Edit... files now available HERE
The listening tests we have run in the past have been both fun but also (I hope) informative and insightful. Well I figured it was the turn of opamps again but this time I want to do things very differently. Firstly, I know that being presented with a bunch of files is off putting, and for those of you who are juggling other things in your life, very time consuming.
So what is this test all about and why is it different ?
Well, there has been a resurgence of opamp threads/posts in the last few weeks and I thought it would be a good idea to see if we could really make a real attempt to try and pin sonic differences to a handful of common devices, some highly regarded, others less so. To make the differences stand out (if there are any), the circuit consists of a chain of five identical devices in each channel, each device being wired as a unity gain buffer. To make it even more severe, each of the first four opamps in the chain is loaded with a 390 ohm load. Now lets put that load in perspective. The max peak voltage that each could theoretically see in this test is going to be around -/+3 volts given that a CD player is specified as a nominal 2 vrms output. Give or take. That means peaks of around 7.5ma worst case.
This is the basic circuit.
However, there is more to this test than meets the eye because I'm hoping to see if it ties in or relates to the fact that we sometimes prefer one thing over another simply because we 'know' something must be better, or that we can see what we are listening to and so that influences our choice.
What ! So there's still another zillion files to listen too
No chance.
So here's the plan. All files will made available for those who are keen enough to want to try them all. For those who would like to compare just one or two devices against another (and you can each choose the devices from the list) then I will give you your own personal link to just those files you want to compare. Of course no two of you will get the same file name, I'll randomise each personal set (so you wont be able to compare notes that way)
(We would have to see how all that worked out... because its all extra work and time for me... but for those of you who have been interested in the other tests, well its probably a managable number)
To make it even more interesting... there are no files at all on offer just yet.
Huh ! So how does that work then ?
It works by you being able to look at the list of devices offered and saying how you think they might differ. We have all had it drilled into us over the years how good certain devices are, and how mediocre or downright poor are others. So in the first instance think about what you think you might hear. (You probably have your own ideas, I know I did). Then you can ask for the devices you are interested in and see if the listening reality matches your opinions, and to see if you really can pick one over the other. There is also a level matched .wav rip of the original as something against which a final judgement can be made. That would not be made available until all the initial listening was drawing to a close.
The list.
1/ JRC4560
2/ LM833P
3/ LM4562
4/ NE5532
5/ TLE2072
6/ TL072
So ! What do you all think. Interested ?
The listening tests we have run in the past have been both fun but also (I hope) informative and insightful. Well I figured it was the turn of opamps again but this time I want to do things very differently. Firstly, I know that being presented with a bunch of files is off putting, and for those of you who are juggling other things in your life, very time consuming.
So what is this test all about and why is it different ?
Well, there has been a resurgence of opamp threads/posts in the last few weeks and I thought it would be a good idea to see if we could really make a real attempt to try and pin sonic differences to a handful of common devices, some highly regarded, others less so. To make the differences stand out (if there are any), the circuit consists of a chain of five identical devices in each channel, each device being wired as a unity gain buffer. To make it even more severe, each of the first four opamps in the chain is loaded with a 390 ohm load. Now lets put that load in perspective. The max peak voltage that each could theoretically see in this test is going to be around -/+3 volts given that a CD player is specified as a nominal 2 vrms output. Give or take. That means peaks of around 7.5ma worst case.
This is the basic circuit.
However, there is more to this test than meets the eye because I'm hoping to see if it ties in or relates to the fact that we sometimes prefer one thing over another simply because we 'know' something must be better, or that we can see what we are listening to and so that influences our choice.
What ! So there's still another zillion files to listen too
No chance.So here's the plan. All files will made available for those who are keen enough to want to try them all. For those who would like to compare just one or two devices against another (and you can each choose the devices from the list) then I will give you your own personal link to just those files you want to compare. Of course no two of you will get the same file name, I'll randomise each personal set (so you wont be able to compare notes that way
)(We would have to see how all that worked out... because its all extra work and time for me... but for those of you who have been interested in the other tests, well its probably a managable number)
To make it even more interesting... there are no files at all on offer just yet.
Huh ! So how does that work then ?
It works by you being able to look at the list of devices offered and saying how you think they might differ. We have all had it drilled into us over the years how good certain devices are, and how mediocre or downright poor are others. So in the first instance think about what you think you might hear. (You probably have your own ideas, I know I did). Then you can ask for the devices you are interested in and see if the listening reality matches your opinions, and to see if you really can pick one over the other. There is also a level matched .wav rip of the original as something against which a final judgement can be made. That would not be made available until all the initial listening was drawing to a close.
The list.
1/ JRC4560
2/ LM833P
3/ LM4562
4/ NE5532
5/ TLE2072
6/ TL072
So ! What do you all think. Interested ?
Mooly, great initiative.
Frankly, I don't expect to hear much difference at all with music, if any, but I would like to see this bias confirmed Other than that, the JRC will sound cheap, the LM833 fresh and updated, the LM4562 clean and analytical and it will pick up your cell phone before you do, and the NE5532 will sound bog standard HiFi for the masses.
The filthy ground will affect all op amps similarly. I would expect some audible differences to show up with dirty supply lines, but even then only when playing silence.
Frankly, I don't expect to hear much difference at all with music, if any, but I would like to see this bias confirmed
Other than that, the JRC will sound cheap, the LM833 fresh and updated, the LM4562 clean and analytical and it will pick up your cell phone before you do, and the NE5532 will sound bog standard HiFi for the masses.The filthy ground will affect all op amps similarly. I would expect some audible differences to show up with dirty supply lines, but even then only when playing silence.
Last edited:
Mooly, you sure have good idea making this thread and the LTSpice.
I amp very interested in this. I don't have a preference because I never even try before to form an opinion. I bought the LM4562 just because everyone here told me. I never even get to listen to it yet. So I should be the perfect subject for blind test.
Particularly from my experience sneak oil really does not work on me, commercial does not work for me either. I listen and just judge it. Case in point, my most favorite amp is YBA, when I look at the schematic and show it here, I got nothing by negative comments. In contrast, I was not impressed by Mchantosh nor Krell where they are regarded highly. I think I can be very impartial.
Post the files like soundtrack1, soundtrack2....... and just pure blind test.
I amp very interested in this. I don't have a preference because I never even try before to form an opinion. I bought the LM4562 just because everyone here told me. I never even get to listen to it yet. So I should be the perfect subject for blind test.
Particularly from my experience sneak oil really does not work on me, commercial does not work for me either. I listen and just judge it. Case in point, my most favorite amp is YBA, when I look at the schematic and show it here, I got nothing by negative comments. In contrast, I was not impressed by Mchantosh nor Krell where they are regarded highly. I think I can be very impartial.
Post the files like soundtrack1, soundtrack2....... and just pure blind test.
May I make another suggestion? Do a blind test with a pull down resistor to -V to force the amp to run at pure Class A. It should be easy as you only run +/-3V max, a 5K pull to -15V should be good enough.
Far as all the circuit of opamp they posted, all are push pull and have crossover distortion. I am curious how much improvement you get by forcing it to run at class A. Of cause blind test.
Far as all the circuit of opamp they posted, all are push pull and have crossover distortion. I am curious how much improvement you get by forcing it to run at class A. Of cause blind test.
Thanks.
You mean its been drilled into you that the LM4562 is the audibly superior device. Which it has with me I must add.
The files are all done and dusted and so any Class A testing would have to be for another day.
I'll probably post the complete set of files tomorrow, and then if any of you want to try just a couple of devices then let me know and I'll give them different file names and send them to you.
Hi Pano,
yes there is a .wav reference file which I'll put later.
I was really interested in seeing if any audible traits in the IC's showed up by being able to compare one vs another, as much as comparing each to a reference. This way the recording chain remains a constant with only the buffer group changing.
yes there is a .wav reference file which I'll put later.
I was really interested in seeing if any audible traits in the IC's showed up by being able to compare one vs another, as much as comparing each to a reference. This way the recording chain remains a constant with only the buffer group changing.
No worry, I am very skeptical about all these, I have absolutely on opinion on opamps at all.
I agree to buy the LM4562 because it's BJT front end, 55MHz, high drive capability, low noise and well spec distortion. 55MHz implies higher loop gain at 20KHz and no worry about slew rate.
I kind of think the opamps might sound closer to each other with the pull down resistor that make the opamps running in Class A.
Last edited:
Yes, of course. But we have to remember that almost 30% of the sound will come from opamps that are NOT what we think we are listening to. Is that fair to the LMxxx opamp when we don't know what else is in the signal path?
I am raining on the parade, but it's important.
Not at all, and I'll try and explain why. And I have to chukle as I see you 'automatically' picked the LMxxx above to express the point, rather than TL0 or JRC
Many rolling opamps often use phrases like 'dark' or such a device was 'to bright' or 'no bass'.
This kind of trial should at least reveal something for those folks.
That's at one extreme of the possible opinions, the other is the far more subtle where for example the poor drive ability of some devices are criticised, or that they are noisy, or have 'obvious' distortion etc. By loading the devices as in this trial we should be able to confirm that the poor drive for example is an obviously audible issue... or not. And this is where you can all ask for the files yourself. If you want to try xxx against yyy then I'll let you have just those two to compare.
Although part of the audio chain remains an unknown, the same is true of the actual material used, and the final amplification and speakers/headphones used to evaluate it all, be it a 1950's recording via a valve amp or Beyoncé via Class D. They are all valid listening chains to evaluate what has been inserted in the middle.
Enjoy
LM was the first thing that came to mind. Tho I understand the above post, I still don't think it's fair to classify the sound of a device when it makes up only 70% of the circuit. Wouldn't the sonic character by only about 70% of what you hear?
In a device like a Behringer DCX, there would be about 7 extra opamps inline, reducing your DUT to 42% of the circuit.
Finding differences is another matter. Change 70% of the circuit and hearing a difference is plausible. Change ~ 40% and maybe not so easy. But if no one can really hear them, then it's moot.
To be fair, we need two control tracks. One straight from the rip file, and one looped only from DAC to ADC.
Tho I understand the above post, I still don't think it's fair to classify the sound of a device when it makes up only 70% of the circuit. Wouldn't the sonic character by only about 70% of what you hear?In a device like a Behringer DCX, there would be about 7 extra opamps inline, reducing your DUT to 42% of the circuit.
Finding differences is another matter. Change 70% of the circuit and hearing a difference is plausible. Change ~ 40% and maybe not so easy. But if no one can really hear them, then it's moot.
To be fair, we need two control tracks. One straight from the rip file, and one looped only from DAC to ADC.
Lets just see how it all pans out because the files are done.
One very real issue in any trial like this is that it always ends up with files being analysed rather than listened to and as such the ripped file will always be detectable. That is something I want to avoid. Of course at the end of the test we can all compare to the rip and see how well or otherwise things are.
One very real issue in any trial like this is that it always ends up with files being analysed rather than listened to and as such the ripped file will always be detectable. That is something I want to avoid. Of course at the end of the test we can all compare to the rip and see how well or otherwise things are.
Full Set Of Files
Here are the files.
FILES: FULL SET
(These are on Dropbox. For those not familiar with this, you do not, need to sign up or register. Just close the window asking if you have or want an account and the 'Download Files' option will be available)
Here are the files.
FILES: FULL SET
(These are on Dropbox. For those not familiar with this, you do not, need to sign up or register. Just close the window asking if you have or want an account and the 'Download Files' option will be available)
I'm most interested with LM4562 because in previous listening test I found LM4562 (Pomegranate) was very dynamics and "high end" but at the same time I feel something is wrong with the sound (so I put it at the bottom). I suspected stability issue or something.
In current test I found one opamp has similar signature. Very dynamics and clean/detail but I feel something is wrong with the sound. The pitch and flow of the music is not good. As if there is phase reversal.
Attached are FoobarABX between C&A, B&D, E&F. All with 8/8 at first trials.
EDIT: removed the ABX pictures because they have opamp names as the filenames.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.