Using TDA1543A with Teensy 4.0

[PaulS]

Hallo Frank,

Thanks. Tested your code and the sine looks fine on the oscilloscope.
Hooked up the logic analyzer and saw this:



So apparently the TDA1543A needs japanese (2's-complement) data as 32+32 bits per sample? WS = 44.1 kHz, BCK = 2.84 MHz.

Paul

 

[Frank B]

Hallo Frank,

Thanks. Tested your code and the sine looks fine on the oscilloscope.
Hooked up the logic analyzer and saw this:

View attachment 25279

So apparently the TDA1543A needs japanese (2's-complement) data as 32+32 bits per sample? WS = 44.1 kHz, BCK = 2.84 MHz.

Paul

I've seen this format the first time.
After WS, it needs 7 dummy cycles (Pazis code set them to '1' - I removed that since it doesn't seem to be needed - see previous post), then 16 data bits, the rest gets ignored.

 

[PaulS]

From the www.diyaudio.com forum, message #2 and #3 seem to support that dummy cycle theory.

Paul

 

[Frank B]

indeed, yes.

 

[PaulS]

From the datasheet of the Sony CXD2500BQ [CD Digital Signal Processor] that was used in Marantz CD players and could drive a TDA1543A directly:



The chip outputs data at 48*LRCK rate [2.12 MHz]. The first 8 bits are ignored. And apparently the TDA1543A ignores bits >24.

The Philips SAA7345 signal processor, used in Philips CD players to drive the TDA1543A, does the same:



Paul

 

[Frank B]

Year, Philips forgot to mention that in their TD1543A Datasheet...

Or is that a std japanese format?

 

[PaulS]

Thinking about it a bit longer, I think what the TDA1543A needs is a right-justified I2S signal. So the edge of the LRCK [aka WS] signals the end of the incoming left or right data.
That data can be 16 or 24 or 32 bit long [or 100 bits for that matter] - the TDA1543A just accepts the last 16 bits before the LRCK edge and discards all previous bits.
Having said that and looking at the logic analyzer snapshot: this means that the 8 least-significant bits of our digital sine word are seen as the 8 most-sigificant bits of the sine word since the last 8 bits before the LRCK bits are zero... Maybe the picture below makes it more clear. Is the TDA1543A only looking at the data in red rectangle?



I will check the actual analog signal in detail on the scope tomorrow. My brains are done for today...

Paul

 

[Frank B]

The 16 data bits are in the middle of the 32 Bit bit (8 msb and 8 lsb ignored..)
Or it works with (2x)24 bit frames , too. Have not tried that. Edit: The SAA7345 seems to use that.

Would be nice to know: Does the PT8211 work too with this format?
I remember the issue that (other than advertised) it does not output left and right in sync. Maybe it does, with this format?

 

[PaulS]

Yeah, I guess the TDA1543A just assumes 24-bit right-justified 2's-complement data [some EIAJ format] and simply discards the first 8 bits [and whatever comes after the following 16 bits]

Would be nice to know: Does the PT8211 work too with this format?

Unfortunately not:



Yellow trace: TDA1543A, 1KHz, full scale output. Purple trace: PT2811.

Gruesse,
Paul