if you will only be using singular clock multiples and dont need switching between 2, then no you dont need the dual XO board. i'm not aware of any significant improvements in the clock buffer, or PSU on the dual board, so unless Ian wants to correct me there I say nope, you dont need it.
However these will not be around forever and I consider this dual XO board to be a very well designed piece of kit for a good price that works perfectly with the fifo buffer, I would be tempted to get it just for possible future needs. that being said you could wait for the Si570 multispeed DSPLL clock board, provided it goes ahead, it didnt get a lot of interest that thread, I hope that doesnt mean its not going to happen, sometimes you just gotta spoon feed them Ian lol, if you build it they will come
However these will not be around forever and I consider this dual XO board to be a very well designed piece of kit for a good price that works perfectly with the fifo buffer, I would be tempted to get it just for possible future needs. that being said you could wait for the Si570 multispeed DSPLL clock board, provided it goes ahead, it didnt get a lot of interest that thread, I hope that doesnt mean its not going to happen, sometimes you just gotta spoon feed them Ian lol, if you build it they will come
Ok, thanks for the info; since cash is low I'll go with the FIFO board this time and hopefully Ian forges ahead with his next gen project.
I was poking around he thread but didn't find anything related to this question - any idea what improvement in clock jitter (or what the maximum clock jitter is) with this FIFO board and the attached XO? I understand the Teradak is around 200ps, so do you think it's a magnitude better?
Thanks,
Gary
Sounds like the same as what I'm doing - so whether a person uses a Teradak USB to I2S, or the Twisted Pear USB to I2S, the FIFO board reclocks it. But that was my question - is the clock on the reclocker better? Less jitter? By how much? I guess that's the object of reclocking, but I'm just curious on the numbers.
better than reclocked, the MCLK (provided thats the signal thats important for your dac) is thrown away and a new one generated in its place. this is much better than reclocking with clock recovery
sop to further that and explain why there are no numbers on jitter performance, since the incoming clock has literally NOTHING to do with the output clock jitter and the fifo buffer by itself doesnt really improve jitter a great deal, the improvement to be expected is directly related to the quality of clock you install. that; as has been explained in the manual, is the very crux of it. yes there is a high quality power supply and a local, large fifo buffer, proper terminated i2s lines, controlled impedance layout, well designed clock buffer etc. so everything is in place to get the most out of whatever clock you use, even the stock one, but as the MCLK is output from an isolated 'island', where it is buffered and then fed directly to your dac without going anywhere near the fifo, the phase noise/jitter of the clock is essentially the jitter/performance of the i2s output. the other lines are output from the fifo board too, clocked out with very low jitter flipflops controlled by that same MCLK
I hope that makes sense
the clock on the board is for testing, you should replace it with a clock of your choice; Ian could not hope to satisfy everyone with his choice of high end clock, so he left it for the builder to decide.
so in answer to your question, there is no answer to your question, it depends completely on the clock you use, i expect the numbers on the stock clock are not that amazing, you are supposed to replace it, like amps and such you buy with socketed cheap opamps, but otherwise very good quality parts
this is all covered in the manual afaik, so RTFM =)
sop to further that and explain why there are no numbers on jitter performance, since the incoming clock has literally NOTHING to do with the output clock jitter and the fifo buffer by itself doesnt really improve jitter a great deal, the improvement to be expected is directly related to the quality of clock you install. that; as has been explained in the manual, is the very crux of it. yes there is a high quality power supply and a local, large fifo buffer, proper terminated i2s lines, controlled impedance layout, well designed clock buffer etc. so everything is in place to get the most out of whatever clock you use, even the stock one, but as the MCLK is output from an isolated 'island', where it is buffered and then fed directly to your dac without going anywhere near the fifo, the phase noise/jitter of the clock is essentially the jitter/performance of the i2s output. the other lines are output from the fifo board too, clocked out with very low jitter flipflops controlled by that same MCLK
I hope that makes sense
the clock on the board is for testing, you should replace it with a clock of your choice; Ian could not hope to satisfy everyone with his choice of high end clock, so he left it for the builder to decide.
so in answer to your question, there is no answer to your question, it depends completely on the clock you use, i expect the numbers on the stock clock are not that amazing, you are supposed to replace it, like amps and such you buy with socketed cheap opamps, but otherwise very good quality parts
this is all covered in the manual afaik, so RTFM =)
Last edited:
Settle this now and let us have it! Just keep whatever clocks, power, and XMOS/MCU open for us to superceed if we shoulg feel the urge.
But then again we could just wirewrap to our choice... There will always be the next generation (even if their tast of music is very strange).
Brgds
But then again we could just wirewrap to our choice... There will always be the next generation (even if their tast of music is very strange).
Brgds
Last edited:
Supports 44.1 KHz, 48 KHz, 88.2 KHz, 96 KHz using XOs with different frequencies, MCLK is fixed at 256*Fs. This is for the single clock board.
Dual clock is somewhat more flexible, but the clock has to be a multiple of 44.1 or 48kHz.
Albeit it is asynchronous, the clocks must be within reasonable tolerance from the input clock, so not to over- or underrun the FIFO. It's not an asynchronous sample rate converter!
So....no, 12MHz just doesn't do it.
Dual clock is somewhat more flexible, but the clock has to be a multiple of 44.1 or 48kHz.
Albeit it is asynchronous, the clocks must be within reasonable tolerance from the input clock, so not to over- or underrun the FIFO. It's not an asynchronous sample rate converter!
So....no, 12MHz just doesn't do it.
actually thats correct, from memory the clock that comes with the board is for 24x integers too, so if you are using CD material only, you will need to buy another clock anyway, because as I hinted, but zinsula stated, this is not a sample rate convertor, it will not do that, change the sample rate in the fifo and you change the speed of the audio. so while its asynchronous wrt the incoming clock tick, the speed is not altered, just realigned to a better clock
it makes sense that Ian left a couple of secrets for the dual XO upgrade board, but the standard single one will do just fine. didnt realize it was stuck on 256fs though, I read it but wasnt really paying much attention as i'm using the dual XO board, with a view to the Si570
so for 44.1 you'll have to get a suitable clock to get started
Hi,
Do you have the FIFO with the stock clock board? Can this style of XO be adapted to the PCB?
530PC11M2896DG Silicon Labs Standard Clock Oscillators
I know it's SMD, but other than going to Tent Labs which appears to have the module that will plug in, it's one of the only reasonable alternatives.
Thanks,
Gary
no
best to RTFM, thats a DSPLL, not a normal clock and its 6 pin, not 4 pin. its a relative of the one Ian is working on for the next clock board (Si570), but its nowhere near a drop in replacement as its a completely different device called a DSPLL type clock. just buy the recommended crystek, its only 10 more than the one you linked
heres the 45.1584 crystek and here is the 22.5792
you'll need an adapter, but perhaps Ian can help you out. if not someone should really do a batch PCB fun for the adapter to suit
Last edited:
Could be a bit of an issue for someone who wants to get better performance with just the FIFO board if there's no higher performance 11.2896MHz crystal that's suitable electrically or physically. Or will the 22.5792MHz work on the single FIFO board to get 44.1KHz I2S ?
Last edited:
both of them work for 44.1, even 90.14xx will work (notice all are multiples of 11.28xx). it depends on your dac. the fifo board will accept 128, 256 and 512x FS, but the single XO board will only output 256x FS unlike the more flexible dual.
but look, here I am doing your research for you, not even sure which dac you are planning to use it with and all of that was in the manual. it seems like we might have a problem here, in that you dont understand the manual, or how i2s works so reading it does not mean anything to you? thats cool, everyone has to learn sometime.
Ian and the thread can only do so much and I/we do want to help, but I know that I would rather help you help yourself; if you want to apply this in your dac, you should be making it your business to find out by understanding the (rather simple really if you take the time) way the sample rate, clock and i2s signals relate to each other and how that applies to your dac. otherwise IMO you have bitten off more than you can chew. its one thing to say I dont read the manual unless there is a problem (silly way to go about it anyway if you were serious) but if you then expect the thread to keep checking the details and regurgitating them for you..well... it gets old pretty quick
but look, here I am doing your research for you, not even sure which dac you are planning to use it with and all of that was in the manual. it seems like we might have a problem here, in that you dont understand the manual, or how i2s works so reading it does not mean anything to you? thats cool, everyone has to learn sometime.
Ian and the thread can only do so much and I/we do want to help, but I know that I would rather help you help yourself; if you want to apply this in your dac, you should be making it your business to find out by understanding the (rather simple really if you take the time) way the sample rate, clock and i2s signals relate to each other and how that applies to your dac. otherwise IMO you have bitten off more than you can chew. its one thing to say I dont read the manual unless there is a problem (silly way to go about it anyway if you were serious) but if you then expect the thread to keep checking the details and regurgitating them for you..well... it gets old pretty quick
Yeah....I read the manual. It's not clear, so that's the reason I'm asking. You don't have to respond....no problem.
Hi Roger57,
Sorry for the late reply. qusp, zinsula did perfect answers. Thank you so much guys!
If you just want to stick with only one Fs, says 44.1Khz, then you just need the FIFO KIT which attached with the single clock board originally. Maybe I could try to upgrade the software to make it support another default X Fs setting, which might be 512*fs, in this case, you get chance to use the 22.5792MHz CCHD957( the RMS period jitter is less than 2ps) on the single clock board. I know it's very hard to find a really nice 11.2896MHz oscillator.
Again, FIFO makes good clock working, as well as the bad. So, finding out a really nice clock is a even more important job than the FIFO itself.
Have a good night.
Last edited: