how to combine 2 DAC channels? (mono with DSD1792)

[[IDC]Dragon]

I'm building a 3 way digital crossover. The DAC part is planned to be handled by two DSD1792 chips. So one of the stereo chips is only used half.
The datasheet says the SNR improves a few dB when operating in mono, so I could do that, but I don't know to connect the outputs in this case. Simply tie the current sources together?

My I/V stage is the textbook opamp circuit.

Thanks,
Jörg

[Ken Newton]

The answer to your question is contained in Fig.36 of the DSD1792 datasheet, page 33. That solution is for balanced output.
http://www.ti.com/lit/ds/symlink/dsd1792.pdf

For a very nice little single-ended output op-amp based solution, I suggest the circuit of Fig.30 in the AD1853 DAC datasheet, page 16. Don't forget to adjust the vaue of the I/V feedback resistors to account for the 4x greater output current of the DSD1792 over the AD1853.
http://www.analog.com/static/importe...ets/AD1853.pdf

[[IDC]Dragon]

Single ended is required.
The TI circuit would need another differential amp to achieve that, versus the AD folks indeed connect the current outputs (I was rather joking when suggesting that).
I'm tempted to think a 3rd opamp in the chain is not gonna improve things?

Jörg

[SoNic_real_one]

If you use good OpAmps there is nothing to worry about. I would go on the route with 2 separated I/V converters and summing voltage circuits after that - just to keep with the TI schematic (fig 32). There are good reasons (related to internal design of the DAC) to separate the I/V stages and not to connect the outputs directly toghether.
Look on the schematic of Denon DVD-5910 for a proper stage for the DAC (PCM1792 is almost identical). You don't need to use exactly the same OpAmps... I would ditch the NJM2068 for LM4562. Even the OP275 I would try to replace it with LM4562 (slightly less slew rate but better noise and less distortion).

[Ken Newton]

Originally Posted by [IDC]Dragon

Single ended is required.
The TI circuit would need another differential amp to achieve that, versus the AD folks indeed connect the current outputs (I was rather joking when suggesting that).
I'm tempted to think a 3rd opamp in the chain is not gonna improve things?

Jörg

Also, take note that Analog Device's i/v circuit is referenced to the +2.78V AD1853 Vref, while I believe that for the DSD1792 the i/v circuit should be referenced to ground.

[[IDC]Dragon]

Originally Posted by Ken Newton

Also, take note that Analog Device's i/v circuit is referenced to the +2.78V AD1853 Vref, while I believe that for the DSD1792 the i/v circuit should be referenced to ground.

Yes, I've seen that.

Originally Posted by SoNic_real_one

If you use good OpAmps there is nothing to worry about. I would go on the route with 2 separated I/V converters and summing voltage circuits after that - just to keep with the TI schematic (fig 32). There are good reasons (related to internal design of the DAC) to separate the I/V stages and not to connect the outputs directly toghether.
Look on the schematic of Denon DVD-5910 for a proper stage for the DAC (PCM1792 is almost identical). You don't need to use exactly the same OpAmps... I would ditch the NJM2068 for LM4562. Even the OP275 I would try to replace it with LM4562 (slightly less slew rate but better noise and less distortion).

Thanks for the schematic excerpt. In front of the final adder is more filtering than I've planned for. The rest is similar. What's the sample rate there? I'll have 96 kHz, hopefully the filtering doesn't need to be so fancy.

In the one DAC I've modded so far (Musical Fidelity V-DAC) the results were good with LME49740 for I/V and OPA2134 for adder / line driver.