ok, so what I want to do is run V3 (S/PDIF) > HD-P2 and at the same time go V3 (analog) > 722 line-in.
I want the levels to be spot on, so I don't have to change them in post and so that the comp will be perfect.

The 722, when going line-in, has a gain range of -6 dB to 18 dB
The 722 manual says that the line-in will clip at +26dBu (with the gain fully down)

When the V3 A/D meters are at 0 dBFS, the balanced analog output is at +25dBu

so, all I need to do is set the 722 line-in to -5dB, correct?  that way, the 722 will clip with a +25dBu signal, which is exactly what the V3 will be putting out over the analog output when the V3 A/D clips.  if I do this, in theory, the V3 levels will be identical to the 722 levels, correct?

I've thought this through and I'm pretty sure about this, but it's always a good idea to check on these sorts of things 

The problem I and others have had, is that when you are running a V3 analog, you only want the meters to flash the first amber.  This results in a very low digital signal via the SPDIF to the other deck.  While your math may be correct, it presumes running the V3 hot, which you do not want to do when using it for analog out.

I realize that when the V3 is run hot, it causes a problem for the analog outputs, because most recorders cannot handle that hot of a signal.  the "first amber", though, is really just a rule of thumb, and is very dependant on what you are running analog into.  As I read the specs, the 722 can handle the analog signal, even when the V3 is run hot.  I'm really just trying to confirm that I'm reading the specs correctly.

Jason --

From the numbers, it sounds like you've got it right.  I just got my new (used) 722 today  -- so I'll start playing around with it this weekend.  I'd like to be able to do just what you are saying and get the V3(digi)>MT and V3(ana)>722 dialed in so I can do some A/D comparisons without needing to normalize levels in post.

Let us know what you find out!!

I finally got around to a quick test of this, because tomorrow night is my big comp: V3 A/D vs 722 A/D.
so, I want the levels to be right on.  I just did a minute long test in front of my home stereo.

V3 > S/PDIF > HD-P2
and
V3 > analog > line-in 722 (-5 dB)

both recorders at 24/96.

after transferring the files to my computer, I went into WaveLab and trimmed the start and end of the two files so that they start and end at the same point.  I then did an analysis to see how the levels turned out.

here are the results:
HD-P2 levels:


and

722 levels:


so, if we look at the average RMS values, the left channel is within 0.04 dB between the two recordings, and the right channel is within 0.06 dB between the two recordings.

I am thinking that this is very good level matching between the two recorders, and that my full-on comparison tomorrow night will be a wonderful comp between V3 A/D section and the 722 A/D section.

but I was still curious about the tiny discrepencies.  I think this is the reason:  because I did not use wordclock to sync the two files, even though I trimmed the two files to start and end at the same place, the files from the 722 was actually slighty longer, as evidenced by this shntool output:



from that, we can see that the file from the 722 is 0.003 seconds longer than the file from the HD-P2, which is roughly 288 samples at 96 kHz.  so, it is entirely possible that the very slight level variations are from the different number of samples being analyzed between the two files.  unfortunately, I do not have a BNC > BNC connector, otherwise I would run word clock out from the V3 to the wordclock input on the 722 (my V3 is not an opti-mod). 

anyone have any other suggestions as to why the levels are slightly off.  (of course, with levels that close, I don't think people will really be able to detect that small difference, and that won't affect what people think of the comp.)

BUT, I think my comparision will still be valid, even with the different digital clocks.  it looks like the levels will be spot on, and I have no intention of ever mixing the two sources, so the slight time drift will be a non-issue for me.

so, this brief test of mine indicates that the published output levels on the V3 and the published input levels on the 722 are spot on.

and as a side bonus, this was the first time I ever used the analog outs on the V3.  by running from both the analog and digital outputs, I was able to verify that Grace did indeed fix the polarity of the digital output relative to the analog outs.  not that I ever doubted it after getting the unit back from them, but it's nice to see right in front of me that the two files were of the same polarity

Ok, I just pulled up my Bustle files and used CoolEdit to analyze them.

Average RMS power, V3:
L:  -19.29    R:  -19.51

Average RMS power, 722:
L:  -19.07    R:  -19.27


So 0.22db variance on the L channel, and 0.24db on the R channel.

Not quite as accurate as you found, but far close enough for my taping needs.  I didn't put a huge amount of effort into trimming the files exactly though, so that might explain the difference.  I was fairly close with the trim points, but the files still varied by about 200ms total in length in the end.

Very good read, but keep in mind that by attenuating the 722 source by -5db you are changing the quality of the signal going to the 722 AD stage. 

After reading the responses and notes from the AD comp I think it is safe to say that the V2/3 pre amp is what we all knew it was and the 722 AD is slightly superior then the V3's. 

Wonder what specifics internally make the 722 AD sound slightly better then the Lunatec?

There is defintaly something weird about that Grace AD.