AD1955 192 kHz DAC with async. upsampling

[costiss]

i am soon launching on a new dac on the analog devices ad1955 192 kHz chip. ic chain shall be: dir1703, ad1896, 2 x ad1955, 4 x opa627bp, 2 x ad8610 + at90s2313 mC to load the registers of the ad1955s to have them operate in mono mode. No complex filtering, since all signal is upsampled to 192 kHz by the ad1896 (sample rate conv.). outputs will be se, in the tpology ads sheet proposes . cap values before and after the differential mixer(?) -ad8610- could be lowered a bit to increase the bandwidth (75khz).. no external digital filter seems to be required to enable mono mode , is that right??.. any comments- remarks are welcome..
pcbs on the way, analog part of the board (ad and the ops) has been designed, but the analog ground plane. anybody with a jung reg for 5v?..
opinions?
nnsense?

[BrianGT]

Originally posted by costiss
i am soon launching on a new dac on the analog devices ad1955 192 kHz chip. ic chain shall be: dir1703, ad1896, 2 x ad1955, 4 x opa627bp, 2 x ad8610 + at90s2313 mC to load the registers of the ad1955s to have them operate in mono mode. No complex filtering, since all signal is upsampled to 192 kHz by the ad1896 (sample rate conv.). outputs will be se, in the tpology ads sheet proposes . cap values before and after the differential mixer(?) -ad8610- could be lowered a bit to increase the bandwidth (75khz).. no external digital filter seems to be required to enable mono mode , is that right??.. any comments- remarks are welcome..
pcbs on the way, analog part of the board (ad and the ops) has been designed, but the analog ground plane. anybody with a jung reg for 5v?..
opinions?
nnsense?

If you want an easier design path, requiring no microcontroller, you might want to consider the TI PCM1794 for the dac. This has most all the features of the ad1955, but uses HW control instead of SPI/I2C.

I am designing a similar dac, and have so far worked out the analog i/v stage for the dac:
pcm1792 I/V stage layout
and I will also be using jung regulators for the 5v dac, and also + and - 15v. Here is my status on them:
another jung super regulator pcb layout
I am concurrently working out the interface with the digital side with the AD1896 upsampling the data. If you want to run full 192kHz, you will have to use the AD1896 in slave mode. An alternative to this would be to go with the TI version of this part, then SRC4192 (SRC4193 is the SPI version instead of HW control). The SRC4192 is pin compatible, but supports a few more features, such as 192kHz resampling in master mode.

If you want more information/want to coordinate some efforts, drop me an e-mail.

I have had a hard time deciding between HW control and SPI control for the 2 chips, as I will already have a PIC microcontroller in the project for my cdpro2, but if I go for HW control, then others can adapt this project for their own needs, and it can be a stand-alone dac, apart from my cdpro2. The HW control parts are much more DIY friendly

Mono mode doesn't seem hard to implement, but I am wondering if it is overkill, since the dac already has differential output. I also thought about making a board using the PCM1794, where one board could be used for stereo, or 2 boards could be used for mono, simply inverting the L/R clock for the second board.

--
Brian

[costiss]

digital audio TI parts are a real pain to get in greece, thats why src is probably out of the question.. sampling it from ti doesnt seem to be possible (refered to as non stock).. i think im going on with ad1955 in mono mode.. it wont be too difficult.. just to load one register... inputs are probably going to be 1 optical + 1 aes/ebu, maybe an spdif, too.
have u got to a decision about the master clock?
thats what ive got to with the board..
please comment..
i dont want to print the board three times to finally get it right...
(this has happened....)

[Elso Kwak]

Originally posted by costiss
digital audio TI parts are a real pain to get in greece, thats why src is probably out of the question.. sampling it from ti doesnt seem to be possible (refered to as non stock).. i think im going on with ad1955 in mono mode.. it wont be too difficult.. just to load one register... inputs are probably going to be 1 optical + 1 aes/ebu, maybe an spdif, too.
have u got to a decision about the master clock?
thats what ive got to with the board..
please comment..
i dont want to print the board three times to finally get it right...
(this has happened....)

Hi Costiss,
Are you using Boardmaker?
Picture looks familiar to me.
If the part is not in stock no samples are available. This situation may change over time.

[costiss]

yes.. its boardmaker im using.. old program (dos) but i think it is easy to use, really functional and has a nice library editor..
i have already ordered the ad1896 and have the 1955 in hands.. also the opamps, too..

[Elso Kwak]

Originally posted by costiss
yes.. its boardmaker im using.. old program (dos) but i think it is easy to use, really functional and has a nice library editor..
i have already ordered the ad1896 and have the 1955 in hands.. also the opamps, too..

Hi costiss,
I love it!

[BrianGT]

Originally posted by costiss
digital audio TI parts are a real pain to get in greece, thats why src is probably out of the question.. sampling it from ti doesnt seem to be possible (refered to as non stock).. i think im going on with ad1955 in mono mode.. it wont be too difficult.. just to load one register... inputs are probably going to be 1 optical + 1 aes/ebu, maybe an spdif, too.
have u got to a decision about the master clock?
thats what ive got to with the board..
please comment..
i dont want to print the board three times to finally get it right...
(this has happened....)

I haven't worked out the clocking yet. I have the chips all on dip adapters, so I will prototype the whole thing before I get the board layout made .

As for the layout, How are you running power to the last opamps? They seem boxed in. Also, how did you come to the decision to use 2 different types?

--
Brian

[costiss]

i corrected the layout to provide powe to the 8610s.. just switched two tracks to top layer... ill put the 4 i/v opamps on bases to have some flexibility.. i thinkthe8610 is the sure bet for the differential to single ended conv... the 627 is just the initial choice for the i/v..
added the ground plane...
anybody knows if i could have two seperate avdd running to the psu with seperate LC filtering after the reg?...
im putting the receiver, upsampler and microcontroller on a seperate board (maybe the master clock , too), connected with flatline to the dacs board (as in the elektor dac 2000)
could this cause oscillations? the i2s lines will have an 22r resistor in series with the cable..
the pcb with the top ground plane..
a jung reg schematic for 5v?

[costiss]

im back.. after some time.. have been experiencing some problems with my pc..
havent been around for quite some while..
i am getting the dac&i/v board on friday, still waiting for the ad1896...
i now have to go to the rec. - upsampler section..
thats the final board:..

[takashi]

dear all, do your DAC finished?

i use ad1955 with pcm1738/1730 i can hear the differents!
but it have a trouble in using 1896.

in the mode selection, i cant get a "normal sampling" 44.1k and 192k without changing the oscillator.
and "bypass" pin seems works far away from my expected

since i set the mode to 512Fs / 256Fs(lets ignore 768Fs), when the Rclk is 22.5792Mhz, the LRCLK out can be selected to 88.2k and 44.1k.(cannot reach to 192k of course)
and then i use and external Rclk, 49.XXXMHZ, i can get the 96k/192k selectiable in LRCLK out.
but i cannot restore to 44.1/48k by any mode without changing the oscillator...

in fact in the spec. there is a "bypass" pin, i suppose it can bypass the I2S to output port, but i cannot get back to 44.1/48K when changing this pin high / low

i also find that SRC4192 can do that , since one of the mode is 128Fs, so if i use fix 24.5796Mhz , i can get 48k, 96k and 192k by changing modes. no matter the bypass pin work or not...

i want to get all sampling rate is because i want to compare the differents.

can you share the experience / finding?