Thread: HD Radio Optical Digital Output Hack

So after some research, I figured out the basics of how the DMHD-1000 radio is designed. From the antenna input - the AM/FM signal is filtered and tuned-in by a Philips microcontroller and the in-band digital FM signal is repeated to a TI IF decoder. The AM/FM/IB-FM signals are all MUXed into a TI baseband controller (BGA) - The baseband controller also formats the GUI for the LCD (channel/band/RDS info) and translates I2C bus I/O from the remote and the other decode/controller chips on the bus. Now all I need is a data sheet or reference design manual for the TI baseband chip so I can put a scope on the pins that have the desired SYNC, SDATA, SCLK and MCLK signals. Mitch has informed me the MCLK is probably different than that of an XMDIRECT. Does anyone have access to the Texas Instruments TMS320DRI300/350 datasheet with signal information?

Summary of files-
dit4096.pdf (538.5 KB) - TI datasheet on XMPCR Toslink adapter chip.
CS4344-45-46-48_F1.pdf (518.6 KB) - CL Datasheet which represents the closest D/A converter I could find.
CL4334b.jpg (79.5 KB) - Image (Blurrycam,sorry) showing via tap rework to the suspect D/A converter.
pcr pinout2b.jpg (91.2 KB) - Image (Blurrycam,sorry) showing how I connected the Toslink module to the D/A converter.

After some painful via rework, I connected the XMPCR Toslink board to the D/A converter on the DMHD-1000 (The clear stuff is epoxy resin because if you try to move these wires without it, you lift the via/trace and destroy the connection.
I confirmed this is the D/A converter chip (but not its part number) by listening to the analog output on the Cirrus Logic pins of the TSSOP10 package that corresponded to what I suspect is the right data sheet for this converter. The only markings on this chip are two four-digit characters (344D 0632). No data sheet exists for this exact part number that I could find, but it is the only 10-pin TSSOP solution that was close to the part number scheme and seems to match at least on the right side of the package, the pinouts described in the datasheet.
The other indicators that I am close to getting this working are the radio did not blow up or short out, the radio had retained the same audio quality and functionality before I tried to hack it and the TOSLINK light came on. I connected it to an optical input on a DENON HT system and could not get audio to be detected.
So it looks like I am not as close as I thought... just because I got a light does not mean workee workee yet.

This is where I need some help from anyone who understands the XMPCR Toslink adapter.

First, here is how I connected the wires -

SYNC=LRCK
SDATA=SDIN
SCLK=DEM/SCLK
MCLK=MCLK

On the left we have the XMPCR Toslink board signals and on the right we have the D/A converter input signals.
(If I connected this wrong, please let me know.)
Second, I suspect three other possible issues for no Optical audio output being detected:
a. It is possible the data signals are not flipping the logic in the dit4096 due to a voltage threshold issue (let me know if the resistor values on this converter board need to be changed)
b. The dit4096 may not have enough current on the 5v line to work properly (what is the current load for this adapter?)
c. The data signals from the TI Baseband chip are not in the correct format for the dit4096 driver chip (I don't think this would be an issue as the CS4344 and dit4096 have similar signal sync specifications - but I could be wrong)

Next Steps -
I have supplied as much information as I can at this point, I hope someone in the community can help me out here. Hopefully the attachments will help make troubleshooting this easier.

Well this is impressive, good luck

First off, Massive props to Mitch. Without his help, this project would have stopped at my last post.
I got the MCLK for the D/A converter on a scope and it looks like the clock is at 11.2612 MHZ. This is close to the 11.2896 specified MCLK in both datasheets. This means the audio data coming from the TI Baseband chip is most likely in 44.1khz mode. I think the difference in measurement could be that my scope is out of calibration and I only had it at 20ns resolution.
I decided to connect the optical output to my M-Audio transit and recorded the output. There was audio, but it sounded way over-driven. I recorded the audio and included is a picture of what the captured audio looks like. It would seem the audio is losing sync at a very fast interval and I was not hearing distortion, but audio being pulsed on and off very fast. The audio dropout interval is .0089sec with an ON interval of about .00262sec. Does anyone know what could cause this?

Ive got nothing useful to add since your electronic skills are more thorough than mine, however I just wanted to say that this project looks incredible and I hope you figure it out.

With help from Mitch, an o-scope and a test jumper rig - the secret sauce has been discovered. Below you will find a screenshot of the analog output of my soundcard from the toslink source and my jumper test rig to flip through the MCLK and format selections. I verified there are no audio artifacts, and it is clearer than the standard analog out (by the way if you use the RCA connectors you may want to ground the RCA headers as these cheap connectors introduce a horrid sound over analog when not grounded properly). On the scope, the output signals were as follows:

MCLK 2.5v 11.28Mhz (Constant)
SCLK 2.5v 2.82Mhz (Average)
SDATA 2.5v 200Khz-1.4Mhz (Variable)
SYNC 2.5v 44.1khz (Constant)

This simply confirmed I had the right signals and connected the wires correctly, but gave me little else. Mitch suggested fiddling with the MCLK selection on the Toslink adapter which required lifting pins 4 and 5 on the DIT4096 which is not an easy task. I decided to also lift pins 9 and 10 for audio format selection as these were the only other values I could change. I connected all four lifted pins to blue-wire then jumper headers and glued them in place. I also connected 8 more jumper header pins (4 for GND and 4 for VCC) to give me the ability to cycle through all 16 combinations. Turns out, all I had to do was change the MCLK mode to 384xFs which the DIT4096 datasheet does not mention supports 44.1khz audio mode. This is another instance where the datasheet can be misleading.
So, for future reference, if you want to use an XMPCR Toslink adapter on a DMHD-1000 HD radio, connect the main pins as I did above and rework your toslink adapter to swap pins 4 and 5 on the DIT4096 (Or lift pins 4 and 5 on the DIT4096 and blue wire pin 5 to +5 and pin 4 to GND.)
Thanks to Mitch and the community for your interest in this project - Now I can use Mitch's HD radio cable and software to integrate my XMDIRECT and DMHD-1000 both with Digital Optical Audio out through my toslink switch.

For those of us playing along at home, can you confirm what types of broadcast are output over the toslink. Are only the HD channels available, or does the tuner digitally encode the analog FM and AM signals also?

Many thanks to you and Mitch for this modification.

Fortunately, TI and iBiquity decided to make this HD radio solution as cheap as possible and chose to digitize AM and FM and Encoded HD FM all on the same Baseband chip. So to answer your question, all bands can be heard. I verified that all AM/FM/HD-FM signals were being repeated as I was concerned about that also until I understood the design of the HD radio. Take a gander at the only documentation I could find on the TI baseband chip - the block diagram shows the light at the end of the tunnel - One final D/A coverter before audio output (which forces anything this radio puts out through one chip)

Incredible.