I'm not quite sure, Dejan, I think I might have to spend a week or two going through all the literature, to determine what should be the correct number of trials to conduct - you don't wanna get this sort of thing wrong, ya know ...
Jan, you are correct. Unfortunately, I don't have any LME4562 IC's and I don't have the circuit built to test dual IC's, at least yet. However I have tested other IC's with loading (not even as low as 100 ohms) and I found significant differences in the IM spectrum in level and order of distortion with loading.
Frank, I am not criticizing anyone’s work and contribution here, I try to spot the limits of realistic testing with these files.
X,Y were uploaded as 48k sample rate files.
This means that single sample accuracy alignment is bound to be 21 us (20.83us).
This limit is set at the recording end and we can’t time align the recorded events to a smaller time step at a later time.
By upsampling the files at the replay end, we can time align to finer time steps the recordings as a whole but not the recorded events themselves.
For to achieve a time step of 1.3us, the recording or the (wrong move) up conversion has to be done at 768k (getting heavily anal about it I would say).
I repeat my opinion. Results from such tests on these files can not serve as supporting evidence in any direction. These files can only be tested acoustically, in a fashion like this:
I have expressed my listening opinion, after playing back 6-7 times X and Y files through headphones but it was an informal sighted test.
George
I suppose that, if i had to make a physical comparison, i would had build two samples of this preamp, one loaded, one not, and recorded the four tracks simultaneously for no clock differences ?
The differences you are finding are mostly due to level mismatch and/or time alignment. It took me quite a bit of work to get the null I did.
You need to get the levels right and especially get the timing right for a null. And there seems to be a drift of about 2 samples over the length, not that a drift like that would affect the midrange.
When I did get everything lined up, the null was deep. I had to amplify the result by 40dB to hear it. Then it was mostly hiss, with a some of the loud musical parts sticking thru. The noise is easy to explain, the other parts not. But it's all so far down that I really doubt it could be heard - at least not with a casual listen.
I agree with John Curl, looking at harmonics of sine and dual sine would be very interesting. I would expect to see differences. I tired to extract those for the recording, but could not.
After the first decent burn, you will assume you'll burn your arm every next time.
And gone is your objectivity, with the wind.
John I know how that feels, it seems to have to do with having a long history of testing, testing, testing. I find myself procrastinating (sp?) much more than I used to.
Yet, when I do get up (figuratively) and hook up a simple DIL8 socket to the test unit I find that I do in 30 minutes which I have postponed so long...
Jan
Last edited:
Haha, yeah there was a lot implied, but the important one 'do you hear a difference' was NOT implied nor asked - in fact, you were TOLD there is a difference.
One guy didn't even reply to any question but waxed lyrically about how lifelike # X (iirc) sounded.
jan
I took delivery of a used dScope III a week ago. Sort of poor-man's AP, very similar.
I must say, all those options and functions are daunting.
But hey, now that I shelled out the dough, I better start using it!
But it will take an investment in time before I get something useful out of it.
There's just no other way.
Jan
I'm hoping to re-run this test in the next couple of days. I've actually built a test circuit, all soldered, star grounded (which I'm hoping will prove successful) and have two highly regarded opamp types at my disposable to try out. These are singles, not duals.
I would run the test along similar lines but would also include (if you think it would be of benefit) some 0db test tones including a 250Hz + 8020Hz and 11kHz + 12kHz IMD test track. So those of you that like using software to examine the detail could analyse the distortion spectrum of each and then try and correlate that to a listening test.
(And I've been reading all your posts... which are informative and just, well interesting 🙂)
I would run the test along similar lines but would also include (if you think it would be of benefit) some 0db test tones including a 250Hz + 8020Hz and 11kHz + 12kHz IMD test track. So those of you that like using software to examine the detail could analyse the distortion spectrum of each and then try and correlate that to a listening test.
(And I've been reading all your posts... which are informative and just, well interesting 🙂)
I don’t understand what you mean by “physical comparison” Esperado, is it detailed analysis?
Simultaneity in time per your suggestion (although different from sampling clock difference) can assure time synchronization, timing and amplitude alignment and if preamplifiers are supplied from the same psu, can ensure also psu noise cancellation during the subtraction process
All these, provided recording done by a single, true (one AD per input) multi channel device
I hope that Mooly will perform his new recordings in this way (for to ease the set-up, why stereo? Single channel is OK for such test)
Thank you Pano for clearing this up.
George
Sorry, English, age, all those sort of things: ;-)
I Wanted to mean measurements by intruments or softwares. But you understood.
- Status
- Not open for further replies.