Topic: [COMP] New V3/722 A/D comp. w/ clock isolated

[easy jim]

tapers.org/drive1/v3-722comp.dnove2007-04-07
p: ftp4all
u: ftp4all

Thanks to thepassionofMusic for help recording the sources and supplying the V3/722, and thanks as well to terrapinj for putting the comp. files on the tapers.org public FTP space.
------------------------------------------------------

thepassionofMusic and I were recording a Delta Nove show and used his V3 as the pre for my AKG c452eb/ck1's that I had clamped up on a utility pole in the venue. 

The V3 digital out via s/pdif to my MOTU rig was the main plan to be a part of a 6 track multi, and Justin was running analog out to his 722 to back up that source for me and have something to take home.  He had a BNC cable on hand to sync his 722 to the V3 via wordclock.  Thus, this comp. was born.  Because the 722 and the MOTU were both synced to the V3's clock, this is a true comp. between the V3's and the 722's A/D stage.

Both sources were lined up in AudioDesk 2.04 and normalized to the same level (.1 db accuracy) by adding gain.  They were then both resampled/dithered from 24bit/48khz to 16bit/44.1khz in AudioDesk using the "best" quality setting, exported as .aiff files and FLAC'd w/ xACT.

I'll post the answers to which source is which some time next week to give folks a chance to pick which is which without being biased by knowing ahead of time.
----------------------------------

Delta Nove
4/7/07
The Blue Cafe
Long Beach, CA

1. Ringtone/Matches...
2. Wizard (Black Sabbath cover)

DAUD: AKG c452eb/ck1 (DIN, clamped to utility pole ~ 12' high, 20 - 25' from PA) -> Lunatec V3 (set at 20db with the trim knobs around 12 o'clock)

Alternate sources:

-> Sound Devices 722 @ 24bit/48khz (via analog line in @ -5db gain + V3 wordclock out)

-> MOTU Ultralite (digital in/clock synced to V3 via s/pdif) -> PowerBook G4/AudioDesk 2.04 @ 24bit/48khz
 -------------------------------------------------------------

If you want want to check out the 6 track stage+PA+SBD mix, it is here:
www.archive.org/details/dnove2007-04-07.matrix.flac16
www.archive.org/details/dnove2007-04-07.matrix.flac24

[easy jim]

Thanks for the comp!

Because the 722 and the MOTU were both synced to the V3's clock, this is a true comp. between the V3's and the 722's A/D stage.

Question..  If the 722 clock is better than the v3 clock, wouldn't this potentially taint the performance?

I think that is a good question, which I also pondered a bit.  The quality of the master clock in an A/D conversion can make a huge difference in the sound.  A/D conversion is (at least) a 2-part process: (1) analog signal is converted into bits (2) in reference to a master clock.  This comp. just isolates the 'first' part of the processes, conversion of the analog signal to bits.

My gut feeling is that the 722's clock is better, but since the V3 only has wordclock out (but not wordclock in) we did what we could with the resources at hand.  If we could have done the test in reverse config., with the 722 providing the clock signal, that would have been the first choice.  Since we were able to isolate the clock as a part of this comp., it seemed worthwhile to share and see what others think.  Having prepared the comp. and listened to each track a few times, I heard a noticeable difference.

Maybe someone a bit more technically inclined can answer whether or not the (potentially) lesser clock from the V3 would taint the performance in this test. 

[JasonSobel]

a couple of thoughts.
first off, what makes you guys think that the clock in the V3 is of lesser quality than the clock in the 722?  I think both are very high quality clock sources and it'd be very hard to hear the difference between the two clocks.  on the other hand, I'm not sure how one would go about testing the quality of the two clocks to compare them, so it might be very difficult to ever really resolve this potential issue.

but, on the other hand, I'm glad that you synced the clocks.  for people who want to hear a comp between the V3 A/D and the 722 A/D (each device using their own respective clocks), I've already done one, and it's here:
http://taperssection.com/index.php/topic,81297.0.html

anyway, I haven't gotten a chance to listen to this one yet.  I'm downloading the files now, and hopefully tomorrow I'll get to check it out.
is there any reason why you normalized both sources?  running the 722 line-in at -5dB should have produced two recordings with identical levels, so I was just wondering about that.  thanks for posting the comp...

[easy jim]

I think both are very high quality clock sources and it'd be very hard to hear the difference between the two clocks.

Agreed.  We did not really go in to the show planning to do the comp.  It just grew out of Justin and I realizing during the show that we already had things conveniently configured, and we were aware of your other comp. and that this one would have that one extra variable isolated.

is there any reason why you normalized both sources?  running the 722 line-in at -5dB should have produced two recordings with identical levels, so I was just wondering about that.  thanks for posting the comp...

Without adjusting the gain, they were about .5db off (+/- .1db) from each other if I remember correctly, so I decided to go ahead and raise the gain on both so they would be 'normalized' to the same level.  I figured .5db was enough of a difference to be audible and that it may potentially skew the listener's preference.  I also thought about just using the raw 24bit sources too, but decided against due to file size.  I could reproduce the comp. at any time though, should it be requested, without adding gain to the files to normalize and/or in 24bit.

[JasonSobel]

well, I took a quick listen this morning before I left for work (on my headphones, Grado RS-2's.  My wife was still sleeping, so I couldn't put it on the stereo).

I think A = V3 and B = V3 (analog out) > 722
and I prefer A.  I think it sounds smoother and more natural.  of course, the differences were fairly subtle, as I expected.  the bass in B was slightly more pronounced, but it seemed like too much.  I think the bass fit in better with the overall sound in source A, which is why I said it sounded more "natural".  of course, without having been there, it's hard to say for sure which one is more "natural", but anyway, that was my sense.
thanks for posting this comparison.  I think it provides similar information as my comp, and it helps to have more than one data point.

one last thing that I'm curious about though:

Without adjusting the gain, they were about .5db off (+/- .1db) from each other if I remember correctly, so I decided to go ahead and raise the gain on both so they would be 'normalized' to the same level.  I figured .5db was enough of a difference to be audible and that it may potentially skew the listener's preference.  I also thought about just using the raw 24bit sources too, but decided against due to file size.  I could reproduce the comp. at any time though, should it be requested, without adding gain to the files to normalize and/or in 24bit.

I agree that a difference of 0.5 dB might be audible and could affect the results.  when you tested the files, were they cut at exactly the same spot?  if one of the files was a longer than the other, than the average RMS value of will vary.  but if they're both trimmed to exactly the same length, I bet the level differences between the A sample and the B sample would be smaller.

[it-goes-to-eleven]

first off, what makes you guys think that the clock in the V3 is of lesser quality than the clock in the 722?

Better, worse, it doesn't really matter. Just that it is different. Even if it was running off of another 722, we'd have the unknown of whether the cable was introducing jitter.  Fwiw, I believe the timecode model 7xx recorders do have better clocks in them.

I am also wondering whether running the 722 line in at -5 is a compromise.. I really want to do a -5 vs. 0 722 comp.  My recent experiences with the r09 have left me with no doubt that the internal attenuation has issues.  But the r09 can sound really good if run optimally (and even when not).

I've also been thinking about ways to do a level comp with the 7xx. Given a music source that is -14dBFS RMS avg when peaked at -1dBFS, does it sound better when peaked at -8/-21RMS and raised in post to 0dbFS?  In this case I'd assume the 722 is run at 0 and the gain is changed on the pre.

Again, not picking at Jim's comp. Just trying to understand the factors that may contribute to best sound.

[easy jim]

one last thing that I'm curious about though:

Without adjusting the gain, they were about .5db off (+/- .1db) from each other if I remember correctly, so I decided to go ahead and raise the gain on both so they would be 'normalized' to the same level.  I figured .5db was enough of a difference to be audible and that it may potentially skew the listener's preference.  I also thought about just using the raw 24bit sources too, but decided against due to file size.  I could reproduce the comp. at any time though, should it be requested, without adding gain to the files to normalize and/or in 24bit.

I agree that a difference of 0.5 dB might be audible and could affect the results.  when you tested the files, were they cut at exactly the same spot?  if one of the files was a longer than the other, than the average RMS value of will vary.  but if they're both trimmed to exactly the same length, I bet the level differences between the A sample and the B sample would be smaller.

The files were cut at the same exact spot.  I lined both sources up on the time ruler and cut them together before raising the gain on each and exporting.

I do not 'normalize' based on average RMS values (AudioDesk and Digital Performer do not provide this option, though I doubt I'd use it if they did).  Since I am not anti-compression, when used lightly and appropriately, I will use compression (generally never more than 2:1 with the threshold set around the average peak levels) when mastering multi/matrix recordings, and in rare cases 2track SBD sources. 

With 2track AUD sources like those used in the comp, however, I will only evenly (unless L/R adj. is called for) raise gain on both channels up -0.2dBFS - 0dBFS.  In this case, I raised the gain on both sources so the maximum transient peaks for each track had matching dBFS levels (+/- 0.1dBFS)

Are y'all ready for me to post the answers?

[easy jim]

And the answer is:

A = v3
B = 722

[jerryfreak]

is there a link for this?

[dmccabe]

My 2 cents info on A/D comparisons...

I help run the MOTB project.
We currently have the following hardware:

Lunatec V3 (with special mod for LINE IN (not mic in) from M. Grace)
Mytek 192
Apogee Rosetta 200
Bechmark ADC1
MOTU 896

and our listening environment includes:

Presonus Central Station
Benchmark DAC1

Sennheiser HD650 headphones
Mackie HR824 monitors

After EXTENSIVE testing... our ears like the Benchmark and the Lunatec V3 the best.
We also have the Sound Devices 722. It ranked the lowest in quality for A/D... not even close to the Benchmark, Apogee, Mytek or even the MOTU!

I keep seeing these posts raving about the A/D on the SD722. Not to my ears. I wonder what listening chain those who rave about the SD722 are using.

The one thing we are missing in our test is a good word clock... like the Apogee Big Ben. Our group will try to borrow one and do a test with all the A/Ds we have -- clocked to the master Big Ben. We have some rare master analog reel to reel tapes coming soon, and we can post a sample song done on each A/D to post for everyone to compare.

*FYI, the special mod for our V3 (done by Micheal Grace) was for LINE IN only (we use the V3 as the A/D for analog conversion of master tapes), it does not affect the MIC IN. The default setup on the V3 MIC IN is as good as it can be. Anyone who needs some info on the V3 mod can PM me.

[easy jim]

is there a link for this?

It is on the tapers.org FTP site Jaime.  Check the top of the original post for the FTP info.

After EXTENSIVE testing... our ears like the Benchmark and the Lunatec V3 the best.
We also have the Sound Devices 722. It ranked the lowest in quality for A/D... not even close to the Benchmark, Apogee, Mytek or even the MOTU!

As far as I know, the MOTU interfaces use the same convertors as Apogee units and most other top dog A/D devices.  IMO the clock is the main place where the differences come into play. Once I strated feeding a digital input from an Apogee MiniMe to my MOTU interfaces, enabling a lock/sync to the Apogee's clock, there was a very noticeable difference to my ears.  Same converters, different clock.

A MOTU + an Apogee/Mytek/Lucid master clock seems to me to be an excellent (and cost effective) route for multiple channels of high quality A/D w/ paying the big bucks for Apogee/Mytek/RME/Lucid/Benchmark/Prism multi-channel converters.  My $0.02

If there has been a way to do this comp w/ noth sync'd to the SD 722 clock as well, it would have been a truly comprehensive (and interesting) comp IMO.