Drive NOS AD1865/62,PCM1704/02/63,TDA1541 from FIFO: Universal I2S-PCM driver board

Last ideas:

- some capacitors seems to be missing around PCM1704 (C39, C41,...) - these seems to be +- decoupling of digital supplies. Are these soldered on bottom-side of the PCB, if not - why do you removed them? To keep your wires on top - try to solder these caps on the bottom-side (watch for polarities)
- is DF1704 powered? without input it's outputs are probably low - being connected to the DACs these might interfere with the other signals. If it's easy (e.g. unsoldering a ferrite-bead) try to un-power DF1704
 
Member
Joined 2009
Paid Member
Last ideas:

- some capacitors seems to be missing around PCM1704 (C39, C41,...) - these seems to be +- decoupling of digital supplies. Are these soldered on bottom-side of the PCB, if not - why do you removed them? To keep your wires on top - try to solder these caps on the bottom-side (watch for polarities)

The +- psu goes directly to PCM1704 so C39, 41 etc aren't necessary

- is DF1704 powered? without input it's outputs are probably low - being connected to the DACs these might interfere with the other signals. If it's easy (e.g. unsoldering a ferrite-bead) try to un-power DF1704

Nothing more of the obard is powered only the PCM1704
 
Last edited:
Must read articles about bypass caps roles: "Successful PCB grounding with mixed-signal chips": part 2
Also read part1 and part3.

Even if it goes directly from PSU they should be there, so as a last try, you could try to put them back (to the bottom to leave your wires on top) at least for one DAC.

edit: link corrected
 
Last edited:
Must read articles about bypass caps roles: "Successful PCB grounding with mixed-signal chips": part 2
Also read part1 and part3.

Even if it goes directly from PSU they should be there, so as a last try, you could try to solder them back at least for one DAC.

edit: link corrected

agree, local decoupling isn't a nice too have its a necessity.
With the PCM1704 you will also fid that high vales for the servo decoupling benefit in disabling the stupid BPO, but you need to get 192khz working first correct?

What speed is your master clock ?
 
Last edited:
The datasheet says:

Maximum Bit Clock (BCLK) Rate
The maximum BCLK rate is specified as 25MHz. This is
derived from the 8X oversampling of the PCM1704. Given
a 96kHz sampling rate, an 8X oversampling input and a
32-bit frame length, we get:
96kHz • 8 • 32 = 24.576MHz

192khz is basically 4x so the timing of the bitclock speed should be plenty. The thing that sticks out is you have no fifo and no PLL, did you bypass the reclocking done on the amanero board and run the clock straight to the pcm driver board? If you are trying to reclock twice that may be the problem.
 
Member
Joined 2009
Paid Member
Regal, there is no reclocking whatsoever going on. All 4 signals (MCK, BCK, WCK and DATA) go straight from the WaveIO "as is" to Ian's PCM board and from there to the input pins of the PCM1704 as shown in the pics I posted but here only BCK, WCK and DATA are used. There is no active XO on the 1704 board and DF1704 is powered off, in fact all the board is off with the exception of the DAC ICs.
 
The PCM board I am sure takes the MCK to reclock. The waveIO doesn't have a reclock but the Amerano does. I still doubt that either would work with the Buffalo's on minimal PLL.

I guess what I am saying is you are feeding a DSP-less DAC chip with no PLL (the fifo is a form in way) and that is quite a challenge for 192khz. There is no PLL no "Buffer". Ian said that 192khz is rarely done without missed/skipped samples with most implementations so your asking a lot of a transport.
If you weren't so far away I would be happy to lend you a Fifo . Using USb-I2S without the fifo is not ideal especially straight to a DSP-less DAC.

I have never heard of straight 192khz i2S to a Multibit DAC without some sort of buffer. even the JunDac variaties have a "buffer".

What SPDIF receiver do you have on your DAC? can you input a 192khz SPDIF to the digital receiver then to the PCM board so you have a chance of a clean signal?

Just trying to help, I really doubt the issue is with the PCM daughter board, Ians digital work is has been outstanding.

The ability to handle 192khz is important in this project so as to offer oversampling via computer and pure NOS flexibility. But I think a Fifo is a prerequisit at these sampling rates, there is a lot of "cleanup" required with USB-I2S at high sampling rates.

Wish we had a basic PCM1704 PCB to work on this stuff, I would help but just can't hack up my expensive DAC's.
 
Member
Joined 2009
Paid Member
At the moment I can't afford to invest more on this DAC. When they arrive I'll try again with the cables Ian has sent me but if it doesn't help I'll do with the 96K limit, after all there isn't that much vailable in higher sampling rates.
Thanks a million to everyone for their help.
 
..many Audio-gd DAC's have easy input access to the PCM1704's by simply unplugging the DSP-1 module which needs no "hacking". Let me know if I can help.

Unfortunately my PCM1704 audiogd is an old model with pmd100, no easy way integrate to the NOS PCM board, I wish I could because it has the best analog section I have ever found for the PCM1704, several shunt regs, jfet gain stage, diamond buffer, heavy as heck.

The new model with the DSP-1 (SA-2) looks like it would be ideal for the fifo-pcm board, I wonder if ians spdif+fifo+PCM board would fit in the center section? The DSP-1 itself is quite a crude DSP, jitter/poor oversampling algorithm, but the DAC/analog/power section is top notch.


But if Ians digital boards wont fit, probably best for this Fifo-PCM is justbuild from scratch: draw up a little digital board for pcm1704's and use a Sen analog stage.

The ability for the pcm board to handle 176.4/196/352.8 khz is important so that one isn't trapped to a digital filter and can use the computer to do the oversampling.
 
Unfortunately my PCM1704 audiogd is an old model with pmd100, no easy way integrate to the NOS PCM board, I wish I could because it has the best analog section I have ever found for the PCM1704, several shunt regs, jfet gain stage, diamond buffer, heavy as heck.
Are PMD100 outputs reclocked before connecting to PCM1704?
- If no: go to the below suggestion
- If yes: probably doable with some modification - removing components/cutting traces - that you might not want to do...

Is PMD100 on socket?
- If yes: take it out gently and you are ready to plug-in NOS PCM board outputs via a pin-header(or similar)
- If no: consider adding one :)