Two things: You're assuming that more than one DAC can share the same I2C bus in the first place, and second you're assuming that the master system should be limited to a single I2C port.
Many modern DSP chips have 3 or 4 serial ports. With high sample rates, or with limited clock rates on the DAC chips, there simply isn't enough bandwidth for a single I2C channel to carry information for more than one DAC. The solution is to separate the data onto different serial ports, and then connect each DAC to a private line. I would think that 2 stereo streams at 24/192 would be pushing the limits of a single I2C interface anyway.
I'm not a fan of Sabre, but the two address limit may be perfectly reasonable if you look at the maximum clock rate. I have designed ADC products with 1 MHz sample rates, and with a 20 MHz maximum clock rate, there simply isn't room for more than 1 chip per serial bus. The typical DAC probably allows much more than 20 MHz, but with 24-bit words the maximum clock rate is used up quickly.
From the way you've written I'm getting the impression that you might be confusing I2C with I2S. The number of DACs that can be connected to I2C depends on the number of unique addresses that can be set up. I2C is being used for control, not audio data. Hence bandwidth is not the issue you seem to think it is - the bandwidth required for control is normally very low (potentially even zero) once the DAC chips have been initialized.
the sabre allows 2 addresses to be set within itself for dual mono and other cooperative tasks, so that each dac can be sent the same whole stereo i2s/spdif/dsd etc signal and only decode what it needs to for left or right channel, but that is as far as it goes, more than 2 dacs cannot work together like this for quad mono or whatever, without additional external hardware or software manipulation of the data stream.
yeah diy on the keyboard will be tricky, only started happening about a week ago, but its one of those super thin mac g5 keyboards, doesnt look like its going to be simple to pry apart. i've cleaned it, but sees to be an electromechanical issue underneath
yeah diy on the keyboard will be tricky, only started happening about a week ago, but its one of those super thin mac g5 keyboards, doesnt look like its going to be simple to pry apart. i've cleaned it, but sees to be an electromechanical issue underneath
Last edited:
Thanks for pointing this out. This thread seems to be growing fast, because I somehow missed the discussion of control, although that is very interesting.
You are correct, I was basically saying that lack of several unique addresses on the I2S port would not typically be something to complain about.
Qusp,
when you say more than two DACs, does that apply to the following situation or not?
In the hypothetical case of exa board, plus lets say three sabre boards for six channels, where each board is connected to one I2S output line (I am assuming there are 4 I2S channels from exa board) would you say that we have only one DAC or more, or in another word do you foresee the limitation you described or not?
Thanx
no, as long as there is not a need for each of your 3 dacs to have unique i2c address then there will not be an issue. in this case you could have each dac a unit unto itself without any connection between them at all, because titan will be splitting up the data stream into discrete channels, rather than sending the same data stream to all three dacs and asking them to decipher which data is meant for its own channels and which is for its 'partner' in that case they would use a hardware address to decide which data is meant for it.
so in short, you'll be fine, as titan will take control and send only what data is needed to each dac.
so in short, you'll be fine, as titan will take control and send only what data is needed to each dac.
Hi qusp,
Reading your posts I am getting confused and I don’t know any more if exaU2I can control any DACs
. I am not even sure if the name of my device is exaU2I or titan. Perhaps all this foggy discussion belongs to another thread code named titan. Perhaps the developers behind titan can organize themselves somewhere else and talk about the advantages of their products. Bringing the limitations and complexities of other designs is not productive here.
exa065
Last edited by a moderator:
Hi qusp,
Reading your posts I am getting confused and I don’t know any more if exaU2I can control any DACs
haha, i was under the impression that this was the point, that it didnt control them. they have their own addresses and thats fine, because they have their own i2s stream that doesnt need to be shared with anything else and the dacs need not talk to each other. it controls them in the way that it sends a clock with the data, but thats it, not an i2c connection? titan is not mcu for the sabres as well is it? if it is you have not communicated this to me. maybe you are confused
or maybe its me that is confused and there is a whole other layer of the design i have not been aware of. the i2c limitation has nothing to do with buffalo, or ackodac, it is a limitation of the chip itself, one that can be gotten around by creating some external DSP/mcu that coordinates the efforts of the dacs, or removes the need for the address in the first place.
regarding titan the name slipped out before by accident and then i probably changed back to some silly shortening of your name. my sincere apologies if you did not want the name out there yet.
hmmm well perhaps you can make this decision,
i'm happy to leave it for after there is something more to talk about and start a thread then, but i did not believe it was off topic. on the contrary, once explained it highlights the advantage of your design, as more than 2 sabre dacs can be used in concert in a crossover with multiple chips, which should be higher quality than 1 x 9018
i didnt intend it to go on as long as it did, but the info was first questioned and then i was asked for more detail, i did not think it was off topic/foggy, perhaps you could start a topic and outline the direction you would like it to go? i'll leave you to it, sorry for the disturbance
Last edited:
No silly shortening of my name can come even close to titan. I’ve never used any code names for exaU2I. I will be using version numbers. You must be mixing this thread and exaU2I with something else. What is titan?
Wow, 100 Hz? What, does that give you all of a 50 Hz bandwidth for your audio, or less? At 24-bit bit stereo, you'd have just over 1 Hz of audio bandwidth.
... just kidding, I imagine that you meant to type 100 MHz. What DAC are you using? More importantly, I suppose you haven't hooked it up to an exaU2I yet, have you?
yes, my m key doesnt work, nice catch, stopped working last week and i have to have it in my copy cache to paste into anywhere that needs one, a complete pita.
i'm using the akd12p teflon version sabre dac from ackolabs with opc's d1 inspired mosfet iv stage, nope i havent and wont be hooking it up to the exa unit any time soon. i have another one on its way shortly to test. one that was coincidentally developed at almost exactly the same time, with almost exactly the same specs, thus my confusion. i'm on the interest list for the exa unit though, and will probably try both at some point
i'm using the akd12p teflon version sabre dac from ackolabs with opc's d1 inspired mosfet iv stage, nope i havent and wont be hooking it up to the exa unit any time soon. i have another one on its way shortly to test. one that was coincidentally developed at almost exactly the same time, with almost exactly the same specs, thus my confusion. i'm on the interest list for the exa unit though, and will probably try both at some point
Last edited: