It does indeed. But it's looking increasingly like Joe is not going to step up and do his part to have this done.
On the contrary. I think it quite accurate to suggest that you've been Joe's most persistent critic. As such, he might reasonably be concerned of bias affecting a test conducted by you. Such a reciprocal concern would fair, as you have long been suggesting bias error on his part. Bias doesn't have to be conscious, as I'm sure you would agree.
To be clear here, I have no idea whether Joe is or isn't concerned about bias on your part, but I think that such a concern is fairly raised by some of your comments.
Like I said, when I put results out, I disclose EVERYTHING. If you or anyone else thinks there's an error, there's enough info for replication. When I've done stuff like this before, people always have lots of suggestions on other things to look for- when they make sense, I do them. I don't rig tests- you may not like me or my ultra-rationalist style, but my reputation is for presenting my data honestly and fairly, and if you would ever catch me doing such a thing, you're welcome to go after me for it.
My criticism is focused on the unwillingness to back up claims with data. Here's a golden opportunity for Joe to do so.
My criticism is focused on the unwillingness to back up claims with data. Here's a golden opportunity for Joe to do so.
I can only say what I start with- from there, the results I get guide further work. Harmonic and intermod distortion spectra at different frequencies and different levels, frequency response, multitone measurement, noise with and without signal, that's where it starts. Comparisons are made to the control sample (which initially should match the DUT, but that has to first be confirmed).
That's the start. Where it goes from there will depend on the results seen- is there anything unusual or unexplained? That's where we look.
Until Joe asks someone to do the listening test, I can't really say what the protocol is other than double blind, matched level, and controlled. That person can pick the test format and source material, control switching and master level, and take whatever time he wants. He will need to confirm that under sighted listening, he can hear the differences. I'll be guided by a 95% confidence. If the differences are really as "dramatic" as claimed, that should be easy.
edit: Diffmaker will be of more than a little use.
That's the start. Where it goes from there will depend on the results seen- is there anything unusual or unexplained? That's where we look.
Until Joe asks someone to do the listening test, I can't really say what the protocol is other than double blind, matched level, and controlled. That person can pick the test format and source material, control switching and master level, and take whatever time he wants. He will need to confirm that under sighted listening, he can hear the differences. I'll be guided by a 95% confidence. If the differences are really as "dramatic" as claimed, that should be easy.
edit: Diffmaker will be of more than a little use.
Hello,
I have classic TPA combo BIII on top of a IVY III and I'm willing to try this with 1uF and 2x 3.3ohms. IVY III already has room for these resistors and since I don't use voltage inputs I will put them on R32-R35, replacing original values. I'm going to lift up BIII board some more with additional distancers to disconnect current inputs and use voltage inputs from the same BIII output terminals. On BIII there is also enough room to put1uF capacitors. However I only have some 5mm MKT types, which will fit perfectly. Couple of 1uF MKPs that I have are axial and I will try them also.
Then, there is a question about 0,33F caps, couple of which I bought in local electronic store (I didn't expect this and actually they carry more capacitance values beside 0,33F, panasonic). BIII has four(five, but I removed clock power supply reg) regulators and I wonder where these caps will exhibit most (positive) influence. Anyway I will tried them after the first mod.
Also, I'll try to find capacitor value which will add parallel to R17-R20 to correct frequency response roll off.
Best regards to all.
I have classic TPA combo BIII on top of a IVY III and I'm willing to try this with 1uF and 2x 3.3ohms. IVY III already has room for these resistors and since I don't use voltage inputs I will put them on R32-R35, replacing original values. I'm going to lift up BIII board some more with additional distancers to disconnect current inputs and use voltage inputs from the same BIII output terminals. On BIII there is also enough room to put1uF capacitors. However I only have some 5mm MKT types, which will fit perfectly. Couple of 1uF MKPs that I have are axial and I will try them also.
Then, there is a question about 0,33F caps, couple of which I bought in local electronic store (I didn't expect this and actually they carry more capacitance values beside 0,33F, panasonic). BIII has four(five, but I removed clock power supply reg) regulators and I wonder where these caps will exhibit most (positive) influence. Anyway I will tried them after the first mod.
Also, I'll try to find capacitor value which will add parallel to R17-R20 to correct frequency response roll off.
Best regards to all.
Bingo!
True, true.
Like inventing vapour DACs that I am supposed to supply? Where did that come from?
As I am now accused by Stuart of evasion, then I can hardly be assured of a 'fair trial' - so maybe Stuart should do the right thing and exclude himself from conducting any tests where he is banking on a outcome that he favours? What's the saying? Only a fool puts his head into the mouth of a lion.
If Stuart could give some indication that he is truly interested and has genuine scientific curiousity, rather than behaving like a cop on the beat or acting as a prosecutor, then perhaps I would be more inclined, nay happy, for us to arrange a fair trial (we would need to find suitable DACs still). I could then look upon it as an opportunity rather than a dangerous proposition.
Now he may find it unpalatable that I should question his motives, but he is in a glass house as has questioned mine. We can all see that, right?
But really, where is all this banter going?
1. I am quite willing to submit to a fair test - by a fair person.
2. I think for the short term, trying to get a measurement would be worth a try, with a little bit of help. This may need an outlay of a few hundred dollars.
3. In the meantime there IS another test can be done without any delay, try any of the scenarios in the first six posts. Dare I say it on a DIY forum? Warm up your soldering irons?
4. For listening tests to be possible, we need to find the right test bed, that is, a pair of identical suitable DACs. That's not so easy.
I don't know about any of you guys, but I usually only buy one of most products I purchase, so this one is a bit tricky.
A DAC with USB and perhap SPDIF input too, would be OK. Perhaps a player ? They have generous internal space. But whatever, it would be one that is based on B-B 179x series "current" DAC, Preferably the I/V and post-DAC opamps to be LM4562 as I know they work well with what I have in mind. Perhaps existing lower quality opamps can be replaced, LM4562 are available in DIP and SOIC. We also need enough internal real estate to add components - so tiny USB Pen DACs are out.
Maybe something like a HRT Music Streamer:
I believe some HRTs use PCM1794A DACs. There should be just enough space internally. But other HRTs use "voltage" PCM1793 and they I believe are the budget models, not so good.
The work that will be done to one of them will not in any way change the way it operates or the final output level to be exactly the same. At the toss of a coin, one will have a label "A" on it and the other a "B". No other telling marks.
Once again, for the record, Stuart is quite wrong about me being evasive, nothing could be further from the truth. Just wary and careful.
Cheers, Joe
-
Last edited:
The MKT will do the job. This is ES9018 Sabre DAC into virtual ground, right? Then I have tried the 2 x 3R3 plus 1uF into LM4562 I/V and it works well.
The 0.33F, if they have a larger value, still make sure that they are 5.5V and also they must have ESR not lower than about 40R and 100R max. The ones I use are typical and 75R.
Trust me, that is the least important. And if you use JRiver, then the EQ correction can be done there, see earlier Post #6.
Cheers, Joe
-
This was an unfortunate, yet wildly predictable result of this thread.
How about this: what would make a fair trial such that it would be an opportunity rather than a dangerous proposition? SY laid out teh starting point of his testing criteria already, at least on the bench side of the spectrum (ah, I joke, I joke), and would be administering the ABX, not acting as a testee.
If he's at all like me with this subject, he's probably more interested in (and I shouldn't put words in his mouth):
1.) Is there an audible effect? Is there something significant electrically going on? Whether or not it's neutral, beneficial, or detrimental is another story. (And, probably, at least for younger ears).
2.) Closure of this subject. Of course, when something reaches this level of noise, it's liable to be fact-resistant (again, neutral, beneficial, or detrimental).
This seems to be what led to the confusion. Sounds like you have something to test, but can't ship it. Stuart did offer to pay shipping and others offered to help, as well. We thought you had something to ship.
Just to be sure, you have no DACs that could be tested, or at least not shipped to North America?
Yes I kinda gathered there was some confusion. I don't really have anything here that fits the bill.
I have my Oppo 105 here, but it isn't stock and hence not an option. But I still do measurements on it as it is laid out so well to do it. I will have guidance from Steve in using ARTA and also generating stimuli/files. I think Steve would also like to find something for him to test. Locally, Terry Demol is also welcome to become involved. The more collaboratively we can make this, so much better the chances.
I have downloaded ARTA and it is working nicely. I just used it to test my internal motherboard Realtek soundcard and its distortion isn't low enough according to ARTA's THD+N test should 0.1% or better and I got 0.18% is not good enough. ARTA SAYS 0.1% is usable (not really) and and 0.01% is good. ARTA's own sample is 0.005% and I know what card they used to get that result, so will try to hunt it down.
I also have a ClioFW setup and that can be used as a spectrum analyser - but using it's own software. I would prefer ARTA as it is free (but cannot save files unless you pay) and hence anybody else will be able to try repeat the tests.
I think the latter is important, especially if it actually shows up something cool.
Cheers, Joe
-
I was thinking to the same measurement approach: one stereo channel normal setup, while the another one modified. You already did it. Interesting results so far. Quite predictable the altering of the frequency response. This is a obvious direct consequence of applying this filter.
I intend to measure this way on my system, when the circumstances it will allow me.
There is so an accepted way to do it? Comments?
- Status
- Not open for further replies.