a way to buy and use quality discrete parts through Parasound, I could offer discrete designs at a 'reasonable' price, and that I could always make a BETTER audio design with a discrete design, mostly because I can make it much more Class A, rather than Class B, which almost all IC op amps are set, in order to conserve current.
Now WHY is there such a difference? Well, Class A has a transfer function that resembles an S curve, and Class B has a somewhat uneven transfer function that looks WORSE at lower operating levels, where the Class A transfer function has DECREASING distortion at lower and lower levels. In fact, if the transfer function is reasonably linear, you can PREDICT THE DISTORTION at lower levels mathematically, with great accuracy.
Guess what, my Class A designs reduce to almost NO measurable distortion at really low levels. This cannot be predicted with IC designs.
Now I realize that negative feedback CAN and DOES reduce what we measure with our test equipment, yet does it do this without cost, such as FM distortion, or something we are not yet measuring? Everything in my experience tells me that feedback causes as many problems as it fixes, so I try to keep it to a minimum.
I realize that some here believe differently, I hope to see you in the audio marketplace, because IF you can do what you think you can do, then you can certainly run me out of business. '-)
Now, what is 'wrong' or different about IC's compared to my own designs that are made almost entirely of discrete devices?
First, a little history: Back 40 years ago, solid state audio quality, whether discrete or IC based was marginal to lousy. Tubes really DID sound better in virtually every case.
What to do? Improve topology to get lower inherent distortion? (my approach) Or what about adding lots of extra negative feedback to suppress the distortion? (IC approach)
Both approaches were attempted. First, the Levinson products stood out beyond mid-fi and gave tubes a good run for the money. This was because we used good parts, good topologies, and Class A, to make extra fast, low distortion products.
What about IC's? By about 35 years ago, with the introduction of the 5534, many of the BIG problems between the best discrete designs, either tube or solid state, diminished significantly, BUT not completely. So I kept going with discrete designs, rather than just use IC's. In truth, it would be a lot easier, cheaper, and more convenient to just use IC's at this point. If it works for you, go for it!
I have found, however, when I finally got a way to buy and use quality discrete parts through Parasound, I could offer discrete designs at a 'reasonable' price, and that I could always make a BETTER audio design with a discrete design, mostly because I can make it much more Class A, rather than Class B, which almost all IC op amps are set, in order to conserve current.
Now WHY is there such a difference? Well, Class A has a transfer function that resembles an S curve, and Class B has a somewhat uneven transfer function that looks WORSE at lower operating levels, where the Class A transfer function has DECREASING distortion at lower and lower levels. In fact, if the transfer function is reasonably linear, you can PREDICT THE DISTORTION at lower levels mathematically, with great accuracy.
Guess what, my Class A designs reduce to almost NO measurable distortion at really low levels. This cannot be predicted with IC designs.
Now I realize that negative feedback CAN and DOES reduce what we measure with our test equipment, yet does it do this without cost, such as FM distortion, or something we are not yet measuring? Everything in my experience tells me that feedback causes as many problems as it fixes, so I try to keep it to a minimum.
I realize that some here believe differently, I hope to see you in the audio marketplace, because IF you can do what you think you can do, then you can certainly run me out of business. '-)
Last edited:
isn't it actually 295 Ohms on one, 418 Ohms on the other of the I/V op amps? - and then what happens with balanced earphone takeoffs?
don't forget that the balanced DAC outputs constitute a pair of dependent sources, one side will be pulling the diff-to-SE TPA6120 virtual gnd input node a factor of 0.5 in the opposite direction
and what does 100 pF do with 530 Ohms - 100 pF is too small - the low pass greatly help I/V performance, even 200 kHz 1st order is fine by most audiophile's and there really should be higher order filtering where the I/V pole could be part of a 3rd order LP designed to some preferred alignment
Last edited:
Nobody, here want to run-you out of the business. First because no one hate you, second, you seem to need-it for a living, third, because you are talented.
Now, back to your points.
Agree with tubes and first solid state devices. Well, a lot of mistakes where done, at this time, because a lot of engineers used to think 'voltage', as they were used with tubes, and solid state need a 'current' point of view.
Second, the first solid state devices were so slow, that it was just IM generators and HF demodulators.
It would be a bad idea to drop, discouraged, at this time: Do-you still design with tubes ?
Same thing about the Class A vs Class AB point of view. When the slew rate is low, what you write is true. When the slew rate is fast enough for the amp can correct any level error in nS, what do you fear ?
In fast OPAs and closed loop Amps, we see that the distortion decrease with level as well.
My point of view is not to exclude the two approaches: to design an open loop amp with as low native distortion as possible, and as much feedback you can. With some precautions for not let enter parasitic signals in the feedback loop, or add distortion here with anything else than a 'Resista'
!(Don't kill-me for my poor English)
.
Last edited:
Presumably you have tested the evil AD797 and 'proved' in your usual fashion it generates these HIGHER ORDER distortions. How about comparing a 797 device against Blowtorch for these distortions?
Can you also pontificate on the Quan & Hirata distortions for AD797 as you have claimed ad nauseum that these are important? While you are testing AD797, please test one of your 2 Blowtorches too.
I take it you also have 'proof' of this.
I already asked this, but with a wider scope (any "high end" preamp like BT or Parasound). He doesn't.
AD797 will have high order harmonics somewhere at -125 or -130dB. Many of simple open loop designs will have it much higher accompanied by low order harmonics at -80dB or so. Again, such reasoning is funny and completely wrong and -130dB cannot be audible. If there is a reason of a different sound, it should be found somewhere else than at higher harmonics of -130dB.
John, you do yourself no favours by making such remarks. Anybody who has had a career in almost any industry, or hired others to do jobs at home or work, will know only too well that making a living at something is only weakly correlated with skill in that field.john curl said:
Please offer something better than boasting and anecdotes. "My customers can't be wrong" and "My ears can't be fooled" don't impress us. Why not share your undoubted knowledge with us poor mortals, but without continually insulting us?
Actually if you adopt the alternative 'Cherry' compensation that's in the AD797 datasheet, you'll see high order harmonics magically disappear .. IF the rest of your layout & circuit are up to scratch.
JC's measurements suggest poor decoupling and earthing. Too much concentration on Golden Pinnae bits and too little attention to basic principles.
Clifforest, that was NOT how it was, so far as I know. I seriously doubt that they even opened the case, BEFORE making their listening opinions. This is a MYTH, reverse engineered to EXPLAIN why they decided the amp was not 'ready'. (Direct quote)
The amp did NOT sound right, AND when we removed the IC, it sounded better. Not only with my amps, but with my associate Carl Thompson when we modified his unit. It got quieter, too. Maybe that was what it took. In any case, the MK2 was sonically better than the MK1, even if you just removed the IC and bypassed it.
The amp did NOT sound right, AND when we removed the IC, it sounded better. Not only with my amps, but with my associate Carl Thompson when we modified his unit. It got quieter, too. Maybe that was what it took. In any case, the MK2 was sonically better than the MK1, even if you just removed the IC and bypassed it.
JC,
We are going to disagree. Unless you are driving on a track or under weather conditions where the wise would not, the the performance value of a car changes. I go for comfort, safety, reliability, ease of parts and service, mileage and the ease of reading and using the controls.
I had 1,000,000 miles under my belt without a moving violation or accident. Until the guy driving the import hit me. I walked away (O.K. Limped because I did have my foot on the brake.) He didn't.
Attached is what my car looked like after. (No damage to the passenger compartment!)
ES
Attachments
IOW, you have no relevant and comparative data to back up your assertions.
SY, what do you want? I showed higher order distortion from a typical graph from Samuel Groner's measurements of the AD797, that has been discussed here before.
Admittedly, it is at a high output, but it does show higher order distortions.
It is almost impossible to measure the actual nonlinearity of an AD797 at low levels (typical working levels) because the high negative feedback and the residual noise floor hides it. This is also true with a CTC Blowtorch, but I DO have one advantage. Because I operate heavy Class A, I can predict the amount and order of distortion at low levels, mathematically, using the formula that 2'd harmonic drops directly with level, and 3'rd harmonic drops with the square of the level. Is this what you want?
Admittedly, it is at a high output, but it does show higher order distortions.
It is almost impossible to measure the actual nonlinearity of an AD797 at low levels (typical working levels) because the high negative feedback and the residual noise floor hides it. This is also true with a CTC Blowtorch, but I DO have one advantage. Because I operate heavy Class A, I can predict the amount and order of distortion at low levels, mathematically, using the formula that 2'd harmonic drops directly with level, and 3'rd harmonic drops with the square of the level. Is this what you want?
The arguments flow back and forth, back and forth, no progress is made. The point is, no matter how fussy you are with audio you will always hear distortion in the end result, for one reason or another. The electronic guys will point scathingly at the speakers and recordings as letting down the side, the recording guys will point scathingly at the quality and design of electronics in consumer gear as ruining their good work, on and on it goes ...
When the dust settles, everyone accepts that you always hear distortion in the end result. Otherwise, at least one system would sound identical to another one,, to the "worst" and "best" ears out there.
So, what do you do about it? Some people design and assemble gear so the "blessed" recordings -- yes, all the usual suspects -- sound "magnificent", and all of the lesser cousins sound relatively mediocre, or downright unlistenable to. They've adjusted, manipulated the distortion quotient, spectrum to suit their agenda, which is that only some source types and recordings are worth listening to.
There are many variations and versions of this procedure, tweaking the audible distortion pattern in numerous ways to get the "sound" they want ...
Me, I'm after making every album a powerful and emotionally rewarding experience to listen to, irrespective of how "good" the recording is. Which means that I manipulate the distortion also, happy to admit that, but at least I get the satisfaction when I'm listening to the music, I'm not aware of the system not being "quite right" when certain recordings are on. In my case I focus on reducing those distortions which particularly emphasise the inbuilt flaws of "lesser" recordings, and attempt to make the audible distortion invariant with the volume of playback -- go from headphone to PA sound levels with the "tonality" unaltered ...
My 2 cents worth, ...
Frank
When the dust settles, everyone accepts that you always hear distortion in the end result. Otherwise, at least one system would sound identical to another one,
, to the "worst" and "best" ears out there.So, what do you do about it? Some people design and assemble gear so the "blessed" recordings -- yes, all the usual suspects -- sound "magnificent", and all of the lesser cousins sound relatively mediocre, or downright unlistenable to. They've adjusted, manipulated the distortion quotient, spectrum to suit their agenda, which is that only some source types and recordings are worth listening to.
There are many variations and versions of this procedure, tweaking the audible distortion pattern in numerous ways to get the "sound" they want ...
Me, I'm after making every album a powerful and emotionally rewarding experience to listen to, irrespective of how "good" the recording is. Which means that I manipulate the distortion also, happy to admit that, but at least I get the satisfaction when I'm listening to the music, I'm not aware of the system not being "quite right" when certain recordings are on. In my case I focus on reducing those distortions which particularly emphasise the inbuilt flaws of "lesser" recordings, and attempt to make the audible distortion invariant with the volume of playback -- go from headphone to PA sound levels with the "tonality" unaltered ...
My 2 cents worth,
...Frank
What you appear to be seeking, Fas42, is the equivalent of an old Cadillac sort of design, or pure listening comfort, no matter what. This is a good direction, but it is not my direction in audio design. I want a revealing and clear, effortless design, not a super smooth design. I want the edges to show, and the hidden qualities of a recording to come out in an astounding way. That is different from your path.
- Status
- Not open for further replies.