I finally came to a good conclusion with my setup of the Tascam unit. About a month ago, I got a LPS (so called low noise) and its been running now stable where I had about 5-6 evenings totally focused on testing, listening and enjoying working with my Tascam unit and I felt I didnt get the good sound quality I expected, recording high res material from foobar2000. I still heard the little muddy soundstage and unprecise recording of crystal clear music.
 
Since I have 0 artifacts in noise floor, and I have calibrated my system to play -0.5 db as max tone when the tone is played at full volume, I would say Ive calibrated it pretty well. Rightmark does great measurements.
 
Then yesterday, I thought, hey, what if I plug in AES input from my Gustard u12 which has a 0.1ppm crystal, and mute the digital in, but allow the Tascam unit to sync to the clock signal ?
 
So tonight I have been recording a bunch of songs, with and without AES clock input from my Gustard U12.
 
And voilá. The soundstage got precise and adjusted to just what I am used to when playing through my whole system. For listening and recording, I use same DAC, and rest of the system is the same. I did several heat maps of the songs recorded and they seem to be missing some data around its noise floor compared to the original track, but also I use a 21 bit closed form DAC which gives me a realistic noise floor of -120db, while the original songs can be deeper than that being 24 bit which most likely explains why I miss some data deep down around the noise floor.
 
Even using the Tascam unit as a DAC, with only the Gustard U12 AES clock just staying there sending no data, and taking the analog output of the Tascam unit to my amplifier gives a great potential for sound improvement.
 
Suddenly I really got impressed by the Tascam unit, and not kinda disappointed which Ive been since I got it, thanks to external clock which is way more precise and sets the soundstage and quality to a new level.
 
Tomorrow I will continue with some jitter tests with and without AES clock.

Heres the FFT spectrum of 11khz tone at -0.5db sent through my system to the Tascam Line inputs and recorded.
 
This 11khz test tone is With AES sync, sent through my DAC to analog line input of Tascam unit, recorded.

 
 
This is only with the internal clock of the Tascam unit, sent through my DAC to analogue line input of Tascam unit, recorded.

 
I can hear a difference between recordings in the details, instrument separation and accuracy with and without AES sync.
 
The recorded tracks with AES sync, is damn hard to hear difference compared to the original HDtracks version. I think this is as far as I can push my Tascam unit. My DAC is a 21 bit closed form DAC (not multibit) so comparing the original track would probably miss out on some deep noise floor data, but it wont be audible as were talking data between -125db and -135db if I had a true multibit DAC.

@manpowre You've got some seriously wacky issues going on somewhere!
 
1. You're showing spurious spikes up to about -68dB which is roughly 100 times greater than I'd expect to see even from a much cheaper unit! Could be any one of a number of different problems, from a gain staging issue to the output from Foobar or your DAC.
 
2. In general ADC/DACs never perform better on external clocks. The best units, with the best clock recovery circuitry, perform the same on external clock as internal clock but none perform better. Your result again points to some serious problem with your test/test signal or far less likely, a fault or serious design flaw with your Tascam.
 
Quote:

manpowre said:

  I use a 21 bit closed form DAC which gives me a realistic noise floor of -120db, while the original songs can be deeper than that being 24 bit which most likely explains why I miss some data deep down around the noise floor.

Click to expand...

 
No commercial recordings I'm aware of have a dynamic range greater than about 60dB and most have significantly less. Furthermore, it's actually impossible to record at 24bit or even anywhere near 24bit. The most dynamic of pro recording mics have a range just over 90dB and the most commonly used mics have around 75dB or less (IE. Roughly 12bit or less). The only data you can be missing in 24bit "deep down around the noise floor" is noise (acoustic noise floor, thermal noise, etc.)!
 
G

This is whats nice with forums, you engage with some content and get feedback, and thanks to your post, I went through my fft analyzer, and I found that the test tone was sending signal at +2 db.. I took the test tone down to -1.02 db, and that made the fft different. I have seen that Innerfidelity operates with test tones at -6.5db for similar tests. Hmm, so now these two fft doesnt show much of a difference.
 
Im here to learn :=)
 
So here is the fft with AES sync to the Tascam unit with the tone set to -1.02db:

 
And here is the internal sync, same tone:

 
 
So back to idea board, why does the recording of the AES sync from high res material sound better ? I dont think its imagination, its very clear that the recording with internal clock is more blured soundstage and less details. I even made a playlist with the recordings and moved things around and picked out the internal clock recordings when they were played.

manpowre said:

  [1] This is whats nice with forums, you engage with some content and get feedback, and thanks to your post, I went through my fft analyzer, and I found that the test tone was sending signal at +2 db.. I took the test tone down to -1.02 db, and that made the fft different.
 
[2] So back to idea board, why does the recording of the AES sync from high res material sound better ? I dont think its imagination, its very clear that the recording with internal clock is more blured soundstage and less details. I even made a playlist with the recordings and moved things around and picked out the internal clock recordings when they were played.

Click to expand...

 
1. Ah, +2dB of clipping would certainly account for your previous readings. Your new readings are exactly as I would expect, with the external clock being marginally inferior (noisier) than internal, although down at those levels the difference would obviously be completely inaudible.
 
2. "Imagination" is the most likely culprit. A thorough double blind test is the only reliable way of eliminating imagination as the cause. I believe there is a free plugin for foobar specifically designed for this. There are some potential possibilities other than just imagination but that's what I'd test for first as it's by far the most likely.
 
G

Hello.
I just like to share my fix to the crazy powersupply in this thing.
There are 2 things you can do to the provided supply if you know your way around some solderingtools. Lift the blue capacitor near to the rectifierdiode on the output side. Solder it in paralell with the electrolytic cap instead. This will remove all the "groups" of noise in the spectrum except the top one. To move the last piece of the problem away, replace the 7 pin switch regulator with the 100KHz version. (You can buy them on ebay). They are called A6259H.
With this fixes you might have the slightest residue of problem in the far end of the spectrum when you sample 192KHz, but else, its all gone. Good luck

comment deleted