Pre-/ De-emphasis via S/PDIF

[Limopard]

I started the topic in the DX 3 pro+ thread. Meanwhile I made some observations.

The DX 3 pro+ decoded the emphasis when fed by a Pioneer BDP 450 (coax) but not when connected with a Marantz CD 5001 (coax or optical) or a Sony D-NE 700 (optical).

A Pioneer NC 50 DAB decoded the emphasis when fed digitally by all three sources.

The reason for this behaviour can neither be the source alone nor the dac alone. I'm quite puzzled - can anyone enlighten me?

Test CDs were Genesis s/t (1983, Vertigo), Pet Shop Boys "Please" (1986 EMI). The added brightness without applied de-emphasis is obvious.

 

[mansr]

Could it be "consumer" vs "professional" channel status format?

 

[little-endian]

Reviving an interesting thread about an unfortunately often neglected topic ... (Michael Jackson's Thriller (35・8P-11) is calling as well):

mansr's contribution is a good one as besides, I don't see any other explanation. While there are two places where the pre-emphasis on a CD may be stored (TOC vs. subchannels), the effective flagging via S/PDIF will be the same, but apparently differ between the consumer and the professional format. Page 9 here shows an overview about both.

So the "1 1 0" for the 50/15 µS Emphasis under the professional format might be misinterpreted (and probably ignored) by a consumer device (falling into the "reserved" category and expecting it to be sent as "1 0 0" instead).

On the other hand, that raises the question for me, why both the Marantz CD 5001 and the Sony D-NE 700 should use the professional format at all. To know for sure, one would have to be able to check the raw stream coming via S/PDIF though.

@Limopard: did you have the chance to follow that issue up further meanwhile? At least, the DX Pro3+ seems to support the de-emphasis in general, which is also supported by the information I find on the ESS website about their chips. Since the EX5 also uses the
ES9038Q2M, it should also support it (page 2 is referring to a block diagram showing the de-emphasis feature). On the newer ones however, such as the ES9068A, used by the DX5 (Lite), they don't mention the pre-emphasis support anymore.

DACs - especially the ones with S/PDIF inputs - should not only support the pre-emphasis in general to comply with CD- or DAT-standards, but also provide an option to manually trigger it for those cases where the original information is "lost in translation" so to say. Also for reasons beyond me, no player on a PC whatsoever seems to support pre-emphasised sources, not even when provided with a cue-sheet where it is clearly stated.

So, DAC users, grab your CDs from the early 80s, honor some very fine pre-emphasised masterings and test whether it sounds too bright or not.

 

[Limopard]

The very few CDs with pre-emphasis are fed from the Pioneer BDP-450 into the DX3 Pro+. So they are properly decoded. Aside from that the tracks are stored as de-emphasized FLAC files on the NAS server.

 

[little-endian]

Thanks for the confirmation. Meanwhile, I've ordered an EX5 to test it as it uses the same DAC (the only reason not to get the actually more attractive DX3 Pro+ was it's 3.5mm output jack) and also contacted ESS about the de-emphasis implementations in their DACs.

So according to ESS, in fact their newer models such as the ES9068AS don't feature a de-emphasis as a "default filter", but they provide a tool for custom filters instead. While such a de-emphasis used on early CD- and DAT-recordings seems to be rather simple, the question is if the manufacturers still consider this and take to (little) effort to create a custom filter.

DACs - especially the ones with S/PDIF inputs - should not only support the pre-emphasis in general to comply with CD- or DAT-standards, but also provide an option to manually trigger it for those cases where the original information is "lost in translation" so to say.

Quoting myself here, I see with pleasure that the RME ADI DACs offer exactly that feature according to one of the manuals, here on page 55. Unfortunately, the price tag is severely higher for at the end the same audible performance to be expected to a Topping of course.

 

[little-endian]

Since the Sennheiser HD-6XX come with a native 3.5mm plug anyway, I gave the DX3 Pro+ a try meanwhile.

Despite the decent performance as to be expected, with my common players at least, it doesn't apply de-emphasis. Neither via Toslink nor Cinch; tried a Panasonic DVD-S47 and a Technics SL-PS740A and the CDs Earth, Wind & Fire - Raise! (CBS/SONY 35DP15) as well as Michael Jackson - Off The Wall (EPIC/SONY 35•8P-2). Both, a Denon AVR-4306 as well as a Benchmark DAC1 do it properly with the same sources.

Which right now really pisses me off. They're all jerking off over SINAD > 120dB, channel separation, THD and whatnot and to think one little step ahead and a taking early CDs into account then is too much. So fucking typical, uargh.

I guess I return it out principle, damn it!

The DX 3 pro+ decoded the emphasis when fed by a Pioneer BDP 450 (coax) ...

Another thought which just came to mind: maybe that particular player performs the de-emphasis already by itself and outputs the already de-emphasised PCM-result which also would explain your observations.

@amirm: Can you please from now on - with sugar on top - test all DACs for proper implemented de-emphasis? After all it is part of the CD standard, so those ignorants are supposed to implementent it, especially when the used DAC chips even support this filter out of the box. All those tiny differences in nowadays performance are entirely irrelevant compared to auch a feature which then turns out to be missing, grrr.

 

[Sokel]

Also for reasons beyond me, no player on a PC whatsoever seems to support pre-emphasised sources, not even when provided with a cue-sheet where it is clearly stated.

There is at least a plug-in for foobar the little I searched:

foobar2000:De-emphasis postprocessor (foo_deemph) - Hydrogenaudio Knowledgebase

wiki.hydrogenaud.io

De-emphasis components

De-emphasis components

hydrogenaud.io

 

[little-endian]

Thanks, yes. My comment was referring to the fact that no player on a PC apparently is able to output the raw material while at the same time trigger the appropriate flag on the PC's S/PDIF interface.

While it is of course perfectly possible to post-process the pre-emphasis within the source, it is still totally ignorant by manufacturers like Topping to not take that into account on the receiver's end.

Also, it annoys me that I now cannot just hook up a CD-player and use it with the DX3 Pro+ but would have to rip and mangle around the CDs first. Which successfully takes any haptic collecting pleasure entirely away as then I can just copy and throw away the CDs anyway.

 

[Sokel]

The list is long and includes some long thread discussed albums too,I think we should take it under consideration:

(one of the lists):

Pre-emphasis (release list) - cdHistory

www.studio-nibble.com

 

[little-endian]

Thanks again. The Lionel Richie - Can't slow down (MCD 06059MD) I also have (great album and mastering). One of the oddly rare examples to list the pre-emphasis in the TOC properly which tools such as EAC then also show on a PC.

So I guess we should make a list of DACs or AVRs which on one hand correctly pay some due respect to the corresponding S/PDIF flag and even better allow to manually enable the de-emphasis for those case where the flag gets "lost in transit".

The Benchmark DACs at least fall in the first category and the RME ones even allow to set it manually (while not exactly cheap either, this company appears to put the most thoughts and effort into their product according to their elaborate manuals).

 

[evalrat]

As far as I know Marantz players support de-emphasis. I would consider it logical that the player applies the de-emphasis and removes the flag from it output streams. Same for the Sony I suppose.

 

[little-endian]

Makes sense to me indeed and it might resolve the "Limopard mystery". At the end probably only academically, the solution of the DAC performing the de-emphasis should be the better one as the de-emphasized result of a 16 bit pre-emphasized source could be put into a 24 bit container or otherwise would have to be dithered again in order not to have any cascading effect (same as when decoding lossy floating point stuff).

Originally, the whole point of the pre-emphasis allegedly was to mitigate SNR/dynamic flaws of early ADC/DAC designs so it would only be consequent to perform this calculation on a "higher domain" when done mathematically instead electronically, but yeah ... most of all that often has little to no relevance, especially today and I wonder if a ABX test would be succesful between a pre-emphasized album and one without.

Anyway, early CDs have it and also today's DACs should the respect that.

 

[little-endian]

Another addition which might be interesting for one or the other:

With my Panasonic S-47 DVD-player I discovered that when using the "AV Enhancer" function (whatever that one essentially does besides upsampling the 44.1 kHz from the CD to 176.4 kHz), pre-emphasised CDs get de-emphasized on the digital output via S/PDIF by the player whereas without that option, it natively delivers the unaltered bitstream from the CD, flagging the pre-emphasis accordingly and leaves the de-emphasis processing to the "external DAC of trust", so to say.

So for any with "ignorant DACs", that player or a similar one might be worth getting.

 

[NTTY]

This is an interesting topic.

I recently started testing compliance with de-emphasis curve for the CD players that I review (using a sweep from 20Hz to 20kHz without and with the pre-emphasis flag).

Recently, I made an operator error, not realizing I was measuring the SPDIF output instead of the analog one, when performing that measurement. And I was surprised to see that de-emphasis was applied in digital domain. The sweep showed the expected attenuation with de-emphasis flag "ON", meaning digital stream was modified prior to being sent, or that my input interface (a Motu UltraLite Mk5) is applying the correction in digital domain directly at the input (but I doubt so, since it is not a documented feature).

I grabbed another player to quickly check, and no difference between the two files at the input of my interface, probably meaning this second player outputs the original bitstream and the appropriate bit state code (bit 3, 4 and 5 of the sub-frame, per the IEC 60958-3), or don’t output it but that’d be weird.

I will need to perform more tests, feeding that second player into a DAC that I’m sure applies the de-emphasis when instructions are present in the SPDIF data stream. Best option would be to use an RME, as it clearly shows on screen the audio input with pre-emphasis, but I don’t have an RME… So, I’ll be using a Topping D50III of which ES9039Q2 certainly applies de-emphasis when the appropriate codes bit state is present in SPDIF input data, as per the datasheet. That should be good enough.

Else there might be a way to read the subframes with specific software (I’d honestly prefer). If you know, I’m interested!

 

[MC_RME]

You mean ES9039Q2M? It does not support De-Emphasis at its analog outputs anymore. The fixed De-Emphasis filter was removed. The former ES9038Q2M still had it. Meanwhile several DACs with ES9039Q2M had been tested here at ASR, but never about De-Emphasis working or not.

 

[NTTY]

Thank you for your insights! When reading the datasheet I saw the below, which conforms to the SPDIF standard, and I assumed it would process it correctly.

 

[MC_RME]

To understand this first the confusing 'payload' has to be exchanged with 'Channel Status' (payload would describe the exact opposite - the audio data within the SPDIF stream).

IMHO this table is a general info on what the Channel Status of the incoming SPDIF signal means, as help to decode it. So you could let a LED light up on the front when the Emphasis bit is set , but it doesn't mean that anything else happens.

 

[AnalogSteph]

You mean ES9039Q2M? It does not support De-Emphasis at its analog outputs anymore. The fixed De-Emphasis filter was removed. The former ES9038Q2M still had it.

Now that's mildly annoying. So the 9039 can avoid overs via digital attenuation but drops DE. Granted, your average XMOS chip these days should have enough oomph to implement the requisite filtering, but it introduces a potential source for bugs. Not necessarily a good thing when naming the various digital filters correctly seems to be a challenge already.