Yup. Bear, please, controls are necessary. As long as they are negative.
(Positive controls are best avoided, anyway, as they will not be "appropriate" in this test. It's not really important to find out whether the listener could hear differences others have proven to hear in a controlled test).
http://www.diyaudio.com/forums/digi...native-post-dac-filtering-32.html#post4462498
WTF are you talking about??
It's a proposed piece of hardware.
What "controls" do you want??
A straight piece of wire is not good enough??
You take the box, do whatever tests you wish - control yourself any way you wish.
There is no specific "TEST" specified.
It's a box a piece of HARDWARE intended to increase the measureability of opamp distortion.
If you think the string of 10 will NOT increase the distortion, then say so, and state why.
After that you can do whatever you wish. I proposed measuring it first.
Are you afraid what the measurements might show?
Then listening to it.
AND then if people like YOU just happen to not lie and they hear a difference
between each "channel" THEN perhaps more "controlled" tests (which are more difficult to set up and more time consuming) could be considered.
Aka, one step at time.
And, although it is obvious, self-evident, to some extent this is a crude approximation of the cascaded signal path found in most recording situations, that being multiple opamps, apparently this is an elusive concept? Of course here the aim is to use identical opamps in the (vain?) hope that a "sonic signature" will emerge that even those who have difficulty hearing such things will find it audible and clear. (obviously, a real recording chain is more complex and the opamps are a mish-mash of types and brands - we're not ready to look at that, this is more basic).
Ok now, Stewart?
So, what's the problem?
Are you afraid you and others might detect audible differences that do NOT go away when DBT'd??
And, should it not "work" and show nothing, chalk one up.
Last edited:
Hehe. That might be THE point to mandatorily introduce a positive control. Never hearing a difference ist just too easy...
Consumer protection genes. Know it all, and especially "knowing what is good for others" approach.
Whatever. It is annoying, when others want to state what's good for you, but you have to get over it. It happens all day, and tons of (our) money are spent to fund research which will tell us what we have to do or not.
If what you want is to not have any technical criticism when half-baked ideas are tossed out, this is not the right forum for you.
Nothing terribly remarkable about any of that--of course that's what's going on. So instead of a totally synthetic work-up that looks scary, what's the harmonic structure of ____ opamp driving into ____ load to achieve that kind of current demand?
P.S. Something like a OPA1642 has ~40 dB PSRR at 100 kHz, too. PSU will need to be characterized, but with the bypass, it'd be pretty safe to say < 1 ohm throughout the range of interest.
Bear, let's step back. What do you want to test? I.e. what's the hypothesis being presented? Perhaps from there, a clearer experimental design can be made. This, as it is proposed, has no clear start or end goals.
As far as I can tell, you want to switch between a wire and 2 different 10 opamp-long unity gain chains. You wish to characterize the response of said chains, then circulate a box containing said 2 circuits and some sort of switching hardware to allow the user to pick between A and B on their own time and develop a preference.
It's simply the additional harmonics from driving a 15 kHz sinusoid through a perfect class-B OPS (as referenced to one rail).
My guess was that he wants to compare a wire with a chain of 10 opamps with total gain 1, both from a "measurement" and "listening" perspective. And have some fun in the process.
I would only suggest that, depending on the source and load characteristics, the wire may be "worse" in the sense that the opamp chain could perform some "better" impedance matching / driving. Up to Bear to figure out how he wants to address this.
Daniel (Dr?),
The PSRR and the CMRR are often intertwined by real world circuit impedances. Spray noise around and something else picks it up.
To get 1 ohm of bypass impedance at 100,000 hertz you would need in theory 1.59 microfarads of capacitance. Now when I look up http://www.mouser.com/ds/2/293/e-upj-876311.pdf the 100,000 hertz impedance for a typical 1.5 uF capacitor it is closer to 10 ohms impedance.
If you look at how PSSR is measured for the manufacturer's data sheets it is really a bit tricky and the numbers shown will not be reached in most real circuit layouts.
Bear, it sounds like to you trying to demonstrate that many people can hear a difference between different opamps in some particular case, the one case you propose.
However, the argument that has been going on here is not that different opamps produce different distortion characteristics. The argument is that is more like this: In a simple one or two stage line level audio preamp, the distortion of opamps is so tiny, no human could hear it.
Therefore, some people seem to think you are trying to demonstrate something that has not been at issue here.
Also, people have been asking that you state a question you are trying to answer with your experiment. For what you have proposed so far, it sounds like your question might be something like: What percentage of forum participants can perceive a difference in the sound of 10 opamps cascaded in series? ...Something like that, you could clarify if you are trying to find an answer to some question or not.
Then again, maybe you are just trying to give people and opportunity to hear something for themselves that they may not have heard before. If so, fine with me, but some people here will point out that even if people do hear something different between opamps with your circuit, they still could never hear it in a simple line level audio preamp. So, to them, you will not have done anything to resolve what they perceive as the basis of the ongoing dispute. Something like that, maybe they can elaborate.
However, the argument that has been going on here is not that different opamps produce different distortion characteristics. The argument is that is more like this: In a simple one or two stage line level audio preamp, the distortion of opamps is so tiny, no human could hear it.
Therefore, some people seem to think you are trying to demonstrate something that has not been at issue here.
Also, people have been asking that you state a question you are trying to answer with your experiment. For what you have proposed so far, it sounds like your question might be something like: What percentage of forum participants can perceive a difference in the sound of 10 opamps cascaded in series? ...Something like that, you could clarify if you are trying to find an answer to some question or not.
Then again, maybe you are just trying to give people and opportunity to hear something for themselves that they may not have heard before. If so, fine with me, but some people here will point out that even if people do hear something different between opamps with your circuit, they still could never hear it in a simple line level audio preamp. So, to them, you will not have done anything to resolve what they perceive as the basis of the ongoing dispute. Something like that, maybe they can elaborate.
Ed, yes, you'd want to bypass with something like a 100 nF ceramic (which is almost universally recommended, as close to the pins as possible) + 10 uF electrolytic.
But your point to the overall complicated nature of the interconnections is why real-world tests are needed.
But your point to the overall complicated nature of the interconnections is why real-world tests are needed.
Mark, note that I published my results for the experiment asking the question, "How many IC op-amps (in this case OPA134) configured in unity gain needed to be placed in series before I can distinguish the chain from a straight wire, ears-only?"
The answer in that case was about 6. Test protocol, described in my article, was paired forced choice.
The answer in that case was about 6. Test protocol, described in my article, was paired forced choice.
Jan, "THD" is merely the label that LTSPICE attaches to a number it computes during Fourier analysis.
What the computed number is, in fact, is the RMS average of the harmonics (omitting the fundamental).
The printed value "61.206% THD" in post #2279, is just the RMS average. I copied harmonics 2 thru 11 into Excel to show the calculation, see below.
Attachments
You would need to build a series of individual PCB's optimized for each op-amp being tested with the ability to instantly switch between them. If output levels were equal you might still notice differences but i recon they would follow specs. Ultra low distortion op-amps might sound better (to some) but others who are used to a bit of distortion may well disagree.
It is. The pull down resistor version was ABX tested at Alesis in the late 90's. Greg Timbers of Harman Everest K2 fame uses it too.
- Status
- Not open for further replies.