Seems like JC said something recently about low order harmonic distortion is much less objectionable than higher order. So, maybe the question should be reframed: how do you allow a little low order harmonic distortion while suppressing higher order, in relation to use of high V rails?
I suspect he may say that at least one symmetrical circuit topology works pretty well for him, but let's see.
I suspect he may say that at least one symmetrical circuit topology works pretty well for him, but let's see.
Come on guys, improving a design, all else being equal, through higher voltage rails, can't be too hard a concept to accept.
I started off at Ampex with a single regulated 28V supply. Then, with Mark Levinson (JC-2) +/- 15V for line and +/- 13V for the phono stage. Why those voltages? Because we could buy a COMMERCIAL AC-DC power supply virtually off the shelf.
When I went to building custom power supplies, then higher voltages became possible, up to a point. During product development, I used an adjustable lab power supply that went up to +/- 30V, so I could troubleshoot a design at perhaps +/-15V, and then move it up, once it worked OK. Invariably, the distortion went down, even at nominal output levels, like 1-2V as the supplies went up. One of the reasons was the reduction in the gate-drain capacitance with somewhat higher voltage and then even a relative flattening of the capacitance above 10-15V or so. However, the gate leakage will rise, making extremely high voltages problematic. So +/- 24--32V is what I usually select, then I can place a cascode device in between the supply voltage. This is optimal for both lowest distortion, and reasonably low gate leakage.
Now, I don't ever add even order distortion, on purpose, but even my most balanced topologies have a little even order distortion, because of the differences between complementary parts. Most of the time, today, I just leave the residual even order distortion, instead of going to great extent to null it out in some way, but this is not always so. Recently, I designed a new driver stage for the JC-1 power amp that is more symmetrical, and I will see if it makes any audible difference, in future. The amp will measure better, that is for sure.
Still, IF you don't want to make improvements in your designs, and are happy with most IC's, why bother me about it?
I started off at Ampex with a single regulated 28V supply. Then, with Mark Levinson (JC-2) +/- 15V for line and +/- 13V for the phono stage. Why those voltages? Because we could buy a COMMERCIAL AC-DC power supply virtually off the shelf.
When I went to building custom power supplies, then higher voltages became possible, up to a point. During product development, I used an adjustable lab power supply that went up to +/- 30V, so I could troubleshoot a design at perhaps +/-15V, and then move it up, once it worked OK. Invariably, the distortion went down, even at nominal output levels, like 1-2V as the supplies went up. One of the reasons was the reduction in the gate-drain capacitance with somewhat higher voltage and then even a relative flattening of the capacitance above 10-15V or so. However, the gate leakage will rise, making extremely high voltages problematic. So +/- 24--32V is what I usually select, then I can place a cascode device in between the supply voltage. This is optimal for both lowest distortion, and reasonably low gate leakage.
Now, I don't ever add even order distortion, on purpose, but even my most balanced topologies have a little even order distortion, because of the differences between complementary parts. Most of the time, today, I just leave the residual even order distortion, instead of going to great extent to null it out in some way, but this is not always so. Recently, I designed a new driver stage for the JC-1 power amp that is more symmetrical, and I will see if it makes any audible difference, in future. The amp will measure better, that is for sure.
Still, IF you don't want to make improvements in your designs, and are happy with most IC's, why bother me about it?
And goodness knows opamp design hasn't sat still. Even if many of the newer advantages are less important for home audio (e.g. power). It's just patently ridiculous to keep bringing up issues long since dealt with 40 years ago.
I half wonder if Robert is running some sort of fancy IFTTT algorithm to interject comments about slew rate whenever a topic of opamps comes up. Hey, we all have our things, I guess.
And yet so bereft of theoretical or real (as in from some form of DBT) justification. You know it gets old, right?
I tried an opamp recommended on this thread. I ran it through my evaluation process and posted about it here. Since none of you here actually try a design with high speed opamps; what am I left to think about a post like yours right now?
I'll send you some LT1358's to play with gratis. I will be here exhibiting and you are welcome to come and have a listen.
So? Your move next.
The claim is on you to prove, Robert. And doubly so given you're also (at least by the mechanism you're pushing) saying that a lot of our basic understanding of signals and systems is invalid.
Prefer what you may, but audio signals don't really care about *that* aspect of your beloved opamps. Anyhow, good luck at your audio show.
Prefer what you may, but audio signals don't really care about *that* aspect of your beloved opamps. Anyhow, good luck at your audio show.
I tend to believe that critical audio sonic differences can be created outside direct measurement of some parameter, but still can related to it, inferentially, so 50-100V/us is a good sign that one IC is better than another for some good reason.
No TIM and PIM, we know this for 40 years now.
Wait, make that 50 years, it's 10 years since this was first mentioned as "40 years ago".
Would be interesting to see a list of the breakthroughs in those 40 years,
especially those universally accepted.
- Status
- Not open for further replies.