[kipman725]

I was thinking about a DAC with inbuilt headphone amp and would like to totatly isolate it from the computers power supply. As I understand it I2S is uni directional so I could use opto isolators on it? however the wikipedia article says that the word clock will be 64* the sample rate giving a 2.8MHz word clock this appears to be too high for opto isolators (most 1MHz max). Has anyone got any ideas of a solution to this problem?

The only one I can think of so far is turning the I2S into a parallel interface and opto isolating that and reconverting to I2S but that would require 2 fast micro controllers or FPGA's and lots of opto isolators.

 

[kipman725]

Actually I just found this Product Folder - 6N137 - 8-Pin DIP 10 Mbit/s Single-Channel High Speed Logic Gate Output Optocoupler 10MHz although it appears to be 5V logic so I will need some logic level translation on the input (also maximum input current is pretty harsh).

 

[Clutz]

fwiw, Jocko Homo would argue against it because of the resulting increase in jitter. I don't know enough about anything to say anything more.

 

[error401]

1:1 pulse transformers should be good up to a couple 10s of MHz. They're used heavily in Ethernet where they run at 10MHz.

 

[borisov57]

I have used 6N137 for that purpose years ago and it worked fine, although I do not remember if sound improved using them.

 

[kipman725]

Quote:

Originally Posted by error401 /img/forum/go_quote.gif 1:1 pulse transformers should be good up to a couple 10s of MHz. They're used heavily in Ethernet where they run at 10MHz.

would the propagation delay be more consistent using pulse transformers? hmm the DAC will have jitter reduction circuitry? if not I could design my own although It would mean learning about PLL's etc as I would need a clock at 90degrees phase to the wordclock.

 

[error401]

Quote:

Originally Posted by kipman725 /img/forum/go_quote.gif would the propagation delay be more consistent using pulse transformers? hmm the DAC will have jitter reduction circuitry? if not I could design my own although It would mean learning about PLL's etc as I would need a clock at 90degrees phase to the wordclock.

I don't understand the physical constraints involved, so I really can't tell you what the prop. delay of a transformer would be, or the mechanism that governs it.

They're used in 'proper' S/PDIF receiver circuits, and many claim they improve the jitter of S/PDIF transmission (though I'm not sure what mechanism would allow an improvement). You'd need to hook it up and measure it, but I suspect it would work well.

 

[kipman725]

well the actual propagation delay is not the problem its the propagation delay variance between devices and if some kind of correction circuit was used the drift in propagation delay would be the parameter of concern. I will have to look up how to use these pulse transformers but they may end up easier to deal with than that particular opto isolator due to the weird voltages that particular IC needs.

 

[kipman725]

I did some playing round with pulse transformers from an old pc power supply and they look like they could be an absolute nightmare, the drive requirements are much greater than TTL logic can provide and the signal at the other end of the transformer was very noisy. Although with my limited equipment (10MHz Scope) I couldn't observe any propagation delay. I have however found a hybrid device that looks like it will do the trick:
Analog Devices: Replacing optocouplers with digital isolators (or how a midnight ride could have been avoided) :: Digital Isolators :: Interface