Oznog

I had some ideas about making a digital headset that would need to run stereo digital audio up to the headset, and also the headset mike would send digital audio back.

The problem here is that I'm limited in the types of jacks which would be desirable. Basically I'm thinking 3-pin, 1/4" phono. I wouldn't mind using 4-pin but I haven't seen them made in other than those 3.5mm minis used for hands-free cell kits and camcorder AVI cables. Too small. And DIN is undesirable because it's going to be connected and disconnected a lot and in general people won't want to screw with orienting that connector correctly to get it in.

Well, the headset would have some powerful digital circuitry, I'm looking at over 100mA. So two of those pins must be power and ground, leaving only one wire for bidirectional digital data. The current is far too high to have this be a parasite power device that rectifies power out of the digital signal. There is zero acceptability of putting a battery in the headset.

Is this even remotely possible? Serial comm doesn't normally go both ways. Even so, man 40Ksamples/sec, even 8 bit, is like 640kbaud for the upstream. It'd be much nicer to have 12 or 16 bit audio, too. The mike would be 8Ksamples/sec so the downstream is much less, but still a problem. That's kinda smokin and making it into a coordinated bidirectional scheme sounds a lot trickier. The Dallas 1-wire protocol is far, far too slow.

Any ideas? Or does this spec just rely on weapons-grade Unobtanium?

__________________
I thought what I'd do was I'd pretend I was one of those deaf-mutes.

phalanx

Why not CAT-5?

Certainly allows for high speeds, several pairs of wires for multiple channels, and you can send power over it.

Papabravo

There are numerous examples of half duplex communications, where the transmit and receive data lines are shared. The real problem with this is that the audio will be nearly continuous in both directions. I can see some possibility of a burst transfer up to the headset. While that was playing you could continuously sample and send the microphone data. It might not be unobtanium, but I'm guessing it won't be cheap, simple, and straightforward either.

Is this a one off project, or are you trying to productize this idea?

Oznog

How does CAT-5 become something you can transmit with power on 3 wires? If I had even 4 wires I would have gone CANBus or at least I2S.

This is an app where toughness is important. A std RJ45 is out of the question. Also, again, no connectors that require orientation like this.

__________________
I thought what I'd do was I'd pretend I was one of those deaf-mutes.

phalanx

Cat-5 has 8 conductors. At the data rate you will be pushing coupled with a pretty short cable run, there is no need to use differential signaling so you will have 5 conductors left over once your serial bus is implemented. They can carry more than enough current to cover even the most demanding MCU applications.

For connectors without orientation, you are pretty much SOL since you dismissed all the existing 3 and 4 circuit ones. Off hand I don't recall ever seeing one with more than 4 circuits.

RJ-45 is routinely used in Military applications and has never had a problem surviving shock tests in my experience. You just have to make sure you crimp the connectors properly. RJ-45 is plenty tough.