ADC to IP and IP to DAC

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
For start I wanted to say a hi to all. :)


As title says

I want a SINGLE ANALOGUE input which can do up to 100KHz FLAT.
Then I want this analogue signal to put it into a ADC.
Than, I want to take it out from IP (NET).

On the other side of the NET (IP) there will be a IP to DAC where I could take out this signal as it was (up to 100KHz flat) BUT without much latency.

Simlified, I want a single analogue signal from 0Hz up to ~100KHz to pass it through a NETWORK.

Is it posible? :)

Thanks in advance,
Last edited:
I want a SINGLE ANALOGUE input which can do up to 100KHz FLAT.

Why the 100 K requirement?

BUT without much latency.

What do you consider "much"?

Is it posible?

Maybe. Depends on your latency criteria. An ADC that can do above 96K is not going to be cheap, and you do need to buffer the IP packets to some degree, and that will inevitably bring latency.

If you only want to transmit audio and not some strange high-frequency information for a non-audio application, you could use something like the Barix Exastreamer.
Can't an ADC or a DAC do ONE Input ~65KHz?

It's about some musicians.
They need a bandwidth 0 Hz to 65KHz flat for passing:
a. their musical instumens (0 - 20KHz - yes they want 20KHz - I can't even hear this frequency :confused: )
b. they want "flatness".
c. they have an equipment that has an oscilator ~45KHz. They give to it 0 - 20KHz and they get that 45KHz + the mix from 0 to 20KHz, thus ~65KHz, and their max frequency is 45+20 = 65KHz. (Though I can't imagine where they use a system like that) :confused:
Aren't all or if not all, the majority of the audio cards, 192KHz sampled nowadays?

Pretty common on the DAC side, less common on the ADC side.

How high latency is acceptable in audio? I don't know. :confused:
I don't think that they want "lost packets" so a bit latency I believe it'll not harm them ehhh?.... :D

Problem is that going from 48 KHz to 192 KHz sample rate also quadruples the latency.

If lost packets (resulting in short dropouts) are unacceptable, you will have to buffer enough to allow for retransmissions, so latency could go up into several seconds.
then you cannot have guaranteed buffered data and since you dont have the ability to use the dac as the clock source and send that back, then you have problems. maybe several seconds is overstating it, I dont think you need that much, but certainly 20-100ms is not enough
I think the OP needs to go back and tell his musician friends that unless they can find a way to avoid causality and the Nyquist sampling theorem (and perhaps a few more things set for us by the universe as it is) they can't have what they want. It is almost certainly the case that what they want is not actually what they need so maybe if they can spell out exactly what they need, and where they can compromise, an engineering solution may be possible.
A dedicated Ethernet link could perhaps do it, but you would need to bypass IP and go straight for the physical layer. Not sure if the usual drivers support that, but I suppose you could get a modified driver.

If I recall correctly (I can't be bothered to look it up) a 'thick' Ethernet cable can be up to 500m, and you can have up to three of these connected via bridges? So 1.5km is possible. Remote bridges linked by fibre can extend this. Many years ago I used to play with things like this for the power industry.

Knowledge of the application might help.
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.