Al
Logged
« Reply #35 on: March 27, 2008, 02:14:10 PM »
| Quote: |
Just thought I'd mention there's a Lavry Gold AD122-96 up for bid on eBay at the moment. I had one of the lower bids, but looks like it might run well into the megabucks before it's finished.
Should be fun to watch the bidding action on this prize.
|
Wow this auction ended wit the winning bid at almost 6k for a "pre-owned" unit. http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=14021 5764084 They can be bought new for 7500. JT
Logged
« Reply #36 on: March 29, 2008, 06:38:21 AM »
I have RMAA results for Digidesign HD 192 I/O at +4 dBu with a balanced loop-back connection: 44.1 kHz mode96 kHz mode192 kHz modeThis is one of the best measured results I've seen so far.
Logged
« Reply #37 on: March 30, 2008, 08:20:12 AM »
| Alexey Lukin wrote on Sat, 29 March 2008 20:38 |
I have RMAA results for Digidesign HD 192 I/O at +4 dBu with a balanced loop-back connection:
44.1 kHz mode
96 kHz mode
192 kHz mode
This is one of the best measured results I've seen so far.
|
Interesting, I'm quite surprised... Is there any way of measuring jitter with the 192? My guess is this is where it would come unstuck. Matt
Logged
« Reply #38 on: March 30, 2008, 04:08:07 PM »
| Jerry Tubb wrote on Thu, 27 March 2008 12:14 |
| Quote: |
Just thought I'd mention there's a Lavry Gold AD122-96 up for bid on eBay at the moment. I had one of the lower bids, but looks like it might run well into the megabucks before it's finished.
Should be fun to watch the bidding action on this prize.
|
Wow this auction ended wit the winning bid at almost 6k for a "pre-owned" unit.
http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=14021 5764084
They can be bought new for 7500.
JT
|
Doesn't the lifetime warranty expire when you sell your Lavry Gold? bab
Logged
« Reply #39 on: March 31, 2008, 07:45:06 AM »
Thanks for posting those test results - they line up pretty closely with my own results (I went to the host site's list and compared some other measurements I'd made with Mix systems etc, and they too seem to match).
One of the most interesting aspects is how these measurements hew to Dan Lavry's hypothesis. The higher the sample rate, the worse the performance! Secondarily, while these interfaces measure and perform quite well relative to other units tested, the list is pretty short, and doesn't include the kit ME's typically use. No surprise, Mytek and Sonic measure better over here... (and I'd expect the same from Prism and Lavry). No magic.
-d-
Logged
« Reply #40 on: March 31, 2008, 11:10:46 AM »
| Matt_G wrote on Sun, 30 March 2008 13:20 |
| Alexey Lukin wrote on Sat, 29 March 2008 20:38 |
I have RMAA results for Digidesign HD 192 I/O at +4 dBu with a balanced loop-back connection:
44.1 kHz mode
96 kHz mode
192 kHz mode
This is one of the best measured results I've seen so far.
|
Interesting, I'm quite surprised... Is there any way of measuring jitter with the 192? My guess is this is where it would come unstuck.
Matt
|
Another issue is the loading when connecting various mixing consoles and mic preamps to the unit (ie API and some balanced circuits etc). It seems all sorts of things can happen then that are detrimental to the sound quality. The analog front end does not seem to be very robust or reliable.
Logged
 "T(Z)= (n1+n2*Z^-1+n2*Z^-2)/(1+d1*z^-1+d2*z^-2)" - Mr. Dan Lavry "Shaw baa laa raaw, sidle' yaa doot in dee splaa" . Mr Shooby Taylor
« Reply #41 on: April 03, 2008, 11:51:27 PM »
| Dave Davis wrote on Mon, 31 March 2008 12:45 |
()
One of the most interesting aspects is how these measurements hew to Dan Lavry's hypothesis. The higher the sample rate, the worse the performance! ()
-d-
|
Hi David It's actually not hypothesis but a fact- 96k using the same chip measures better than 192k by average of several dBs of dynamic range BUT , there is a big BUT: 192k does sound a bit more transparent subjectively, at least to me and many respected engineers with good ears. My hypothesis is that the benefits of filters being twice as far from audio band at 192k overweight loss from objective dynamic range reduction. In digital, particularly, all distortion are not equal, some of them can be audible far more than a general quantization noise even when their level is much lower, mostly because they are non harmonic ie. are a modulated "out of tune" artifact of the program, completely non musical and non existent in nature. Michal, Mytek New York
Logged
« Reply #42 on: April 04, 2008, 08:01:48 AM »
Hi Michal... Great to see ya here!
Interesting confirmation and post. I know you and Dan have butted heads in the past, and it drives me nuts when I find myself nodding at your (sometimes) diametrically opposing perspectives. As usual, your honesty and clarity come through in this reply. While many (including Dan) may not be comfortable with your uncertainty, I find it refreshing for someone who understands the math to concede the theory, while maintaining a firm grip on practice.
My observations generally align with your statement here. I might argue the difference in d/r is a consequence of the issues of timing and settling times that Dan claim make quantization inherently less accurate at 1/192000sec - in other words timing differences during record AND playback contribute to the measurement error. As you suggest, 192 GENERALLY sounds better subjectively than 96K on many pieces, while usually measuring worse.
I've always been firmly in the euphonic distortion camp: we frequently prefer lossy media, like vinyl and analog tape, over fidelity BECAUSE of the distortion added. It's possible for something to sound different and better... don't we like calling that "mastering?" As long as the matter remains in this context (processing and presentation vs. fidelity to source) it's difficult to argue with your conclusion. Indeed, even for program destined for 44.1, analog processing loops make 192K capability useful and desirable.
This brings me to one final data point. I think you've pointed out before that 192K playback of 44.1 material often sounds better than the native rate. Regardless I've definitely observed this. I think this points to the underlying truth of Dan's position and your hypothesis of subjective improvement: the benefits of 192K, like the problems of jitter, are to some extent ephemeral, an artifact of the conversion process, rather than an attribute of the source sound. The fact that many prefer 192K recordings to upsampled 44.1, with 192K playback of both, doesn't mean 192K is more accurate or contains secret information our ears require to enjoy. It simply suggests a double layer of this uncertain quantization is euphonically more pleasant than a single layer.
"Uncertain quantization" is a loose term I'm coining to describe the defect Dan rails against. When I think about this set of facts, and combine it with the DSD/SACD experience, I have to wonder whether this subjective improvement is a result of the digital parts behaving in a more "analog" fashion... slew rates and settling of converters are at their limits, making for smoother but less accurate transitions between quantized values. Maybe our ears, amps and speakers prefer a smooth, yet still tightly timed signal, with the added bonus of high frequency "defects" of the original signal present to affect reproduction post-converter.
Thanks again Michal...
Logged
« Reply #43 on: April 04, 2008, 04:28:50 PM »
| Dave Davis wrote on Fri, 04 April 2008 13:01 |
I think this points to the underlying truth of Dan's position and your hypothesis of subjective improvement: the benefits of 192K, like the problems of jitter,
"Uncertain quantization" is a loose term I'm coining to describe the defect Dan rails against.
|
Maybe it sounds counterintuitive, but lesser 192k performance has nothing to do with timing errors or more jitter. If you analyze impementation of a typical real world delta sigma based ADC or DAC , master clock for 96k and 192k is exactly the same- the only difference is where the lowpass filters are (well that's again depends on circuit- but there are some that have only this difference). So if we are talking about subjective sounds differences we ought to focus on filter artifacts first and foremost- not clocking or some "uncertain quantization" (BTW what is the definition?). It's not the speed of quantization either- in typical ADC signal is sampled at 128x48k and then filter is applied to downsample to 96k or 192k so sampling is always at the same speed, the only difference is filter band. Assuming that filters at 96k are inaudible (as some would assume ) is wrong IMO- there is plenty of evidence that changing filter type produces change in sound. DSD sounds best some say- maybe because iit has NO digital filters? Regards Michal, Mytek New York
Logged
« Reply #44 on: April 08, 2008, 01:40:07 PM »
Wow, you guys have gone pretty off-topic.... But still worth!  
Logged
|
