[dwk]

Looks like the behavior does match the emphasis curve - the time constants I grepped for on the web indicate 50us and 15us, which means a low-pass around 3.2k and a high-pass to level it off around 10k. This would indeed put the -6dB spot around 6.4k as was reported earlier, and would level the curve off at a max cut of ~10-12dB.

So, one question I'd ask is why on earth would it ever be reasonable to apply de-emphasis to the digital out, since that's supposed to be carried in the subcode (isn't it?)? Then the question would be why it would be applied to discs that dont' have emphasis to begin with (since very few do).

 

[jefemeister]

Just tried a recording of The Wall which is one of the few albums known to have emphasis. The Toshiba data does not match again. I offer no pretty pictures this time, but I verified that the differences are not from a time offset or constant gain. I'll try with the Pioneer in the near future.

I agree that the roll-off pretty much matches the pre-emphasis curve perfectly. It is absolutely not the job of the transport to apply de-emphasis. The status is noted by a bit in the AES stream, and the Digital Receiver chip either auto applies the filter or puts up a flag so you can handle it later. It's possible that the digital output is actually a repackaged version of what's going to the DACs. I can see how this might be cheaper than taking the digital out before it's received. It would also make sense if the digital output is locked to 44.1 as sample rate conversion would have to occur. It's not possible to rate convert data that isn't unpacked first.

But that really doesn't matter. De-emphasis should only be applied to those discs recorded with Pre-emphasis as noted by a bit embedded in the audio stream.

 

[mekanoplastik]

could you try measuring the data from a standard computer cdrom reader? (from the digital out conector at the back),

i ask this because i am using an old scsi cdrom as transport (nec 602) and i wonder how good this thing really is..

you guys have done a very interesting research...thanks

 

[jefemeister]

I've been thinking a little more about the pre-emphasis theory. Of course the recorded Toshiba data doesn't match the EAC rip for this particular test. De-emphasis filters are not completely standard, so it's probable that it is implemented slightly differently in the Toshiba vs EAC. I guess I will have to do a FFT comparison and see if the highs are still being cut. The data is at home, and Matlab is at work, therefore it will have to wait until tomorrow.

mekanoplastik, my CD-ROM reader doesn't have a SPID/F connection, but I can still probably get real-time recordings from the IDE (ie, recordings of direct playback, not rips). This will be basically equivalent.

 

[jefemeister]

Just came across a nice de-emphasis filter plot in a DAC datasheet. The comparison between this and what I measured is dead-on:

 

[j-curve]

So your player seems to be stuck in de-emphasis mode, but why? I looked at a couple of data sheets for DVD decoder chips but found nothing resembling a "de-emphasis input pin" which might offer the possibility of a dodgy soldering joint. Rather, de-emphasis tends to be activated by a software flag.

There's an interesting thread on Steve Hoffman titled "Toshiba Abbey Road CD on Ebay... " which talks about setting the pre-emphasis bit when burning a CD-R, and some relevant info about CD subcode in Chapter 9 of this summary from Ken Pohlmann's digital audio book.

As to the mystery of why a filter should ever precede the digital output, it seems that whilst the AES/EBU standard provides for transmission of the pre-emphasis bit, this is not properly implemented in S/PDIF(!) (See here). The implication being that any transport with an S/PDIF output takes on the additional job of de-emphasis filtering as necessary. Most ungainly - a veritable can of worms.

OK, how about this theory for a long shot:-
1) Toshiba has pre-emphasis on the brain (after all, they did produce the "legendary" Abbey Road CD, didn't they?)
2) The Toshiba DVD player correctly de-emphasizes pre-emphasized discs, including the S/PDIF output.
3) Pioneer ignores pre-emphasis, or at least doesn't filter its digital output.
4) Your CD-R burning software is setting the pre-emphasis bit for some inexplicable reason.

Point 4) seems highly unlikely, so I don't really like this theory, but it's the best I've got for now.

 

[j-curve]

Quote:

jefemeister: It is absolutely not the job of the transport to apply de-emphasis.

Sorry, I should have included this quote in the previous post, since it appears I've directly contradicted you. When I read this thread the first time I didn't think twice about it, but having seen that S/PDIF page I linked I'm not so sure:- Quote:

But the protocol used in AES/EBU and S/PDIF is not exactly the same and that can cause sometimes problems. The basic data format of AES and S/P-DIF are identical. There is a bit in the channel status frame that tells which is which. Depending upon the setting of that bit, some bits have different meanings. For example, the bits used to describe de-emphasis in the AES/EBU protocol overlap the bits used to implement the SCMS protocol in S/P-DIF land.

Also, I think I need to retract that long-shot theory already. Your initial tests compared a non-pre-emphasis original CD (EAC vs. Toshiba digital output), right? If so, there must be some other reason. I was assuming your testing was based on CD-Rs alone.

 

[blumenco]

I have been reading the threads on this. a good question about your tests is do you use a newer toshiba 3960? Does it have a zoran chip or a toshiba chip (the big chip on the output board). apparently, toshiba has changed the setup, but I do not know when. judging from the dates of your tests, it seems that it is taking place after the switchover. whether or not your unit was manufactured earlier or later is to be found out.

course, they might have done it to cure QC issues. I personally went through 3 players in order to find one that was not defective. additionally, is the players you are using perhaps defective?

I have not researched this too much, but there might even be a

VERY informative tests. While the toshiba does have flaws apparently, I would let it be known that a vinnie modded one RSaavedera's was really really good. almost, and really just a smidgen away from a DAW with a lynx two sound card IMO.

Thanks

Clark

 

[jefemeister]

I have a large Zoran chip, so I must have the older version. As to if specifically mine is defective, I don't know. Do you have a good link to something explaining the Zoran/Toshiba deal? What was the problem that caused the switch over?

 

[blumenco]

well, supposedly, the older toshiba 1523 (I think) chips were better. Your test, as I said is very very interesting. A funny thing about dacs that I have learned over time is that they get it either 100 percent right or really ****ed up as in some of the cheap players.

I forget the specific link to the discussion but apparently, the exact same zoran chip you have in your player is also in the cyberhome dvd player in my little sister's room upstairs...

When I first bought my toshiba, I thought it was a revelation, until it broke. that happened two more times till I finally landed on a player that works. Some of the modds I have seen include placing a heatsink over the dac chip. I never tested the messed up players electronically, but by touch, (I know this is dumb to do on a live circuit) the messed up ones were a touch hotter. maybe the factory went to a zoran chip because of dac heat and hence reliability issues? who knows?

I am still following this product even though the new ones are theoretically obsolete simply because the new sony cd player craze is not as well explored as this one. I am just now getting around to buying some vitamin Q capacitors in order to swenson mod my tosh. I have some doubts about whether or not it will really be an improvement, but hopefully I can end up with some bang for the buck.

AWESOME test, and i guess i am really glad more to see us dorks investigating the engineering and design behind the producs that are far too often bought on blind faith.

Clark