How to switch multiple I2S signals to DAC ?

skibum

Member
2009-12-10 10:42 pm
I am modding/building my own DAC. I have an I2S signal coming out of a DIR9001 (SPDIF) interface and an I2S signal coming out of a CM108 (USB) interface.

I want the BEST method to switch/mux these 2 signals to my DAC chip. Is using something like an 74ACT157 the best way to do it?

I would think the 74ACT157 would add quit a bit of jitter to my I2S signals though.....

Any other recommendations?

Thanks,

Skibum
 
Why would the ACT157 add jitter to your I2S? Here I think the issues will be more to do with ground noise - the more grounds you connect together from different units, the worse its going to get. Have you tried isolators? ADI make some good ones which would work for your data signals (since they'll contain DC). For the clocks, you could try transformers.

But if you're not worried about noise, then yes, the ACT157 will do OK as a mux. I just recall that AC/ACT family is really rather bad at generating switching noise.
 

skibum

Member
2009-12-10 10:42 pm
Why would the ACT157 add jitter to your I2S? Here I think the issues will be more to do with ground noise - the more grounds you connect together from different units, the worse its going to get. Have you tried isolators? ADI make some good ones which would work for your data signals (since they'll contain DC). For the clocks, you could try transformers.

But if you're not worried about noise, then yes, the ACT157 will do OK as a mux. I just recall that AC/ACT family is really rather bad at generating switching noise.

Thanks abraxalito.

I am trying to find out what device would be best suited for switching between two I2S signals. I have been looking at some high speed switches with low ON resistance & capacitance like the Maxim 4715.

I am not sure if switching noise (or switch time) is relevant for my use. The switch will only function when I change from a USB input to a SPDIF input. They key parameters I am interested in is a device that does minimal harm to my I2S signals. I am looking at minimal ON resistance, capacitance and inductance.

Any suggestions would be appreciated!

Thanks again,

Skibum
 

skibum

Member
2009-12-10 10:42 pm
You don't need high speed switching capability, just a quiet and stable multiple pole two throw switch. There are zillions.


The I2S signal consists of a low voltage (3.3v or less) 24 MHz signal. At this frequency the switch on resistance, capacitance and inductance would matter. I agree with you Bill, but I am just trying to find what is the best switch/mux out of all the possibilities out there.

I would imagine all the DAC's out there must be switching the I2S signals in some similar manner. I just don't have any around to open up and take a look at. The DAC I am working on uses a 20+ year old mux to do the job - there must be something better now days.

Thanks
 
I am trying to find out what device would be best suited for switching between two I2S signals. I have been looking at some high speed switches with low ON resistance & capacitance like the Maxim 4715.

I had a quick look at the datasheet - I'd be a bit worried about the capacitance there. Low RDS(on) switches need big transistors, hence high capacitance. From the Maxim website, the 4717 looks to be lower capacitance (15pF) and therefore higher bandwidth - of course it comes with slightly higher on resistance but 4R5 can't be a problem I think. If you check out the bandwidth of the 4715, it looks to be rolling off above 20MHz.

The available packages though do not look very diy friendly.:(

I am not sure if switching noise (or switch time) is relevant for my use.

Right, its not relevant if you use CMOS switches - I was referring to the noise generated on the power supplies by AC/ACT families switching their outputs. So a CMOS switch is definitely a lower noise approach than an ACT mux.
 

wakibaki

Banned
2008-01-08 11:51 pm
3GHz SPDT PCB relay?

Panasonic | Relays, Switches and Indicators | Relays | High Frequency & RF Relays | Panasonic |ARS1512

'Course for ideal performance you'd want to feed it with a 50 or 75 ohm microstrip line, but I don't imagine you're matching the traces to the chips anyway...

I would imagine a 74HC mux will be fine. 74HC151? Good for ~50MHz

Multiplexers / Demultiplexers | Farnell United Kingdom

If you're determined to have a fast device, have a look at CPLDs or FPGAs, speeds of 400MHz are not uncommon. CPLDs are non-volatile. It's a bit of a waste in many ways, but you could maybe soak up some other glue logic into it. You'd have to buy a programmer, I think the Lattice ones are the cheapest entry, but it's a while since I looked.

24 MHz seems a bit high. 192/24 stereo is still <10MHz.

w
 
Why not use a tri-state bus drivers? One such for each set of input signals. Outputs from all are paralleled together, with only one of them enabled (the others tri-stated). These are digital signals at this point (S/PDIF and USB are however fundamentally analog), so don't treat them as analog.
e.g. 74xx244. You choose the logic family based on what you'd like to see for rise/fall times, drive capability, etc. etc.
 
Last edited: