I believe some of you are misreading ticknpop's post. He acknowledges the importance of accuracy/matching for resistors when used in a diff amp or instrumentation amp.
As to incorporating a discrete buffer within an op amp feedback loop, one motivation is to reduce the thermal load on, and potential thermal distortions in, the op amp. Now, op amp designers have done an exquisite job of managing huge swings in chip temperature by careful symmetrical layout, so these concerns are very much smaller than the "early days", when it was not uncommon to see a thermal "tail" on large signal responses. How much of these would be relevant to audio is another issue, although for small geometries the thermal time constants can reach well into the audio range and even beyond.
As to incorporating a discrete buffer within an op amp feedback loop, one motivation is to reduce the thermal load on, and potential thermal distortions in, the op amp. Now, op amp designers have done an exquisite job of managing huge swings in chip temperature by careful symmetrical layout, so these concerns are very much smaller than the "early days", when it was not uncommon to see a thermal "tail" on large signal responses. How much of these would be relevant to audio is another issue, although for small geometries the thermal time constants can reach well into the audio range and even beyond.
So you think "construction details such as steel materials, resistive material, termination etc" are more important than the CMRR and gain variations "once a reasonable tolerance like 1% is met"?
An inverting amplifier built with 1% resistor can have, worst case, about 0.17dB relative gain variation 20*LOG((1+T)/(1-T)), which is about twice the usual gain matching in a "is there any difference" blind test. Add another gain stage, and you are at 0.35dB, something that probably even my father could easily hear.
I'm not saying that better than 1% is an audio requirement, only that resistor tolerances may have a much larger impact than the resistor termination material.
Telling a good story to gullible audio writers. And refusing to use your ears without peeking.
Agree many decades ago, but the LME49990 can swing +/-16V, +/-27mA and drive 600Ohms all day. Doesn't on inspection need any help!
When do I ever talk to audio reviewers? YES, I am a peeker, but mostly I let others listen to my designs for me, AFTER I have designed them. Still, I avoid steel if possible, worry about circuit board material, what brand of resistors, caps, etc. That is how I give added value over my design topology.
If you want one of my designs without the added value, buy from China. Great bargains, good measurements, pretty good sound.
If you want one of my designs without the added value, buy from China. Great bargains, good measurements, pretty good sound.
100 up they are as low as 1.8c at Mouser and even cheaper in higher volumes.
1 up around 8c
Mouser Electronics, Inc.
Cheap and very good. I prefer placing 1206 to leaded resistors any day.
You not read stereophile in the last 25 years?
Surely worrying about the layout and the performance of the passives would be better? FR4 is so good these days up to around 2GHz that there are few justifications for anything more exotic at audio. Unlike 30 years ago when it was not much better than canoe material.
PCB designers ruleMAX got any measurements of the crystals before and after fixing the case to the boards 0V, its something that use to be done a lot mainly for EMI reasons... Not done as much these days (requires manual intervention...) but more designs use crystal oscillators, where the case is already grounded.
https://la3pna.wordpress.com/2014/12/09/any-benefit-to-grounding-crystal-cases/
Some measurements for you at silly frequencies... wonder how on a well engineered design this would have an effect, but if you are drilling through the PCB its not likely to be a multilayer so is a flawed design to begin with. Just curious whether it would have such an audible effect considering the low frequency of operation and levels involved... I have my doubts
Like I said this mod was on consumer/midfi CDPs so by definition single layer board usually phenolic, and sighted final testing all in a few minutes so no time/equipment for measurements.
Dan.
Interesting find, they are actually cheaper than the equivalent 1206 Yageo 1% thick film, very unusual.
Go stock on them while they last, it's definitely worth of.
No doubt it swings the amps and the volts, but it does get hot, and I don't know how small the shifts in other parts of the chip are. As I say modern layouts are very clever, interdigitated whatnots and so forth, so the effects are probably negligible. But one way to assure this negligibility is to offload the big currents to external stages that are not coupled as closely from a heating standpoint.
Amusingly, I just located a modern part that had the power supply connections reversed by letting it heat up and feeling around for the offending IC, in this case an OPA2134, a pretty decent JFET input dual. Five watts of input power to the fairly small board made it easy to find.
I assume the Nat Semi guys who designed this as the 'ultimate audio op-amp' (not the first nor probably the last), were told that it should work to all specs into 600 Ohms, including thermal. As it's EOL not really a worry but not sure what you hook your line out to that is lower that 600 Ohms, unless you are trying to drive headphones with something that was not designed to do that. Unless you've had a few bad experiences with over optimistic data sheets I'm confused.
Whilst I still fancy trying out ADSL line drivers like Scott and Walt's composite paper, I realise its ridiculous overkill for lols. Which is of course why I want to do it
Reading a magazine like Stereophile or TAS doesn't mean that I am on first person relations with the reviewers or writers. I know most of them from audiophile shows that I sometimes go to. They give me NOTHING, not even a free subscription to the mag, and I don't give them anything either.
However, I like to see where I stand with the reviewers, and at the moment I have about 7 products that are listed as Editors choice in one of them and something similar in the other. It is THEIR CHOICE not mine, but it is good feedback that I am on the right track. I just love the MEASUREMENTS of my designs in Stereophile so much that I rarely make most of the measurements myself. Why bother?
However, I like to see where I stand with the reviewers, and at the moment I have about 7 products that are listed as Editors choice in one of them and something similar in the other. It is THEIR CHOICE not mine, but it is good feedback that I am on the right track. I just love the MEASUREMENTS of my designs in Stereophile so much that I rarely make most of the measurements myself. Why bother?
We use the LME 49990 as a low priced replacement for the AD797 that I originally designed into the JC-3 phono stage. Both sound OK in the circuit, another IC was rejected for sonic reasons.
These IC's will sort of drive 600 ohms, and my 20 year old Acura will probably do 120 mph or so, but not as smoothly as a Bentley. '-) Just because an IC can do it, does not mean it is effortless in the effort. This is a good place for a BUFFER, either discrete or integrated. OPPO could use a few buffers as well.
These IC's will sort of drive 600 ohms, and my 20 year old Acura will probably do 120 mph or so, but not as smoothly as a Bentley. '-) Just because an IC can do it, does not mean it is effortless in the effort. This is a good place for a BUFFER, either discrete or integrated. OPPO could use a few buffers as well.
OMG = PPM
This by itself would make for drooling forum members.
Would this make our poorly mastered originals sound that much better ?
I have my favorite 80V/us amp , I like it better than my 200v/us input stages.
So , my poorly mastered originals prefer a "poorer' engineered topology.
Or , the PPM and 200V/us slew matters not with my poorly mastered original.
But , one could assign marketing terms to this , (like ovature e) and make
it seem like magic.
This by itself would make for drooling forum members.
Would this make our poorly mastered originals sound that much better ?
I have my favorite 80V/us amp , I like it better than my 200v/us input stages.
So , my poorly mastered originals prefer a "poorer' engineered topology.
Or , the PPM and 200V/us slew matters not with my poorly mastered original.
But , one could assign marketing terms to this , (like ovature e) and make
it seem like magic.
Last edited:
I doubt there are many consumer audio amplifiers built with better than 1 % resistors even in critical input stages - some pro models posssibly with resistor networks or high accuracy surface mount - and feedback resistors where most are undersized with high ppm drift and suffer from thermal cycling of their value..
Its a good thing I don't live in Salem, just saying
Its a good thing I don't live in Salem, just saying
Last edited:
Feedback dividers can get a little hairy if you're crowded for noise.
All good fortune,
Chris
- Status
- Not open for further replies.