Sony to Philips I2S Format Conversion

Hi,
Anyone has tried the attached schematic? I got sound output from the TDA1543 but sound is way too low in volume.
The schematic is a redraw of Kal Rubinsons.
Should I delay the LATCH instead of the DATA???:confused:
 

Attachments

  • sony-to-i2s-gif.gif
    sony-to-i2s-gif.gif
    7.3 KB · Views: 4,354

rfbrw

Member
2001-10-26 11:51 pm
.
It might be a good idea to define what actual format you are starting with. The Sony format is NOT the same as the Sony/BB format. It would seem that far too often the two get mixed up,not to mention the use of I2S as a blanket term for all formats.

ray.
 
Sony Format

rfbrw said:
It might be a good idea to define what actual format you are starting with. The Sony format is NOT the same as the Sony/BB format. It would seem that far too often the two get mixed up,not to mention the use of I2S as a blanket term for all formats.

ray.


Hi Ray,
The format I start with is the format Sony is using before the digital filter (CXD1244S) in the Sony CDP-X33ES.
From past experience I know this format works well with AD1851 DAC NON-OS, so I guess it is Sony/BB format. Works also with PCM56 DACs NON-OS. I figured out is has room for 24 bits, but of course only 16 bits are used. Bitclock= 48 x 44.1 kHz = 2.1168 MHz.
Anyway this is what I want: use the TDA1543 NON-OS with the above mentioned Sony CD-player.
I agree on avoiding the blanket term for I2S. This has given rise to very much confusion. It should be clear I mean the Philips I2S format as this is used by the TDA1543.;)
Thanks in advance.:)
 

rfbrw

Member
2001-10-26 11:51 pm
.
Though not wishing to sound like the man who, on being asked for directions, replied "I wouldn't start from here if I were you", observation and experience would tend to suggest that what you have is the Sony format with it's fixed 48Fs serial clock and that,as previously posted, calls for the TDA1543A.
Should you still need to derive I2S, you will need to delay LRCK by 8 serial clock cycles and then invert it.

ray.
 
Formats

rfbrw said:
Though not wishing to sound like the man who, on being asked for directions, replied "I wouldn't start from here if I were you", observation and experience would tend to suggest that what you have is the Sony format with it's fixed 48Fs serial clock and that,as previously posted, calls for the TDA1543A.
Should you still need to derive I2S, you will need to delay LRCK by 8 serial clock cycles and then invert it.

ray.

Hi Ray,
I know you don't agree with me but I did not get the TDA1543A working with ANY format coming from the CS8412.
As the AD1851 works perfectly NON-OS with the CS8412 in outputmode 5 and with the Sony I concluded both formats are the same i.e. LSBJ justified, meaning the LSB is justified to the end of the audioframe and the MSB varies with word length.
I have two TDA1543A delivered to me in error. So I guess it is quite easy to try these on the Sony.
I could be wrong but my guess is the TDA1543A is for offsetbinary signals unavailable from the CS8412 but possible with some Sony chips like CXD1125Q when set up in the offsetbinary output mode. Pulling pin 58 high puts out offset binary, low for this pin is two's complement.
I will try to delay the LRCK too.
;)
The TDA1545A is for outputmode 5 of the CS8412. Tried that one too.
I am sorry for the confusion, not my fault but Philips.
Ever wondered what the A meant for the TDA1541? Here it is a quality/selection grade.;)
 

rfbrw

Member
2001-10-26 11:51 pm
.
Elso,
You do not have to accept this, but there was digital audio before the arrival of the CS8412. I cannot see how one can arrive at the conclusion that because the '1543A does not work with the '8412 it must be offset binary not 2's comp. The '8412 was preceded by a number of digital audio I/O chips, most of which were hard to get,very expensive or both. More to the point the relationships between LRCLK, SCLK and MCLK are fixed at Fs, 64Fs and 256Fs respectively on the '8412 which is not the case with I2S and Sony/BB formats and if one takes the time to understand the structure of the two formats, it quickly becomes apparent that the the Sony/BB format encompasses the Sony format.
As for the TDA1545A it states on the second and eleventh pages of the datasheet exactly what is required in terms of formats. Philips put it in writing and they give a picture and short of directly injecting an understanding into the users brain I cannot see what more they could have done bearing mind that these datasheets were not meant for the layman.

ray
 
A

rfbrw said:
Elso,
You do not have to accept this, but there was digital audio before the arrival of the CS8412. I cannot see how one can arrive at the conclusion that because the '1543A does not work with the '8412 it must be offset binary not 2's comp. The '8412 was preceded by a number of digital audio I/O chips, most of which were hard to get,very expensive or both. More to the point the relationships between LRCLK, SCLK and MCLK are fixed at Fs, 64Fs and 256Fs respectively on the '8412 which is not the case with I2S and Sony/BB formats and if one takes the time to understand the structure of the two formats, it quickly becomes apparent that the the Sony/BB format encompasses the Sony format.
As for the TDA1545A it states on the second and eleventh pages of the datasheet exactly what is required in terms of formats. Philips put it in writing and they give a picture and short of directly injecting an understanding into the users brain I cannot see what more they could have done bearing mind that these datasheets were not meant for the layman.

ray


Ray,
No I don't accept this. The TDA1543A is not mentioned in any Philips datasheet, whatever format it is for. I came to the conclusion the TDA1543A is for offset binary format after studying a Philips digital audio file, about all kinds of digital audio chips and the datasheet of the CDX1125 indicates it can be put in the offset binary mode. Also the Aristona CDP of my daughter had a Mitsubishi chip and the TDA1543A.
I got the TDA1545A working with the CS8412.
For a very good friend I want to build a NON-OS DAC in the Sony with a TDA1543.
That's all:hot:
Tomorrow I will try the TDA1543A with the Sony.
I will let you and the forum members know the outcome of this experiment. BTW the datasheet of the CXD2500Q in the Sony mentiones the output is 2's complement format. Are we talking at cross purposes perhaps?:xeye: Please define what you understand as the Sony or Japanese format.:confused:
The last thing I want is to bring confusion. I have several times answered posts on the Audio Asylum and here why the TDA1543A does not work with the CS8412.:angel:
 
TDA1543 with a Sony CDP

Peter Daniel said:
Hi Elso,

I would be quite interested in your findings, as I still have this project to finish http://www.diyaudio.com/forums/showthread.php?s=&threadid=16102&highlight=
and I already got my TDA1543A chip.
I also got sound when connecting TDA1543, but it was very low in volume.
Hi Peter,
I will let you know. I have the TDA1543 and the TDA1543A also.
Actually I bought a Philips CD931 to avoid all these problems.
;) ;)
The alternative is to use the SPDIF interface but I was so happy with I2S direct.:angel:
 

cm

Member
2003-09-05 4:58 am
Singapore
yes - down with SFDIP. I2S is the way to go. The complexity of glue logic for Sony to I2S is probably simplier than SPDIF circuitry ( and a lot cheaper ) once it is understood.

I have downloaded the I2S standard from the Philips website but could not find anything anywhere on the Sony standard. Anyone has it so that I can look thru' the timings and confirm it once and for all? Right now, I am doing the same as Elso - reverse engineering from multiple datasheets.

BTW, there is no need to invert the LRclk - just use the right output as left and vice versa. A few wiring less to do!
 

rfbrw

Member
2001-10-26 11:51 pm
.
Elso,
That you have not seen a reference to the TDA1543A in any Philips literature, you will, I hope, agree, is not a cast iron guarantee that none exists. I have seen reference to it in one of those long gone 1000+ page tomes that semiconductor manufacturers used to produce. It is also listed in the Discontinued Devices manuals produced by D.A.T.A.
I too have a number of Sony datasheets and they define the Sony format the same way Philips do, 2's comp LSB justified data,48Fs serial clock and 48bit L/R frame irrespective of valid data length. If you understand this and the Sony/BB format, it is immediately apparent as to why the Sony/BB format encompasses the Sony format and why the AD1851 works.
You say you do not wish to cause confusion yet, without any evidence to the positive, manufacturer or otherwise, you constantly assert that the TDA1543A is not 2's comp. Worse still, you assert this on the basis that it does not work with the CS8412 as though the CS8412 datasheet were a digital audio tablet of stone. Well, in that case what conclusion should I draw if,for example, I cannot get the SAA7274 to work with the PCM1702?
One might as well claim that as the sun rose after the **** crowed, the sun rose BECAUSE the **** crowed. What possible value is there in constantly answering the same question with the wrong information? Then again, it is said the road to hell is paved with good intentions.

By the way, Elso, the the TDA1543A can be made to work the CS8412, it just requires more effort than most are prepared to invest.

ray
 
Totally Stupefied

Hi Ray,
It works!!! The TDA1543A works with the CXD2500Q in the Sony without any glue logic (NON-OS)!. So you were right and I was wrong about the offsetbinary format for the TDA1543A.

I am totally confused :confused: now as the TDA1543A does not work with the CS8412 in mode 5, but the AD1851 does work with the Sony AND with the CS8412. I simply don’t see the difference in the formats coming from the CXD2500 and coming from the CS8412. Only difference I see is that the CS8412 has a 32 serial clock periods per audio sample and the Sony 24.
I would be grateful if you could enlighten me.

I am very happy you have resolved my problem, and probably also Peter's! Thanks!:angel:
 
Schmitt

Brian Guralnick said:
Replace the HC14 with HC04, or AC04, the HC14's schmit input has an incredible amount of phase noise on the output.
Hi Brian, Ooops you are right. I got this schematic from someone else. Actually in my experiment I used the 74VHC04 and the 74VHC74. I was too lazy to convert the schematic suitable to post on the forum as I had this on file.
AC logic should be avoided like the black pest.
;)
 
Formats

borisov57 said:
Hi Elso,

Format comming from CXD2500Q is two's complement, right justified and with 48 bitclocks per wordclock. I guess ( to lazy to look in datasheet), that CS8412 has 64 bitclocks per wordclock.

Hi borisov57.
That's exactly what I said in post # 14.
One wordclock period is comprised of two samples (left and right channel data).
http://www.diyaudio.com/forums/showthread.php?postid=253040#post253040
;)
 
I do not think so as the CXDxxxx chips output EIAJ. One of the modes of the TDA1541 is I2S, which is similar to to EIAJ except for polarity of the LRCLK, and when the data starts during the rising or falling edge of LRCLK.

To convert EIAJ to I2s, you need to invert LRCLK, and delay it by several clock cycles. To convert I2s to EIAJ, then invert LRCLK, and delay data by several clock cycles. The number of clock cycle delays depends on if you have 16,18 or 20 bit data coming out from the digital filter but I think the TDA1541 accepts only 16 bits data.

You can see a comparision of the waveform of EIAJ versus I2S side by side in the SAA7345 datasheet from Philips.

I have posted the schematic and description of my EIAJ to I2S converter somewhere in the forum about "sharing my digital work" or somethng like that.

I use I2s direct from a CXD1135 to TDA1541A.