[Jahn]

Verrry interesting. Common wisdom is that the Swenson mods improve on the stock analog outs, and since the stock are producing numbers like this as you posted previously that can only bode well for my Tushi. OK I'm back on track on what my next upgrade path can be, thanks for putting my fears to rest!

 

[j-curve]

Building on the above, here's the simplest IIR filter function I could find:-

ET(n) =
+ 26403/65536 * ET(n-2)
+ 30334/65536 * P(n+2)
+ 12794/65536 * P(n+1)
- 4093/65536 * P(n)

where ET(n) is the Estimated Toshiba ouptut sample at time=n and P(n) is the Pioneer (or "pure"

) input sample at time=n. This model has a latency of two samples (ie: looks ahead at the input) and also takes feedback from its own output two samples prior.

Edit: Letting the recursion act on ET(n-1) instead of ET(n-2), the P(n) coefficient drops out altogether and an even simpler approximation is obtained:-

ET(n) =
+ 41596/65536 * ET(n-1)
+ 30331/65536 * P(n+2)
- 6451/65536 * P(n+1)

The "minus one" issue appears to be related to the arithmetic precision of the filter calculation and truncation to a final integer value. There are various ways to approximate it, such as:-

Output(n) = ET(n) - 1
Output(n) = int( ET(n) - 0.1 )
Output(n) = int( 16*ET(n) )/16 to simulate 20-bit arithmetic
etc. etc.

This has to be done as a separate operation to the calculation of ET(n) though, to prevent the rounding errors feeding into the recursion.

I figured it's all being done in integer arithmetic so I converted the coefficients to 16-bit fractions. In fact, the approximation improved when all the output samples were truncated to integers with no rounding. Not perfect, but very close (within +/-1 everywhere I looked).

This filter takes about 6 samples to stabilize but with a bit of reverse engineering I found that by sticking a dummy sample of 235 at the front of the Pioneer data (at time=0) and allowing the estimator function to start at time=-2, the filter accurately predicted the Toshiba's output from the first data point (at time=1). I've no idea how to derive the boundary condition but at least I now understand how the Toshiba comes up with 181 for the first sample when the disc data says 27.

jefemeister has already beaten me to the white noise analysis but for the sake of confirmation:-

The "Difference" graph shown here is the inversion of the filter response, which is -3dB at at 3.5kHz, -6dB at 7kHz and -9dB from 16kHz up. That's quite a roll-off in the treble (should be easily audible).

jefemeister, my phase graph came up quite differently to yours. I'm guessing you measured T(n) against P(n+2), which is quite reasonable given the latency of the filter. [Also gives the cleanest graph, since P(n+2) has the heaviest weighting. My phase graph is a bit dirty, so I'll show mercy and refrain from posting it for the moment.

]

If the analog stage of the Toshiba counteracts this digital filter, one wonders what benefit is gained...

 

[Jahn]

So this is saying that the digital out will be filtered, but the analog out won't? Eh, forget the DAC idea then.

 

[Wodgy]

Quote:

Originally Posted by Jahn Common wisdom is that the Swenson mods improve on the stock analog outs, and since the stock are producing numbers like this as you posted previously that can only bode well for my Tushi. OK I'm back on track on what my next upgrade path can be, thanks for putting my fears to rest!

Not to disturb you unecessarily, but if the analog filter counteracts the digital filter that has been demonstrated here, the Swenson mods would seriously change the frequency response of the Toshiba away from neutral. However, it's still unclear what's really going on, as Jeff and Mr_Radar's units could also just possibly be different.

It's also possible that there is something wrong with RMAA or the configuration of one of the experimenter's rigs, though exactly what is hard to say.

That said, this player is probably not a good choice for outputting a digital signal to a DAC. Mr_Radar's unit failed the DTS playback test yesterday, which suggests that something is wonky with the digital signal of his unit, though perhaps it is not as bad as Jeff's unit.

 

[Mr.Radar]

Quote:

Originally Posted by Wodgy That said, this player is probably not a good choice for outputting a digital signal to a DAC. Mr_Radar's unit failed the DTS playback test yesterday, which suggests that something is wonky with the digital signal of his unit, though perhaps it is not as bad as Jeff's unit.

That test was preliminary and I used the worst brand of CD-R I owned (they're so bad that I've had one fail with several read errors just a few hours after I burned it). I'll re-run it in a few minutes with a better disc.

EDIT: Okay, the second test with a different brand of CD-R's had the same results.

 

[Mr.Radar]

EDIT: TEST RESULTS WITHDRAWN. I performed a test on the Toshiba's digital outputs but in yet another twist, I tried doing the same test with a loopback through my EMU's digital inputs and outputs and surprisingly I get exactly the same results (which indicate that there are problems with the signal), so it seems that my testing methodology needs re-evalutation. I still believe my DTS-WAV and RMAA tests are correct though. I'll do some more testing later, right now I have to go do some other stuff.

EDIT: No wonder the results didn't match! I had my recording application set to 24-bit output rather than 16-bit as it should've. My computer needs to reboot before I can retest (PatchmixDSP crashed and won't start again).

 

[Wodgy]

Thanks for trying this. The plot is really thickening now.

RMAA is an odd, unpredictable beast.

 

[jefemeister]

I've got the impulse response of the (my) Toshiba. Input is a long string of zeros with a single 32767 value stuck in the middle. This is the largest possible positive value. When played through the Toshiba, I get the following. I don't get the proper 0s out from the toshiba, they are mostly all modified to -1. This is to be expected though. Haven't done this test for the Pioneer because I finally have it back in the rack playing.

Wodgy seems to be good at figuring out equations from strings of values, so have at it!

Quote:

-1 -1 15166 6398 4061 2578 1636 1038 659 418 265 168 107 67 43 27 17 11 7 4 2 1 1 0 0

 

[Mr.Radar]

In an update to my previous test I've finally figured out how to get a bit-perfect recording when routing audio inside patchmix, and I've discovered you need to use 24-bit audio for SPDIF connections to measure it the way I am (which means my test won't work for the Toshiba, unless I make a DVD with a 24/96 audio track, which is way too much work for me right now). I recorded my 1234567890 pattern (encoded as a 24-bit PCM file) using the loopback of my EMU card's digital I/Os and internally (no physical connections) and every third number is replaced with a blank byte (hex code of 0x00) so the patter goes "12_45_78_01_34_67_90" and so-on and so-on.

EDIT: Okay, it turns out that Audacity screwed with my test pattern, for both the 16-bit and 24-bit tests (I'll report this bug to the Audacity team). It seems that both sources passed the 16-bit test perfectly, while the EMU at least fails the 24-bit test.

 

[Wodgy]

Mr_Radar, would it be possible for you to try to replicate Jefemeister's experimental configuration, so you can try to replicate his results? It seems that his set up is producing solid, repeatable results, and you both have the same gear (Toshiba and 0404) so it would be interesting to know if you would get similar results if you repeated his kind of experiment. Perhaps you two could agree on a single common audio file, burn it to disc, and record it, and see if the recordings match 1) each other and 2) the original file.

 

[Mr.Radar]

Quote:

Originally Posted by Wodgy Mr_Radar, would it be possible for you to try to replicate Jefemeister's experimental configuration, so you can try to replicate his results? It seems that his set up is producing solid, repeatable results, and you both have the same gear (Toshiba and 0404) so it would be interesting to know if you would get similar results if you repeated his kind of experiment. Perhaps you two could agree on a single common audio file, burn it to disc, and record it, and see if the recordings match 1) each other and 2) the original file.

Good idea. jefemeister, if that response file is less than 5 megs you can e-mail it to me at tknott@gmail.com

EDIT: Sorry to everyone if my posts today seem confusing or vague. I'm really tired and I've had a bad day today. To recap what I did today (relating to this thread): I re-ran my RMAA testing and got more accurate results than I did before, as well as measuring the digital outputs of the Toshiba as bit-perfect (according to RMAA anyways). I then tried an alternate method to measure the Toshiba's digital output, but due to me improperly configuring Audacity my test data was screwed up (though the Toshiba seemed to perfectly reproduce the bad test data perfectly anyways). Based on my testing it seems my Toshiba is bit-perfect. I'll try to reproduce jefemeister's results though, if he sends me his test data.

 

[Glassman]

first of all, try using WaveLab or some other application you can trust, and preferably using ASIO to get around any problems caused by WDM.. mute everything and just place SEND to S/PDIF to your ASIO or WDM out strip and ASIO or WDM in SEND in your S/PDIF in strip, connect digital output to digital input and play & record something, then compare the results, they must match.. now connect Toshiba's digital output, switch Toshiba on, chanage clock source to external in Patchmix and record.. always be shure you're getting good results from digital loopback before you measure external device..

 

[gaboo]

Quote:

Originally Posted by j-curve I've no idea how to derive the boundary condition but at least I now understand how the Toshiba comes up with 181 for the first sample when the disc data says 27.

From what I know CD-A does not define the exact start of the disc. So different drives/players start a different locations. This is the reason for read & write offset options in EAC. Did you somehow measure and compensate for the different read offsets?

Looking at jefemeister's impulse response, the toshiba should mangle initial 0's into -1's. Is this happening to your player, or is it something else? Confusion reigns supreme.

 

[jefemeister]

Quote:

Originally Posted by Mr.Radar Good idea. jefemeister, if that response file is less than 5 megs you can e-mail it to me at tknott@gmail.com

I'm posting wav files of both my whitenoise input as well as the resulting Toshiba 3960 output:

Whitenoise Test Files

above link is a zip file.

 

[Mr.Radar]

I just tried the whitenoise sample that jefemeister provided, and something is up. Just visually inspecting the files (zoomed down to the individual samples) you can tell that the recorded data isn't the same as the reference file. Check out these screen shots (the Toshiba sample is on top of each shot and the reference is on the bottom):

In the top screenshot the Toshiba data was recorded at 16-bit and in the bottom shot it was recorded at 24-bit (in case the Toshiba upsampled to 24-bit and the differences in the samples were due to the driver converting the 24-bit data to 16-bit). Also, for some reason the reference file was always longer than the recording (I set the EMU to use the Toshiba's clock so it couldn't be that samples were dropped because the clocks didn't match).