CS8412 to EIAJ or I2S

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
It's been ages since I used a CS8412, but I've forgotten my basic DIYs and am lazy anyway ... it's easier to pose a dumb query here as I'm sure the answer is easy for many of you...

I've got two DAC types:

Philips TDA1545A (accepts only EIAJ data)
Philips TDA1543, TDA1541A (accept only I2S data).

EIAJ and I2S are, of course, incompatible signal formats. So for a Philips CDP I have that came with stock TDA1543, I had to create a special I2S-to-EIAJ converter (using a bunch of 74HCs, etc) to get the CDP's decoder to "talk" to the EIAJ-based TDA1545 DAC.

However (and TTBOMK), all schematics that I've seen using CS8412 + TDA1545 or CS8412 + TDA1541/TDA1543 have direct connections for the WS (FSYNC), CLK, DATA lines.

How does one config. the CS8412 to spit out I2S or EIAJ signals ... or am I missing something obvious?

Thx!!
 
Last edited:
Stay ON-TOPIC, please!

Use TDA1387T (SOT96-1) which is a smd variant of TDA1545A that can accept I2S signal directly...no glue logic needed.
Right ... but that's not what I asked...

Rather ...
If I want to use a EIAJ DAC (TDA1545A), why, do I have to to CONVERT to EIAJ in a CD player that orig. has an I2S-based DAC (TDA1543, TDA1541, etc.)? I.e., the CD player's decoder only spits out I2S.

But ... (and by all means correct me if I'm wrong)...
Why don't I have to convert to EIAJ (or I2S) for any-format DAC (TDA1454, TDA1543/41) if I am using a Digital Audio Interface Receiver IC, like CS8412, that also spits out the same three-signal lines (WS, CLK, DATA).
Or is there some set pin or other simple config. on the Receiver chip that allows one to select its output format .... or do Receivers like CS8412 spit out some kind of "universal" one-size-fits-all datastream that makes my query moot?
 
The CS8412 has various output modes and yes it can be configured to drive both EIAJ and I2S without needing any glue logic. Table 3, page 24 is the relevant one. You'll note that mode 2 is I2S compatible and I seem to recall mode 5 works with TDA1545A.

Thx ... I'll ck that out ...

One more somewhat-related query:
Do any of these Receiver ICs, like CD841x, also have WS/CLK/DATA as inputs (i.e., in addition to the usual SPDIF-data input)? ... this would allow them to be used as a one-chip I2S <--> EIAJ converter.
 
The almost-"single"-chip design proposed on this forum some time ago does not work, as I noted in that same thread.

I was able to get the following to work:
universal_shifter.gif

...but it's hardly compact or elegant, on my Veroboard version:
An externally hosted image should be here but it was not working when we last tested it.


There is a marketed version that came out not too long ago from diyparadise ...it may use fewer ICs (can't see other side of PCB, so not sure about components used)...
black_crowe.jpg
 
If minimum chip count is your aim then Tam Lin's schematic (left in the comments on my blog) is easily the winner - it would probably need just one extra half-chip to delay WS by a single clock cycle. It is however considerably more expensive and power hungry than my own solution.
 
If minimum chip count is your aim then Tam Lin's schematic (left in the comments on my blog) is easily the winner - it would probably need just one extra half-chip to delay WS by a single clock cycle. It is however considerably more expensive and power hungry than my own solution.
I don't know enough about the Lin design to tell if it'll work. My prev. experimentation with a several diff. designs utilizing fewer logic ICs did not work.

BTW ... from that same blog page, looks like you took a crack at this issue yourself (??) here ...
x1134d1379294813t-i2s-transcoder-p3110065.jpg
 
TDA1545A still any "good"?

Yes, that was my second build of the device whose schematic is shown (albeit with one correction). Still working fine.
Just curious ... all else held equal (which is almost impossible to do)...how would you rate (compare) the TDA1545A/TDA1387 to other DACs (new or classic)?

I've got a bunch of TDA1545A's lying around ... thought they were jellybean DACs ... but they tend to pop up on several serious-DIY posts/blogs (Peufeu, diyparadise, etc) ... so they can't be all bad and maybe worth a project or two. Also, IIRC, it was the last serious R2R ladder DAC Philips produced -- ~1997 -- before everyone went Delta-sigma.

I'm hugely behind the DIY times ... for all I know, the latest ESS or WM DACs -- or even TI's cheap new stuff -- could simply be orders-of-mag. better than these classic DIP-DACs many DIY geezers cling to (too stubborn+old to try anything new -- or they can't see/solder smaller than DIP). But I'm equally behind the times as I haven't been ... uh ... keepin' up by readin' the 'papers -- need stronger bi-focals ;)
 
Last edited:
In my estimation, none of the new stuff (high oversampling, D-S based) holds a candle to the TDA1387 and TDA1545A. Given that they were designed as 'economy' parts, they're almost magic chips. Neither are R2R ladder, they're segmented current source = better dynamically than R2R types due to lower glitch energy. Actually the TDA1545 (non-A) is an older design - I got some samples I think around 1991, this part was I2S.

The only 16bit part which is better than these babies would be the TDA1541A - superior because its bipolar and hence inherently lower glitch than CMOS parts. But hardly suitable for what I plan to do, which is design and build a portable DAC.
 
TDA1545A DACs

But hardly suitable for what I plan to do, which is design and build a portable DAC.
Late 2006, I built a TDA1545A (DIP) portable, partly based on the Monica design. Wasn't all that happy with it ... but then, early 2007, Rudolf B's very-good (discrete) I/V -- designed mostly for the Monica -- was posted both here and on diyparadise. And that was a MAJOR upgrade. Alas, the the whole thing needs 15-24v to work (24v is best). And the whole unit draws ample current ... so one needs a really good battery solution for portable use.

All that said, one could just go off the TDA1545A datasheet, use decent low-power opamps (AD8397) and other SOIC-sized components, plus a few Li-Po cell-phone batts and DC-DC converter with good after-filtering.

---

About the sonics of certain classic multibit R2R DACs ... the sound is BIG, with large soundstage, lots of dynamics and deep, tight bass. Seems that Delta-sigmas moved away from this into more wimpy sound to emulate over-rated analog/vinyl.
 
RMAF 11: Noise Shaping Sigma Delta Based Dacs, Martin Mallison, CTO, ESS

I'd say that the D-S DACs of today more emulate cassette tape than vinyl.
It's funny ... some of the last cassette tape decks -- as well as those rare ones that could R/P greater than 1 7/8 ips, Dual made one IIRC -- were not bad. Dolby HX Pro, etc helped a little.

But back to D-S vs. R2R (heck, we can even throw in MASH and Bitstream 1-bit jobs while we're at it)... what sonic characteristics are you referring to?
Indeed, ESS and others are aware of the most criticisms of D-S DACs, and use these criticisms when developing new models...
RMAF 11: Noise Shaping Sigma Delta Based Dacs, Martin Mallison, CTO, ESS Technology - YouTube
 
But back to D-S vs. R2R (heck, we can even throw in MASH and Bitstream 1-bit jobs while we're at it)... what sonic characteristics are you referring to?

Noise modulation, greyness of tonal colours, missing dynamic contrast.

Indeed, ESS and others are aware of the most criticisms of D-S DACs, and use these criticisms when developing new models...
RMAF 11: Noise Shaping Sigma Delta Based Dacs, Martin Mallison, CTO, ESS Technology - YouTube

Its an excellent presentation, one I've commended to others on numerous occasions. However for all the fixes they've incorporated into their 'Hyperstream' modulator, the noise modulation's still very much there.
 
*Implementation* is the key

Noise modulation, greyness of tonal colours, missing dynamic contrast.
Given some of the advantages of classic R2R designs -- as noted in certain DIY communities, but also some modern portable gear (HiFiMan) and a few high-enders (Zanden), it's curious why Philips, Analog, TI/BB, et. al., no longer pursue R2R (or, at least, manuf. their classic chips). Certainly, the old tool and dies exist.
Maybe the manuf. costs (vs. REAL demand) is too high.
D-S was orig. developed to reduce manuf. costs, and $$ is always the bottom line.
Still, with modern advances in IC manufacturing, SOP-sized R2R chips is not THAT hard (or cost prohibitive).

The most common (and reasonable) excuse -- one that even a lot of DIYers admit to -- is that DAC type (D-S, R2R, etc.) is small potatoes compared to IMPLEMENTATION (clean PSU, better LPFs, I/V and other output-stage elements). PCB layout, passive-component quality and tolerances are important "implementation" factors, too.

I noted earlier that the orig. TDA1545A-based Monica (w/o Rudolf I/V) sounded pretty unspectacular. In fact, at that time, I had a Toshiba mod CD/DVD player in the works-- it used one of those all-in-one Zoran multimedia chips (with D-S DAC of course) ... but the Zoran was unique in that it outputted current (!!), and Toshiba used its resources to come up with a fairly impressive (but cost effective) I/V and high-order output stage, which made certain Toshiba models popular in DIY circles. Anyway, back to Monica .... that $30 (new Wal-Mart price!) Toshiba with a $2 opamp mod blew away almost every CDP and D/A I had at the time.

And I still have that infamous single-bit Optimus CD-3400 from 1994 ... euphonic and colored as it is, somehow Mistumi's IMPLEMENTATION stinks of vinyl and vacuum tubes.
 
Given some of the advantages of classic R2R designs -- as noted in certain DIY communities, but also some modern portable gear (HiFiMan) and a few high-enders (Zanden), it's curious why Philips, Analog, TI/BB, et. al., no longer pursue R2R (or, at least, manuf. their classic chips). Certainly, the old tool and dies exist.

First up, you're not talking here about R2Rs - Hifiman used TDA1543 and Zanden TDA1541A - neither are R2Rs. Philips (now NXP) no longer seem to be designing audio DACs. It appears they all found they could get better measurements for lower cost with S-D designs; chips that could be built on digital processes. R2Rs were always expensive due to the need for laser trimming.

Until we get standardized measurements which more accurately reflect perceived SQ I can't see multibit DACs making a renaissance. The vast majority of designs are done based on DS numbers, not listening evaluations.
 
Segmented current sources vs. Delta-sigma vs. Ladder/R2r

First up, you're not talking here about R2Rs - Hifiman used TDA1543 and Zanden TDA1541A - neither are R2Rs.
I usually use R2R as "generic" name for classic multibit ladder architectures.
But R2R seems only to be specifically mentioned in datasheets for DACs like PCM1704, etc.
Also, MSB Tech, Mother-of-Tone, HyperPhysics seem to limit discussion of DACs arch. to
R2R/ladder, D-S, and "summing amplifier".

In an earlier DIY topical discussion, you also refereed to another type (arch.), segmented current sources (with or w/o DEM), putatively used in TDA1541
Is this what TDA1543 and TDA1545 use, too?

Also, is there a White Paper on Philips use of segmented current sources? It does not seem to be mentioned all that often in datasheets, etc.
Think this one from AD is applicable.

Or select TDA1541/43/45 arch. (if possible) from the Types shown on wiki DAC page:
http://en.wikipedia.org/wiki/Digital-to-analog_converter#DAC_types
 
Last edited:
Further thoughts...

I think the CD (optical disc) medium -- and its inherent, rather messy/flawed way of reading/interpolating/error-correction (all in real time) -- are important drawbacks that compromise sonic performance. (To say nothing of ergonomics or utility or limited usability, such as playback lists)
Since D-S are so popular (for manuf. reasons, as prev. noted), they are incorporated into flash-memory devices and PC soundcards, that have the advantage of cache memory and buffering ... giving D-S the IMPRESSION of sounding superior.
I have heard classic multibit DACs with modern devices (e.g., QLS hifi wav player --> (using I2S) ---> TDA1545A-based D/A (with Rudolf I/V). And that's ... quite a different story ... maybe, it's the transport, stupid!
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.