Sure, there is more than one way. It might a bit involved depending on how good you want the final clocking to be. In other words, IMHO doing it well would likely require designing a small board with an SMD clock module, a voltage regulator, and a clock buffer chip to drive the DSP chip clock inputs. If you are used to doing that kind of thing then its pretty straightforward. Otherwise, maybe not so simple.
Regarding taking the existing clock from one board and using it to drive the next board, its probably going to be more jittery that way. Whether or not you will notice or if it will bother you I don't know. Best to keep wires as short as possible in that case. 2-3cm would probably be considered pretty good.
Regarding taking the existing clock from one board and using it to drive the next board, its probably going to be more jittery that way. Whether or not you will notice or if it will bother you I don't know. Best to keep wires as short as possible in that case. 2-3cm would probably be considered pretty good.
Last edited:
Thanks for your reply!
I can solder smd stuff, no problem... but designing a PCB board is waaaaay above my knowledge.
Any suggestions where to find "pre built" pcb like the one I needed?
I have tried some clock board kits. Tried one like this: https://www.amb.org/audio/omicron1/ Didn't really like any of them well enough to keep. Ended up designing a clock system into a larger board that also does other functions. Anyway for your purposes, I would point out that the clock I linked to is unbuffered which IME is a problem if needing to drive two different boards. I don't know of a similar clock board with buffers. However IIRC an iancanada FIFO_Pi has a couple of buffered clock outputs, but its also a more complicated device than you are asking about now.
EDIT: just recalled that JL Sounds makes an oscillator board that has pads to install a buffer chip. However, I had to reverse engineer it to find out what the unused pads were for. They don't provide any documentation about how to modify it. Also, they use ferrites on the power rails for the clocks and buffer. Personally I would not use them for clocks unless there was a proven need and no other practical choice, but other people use them and seem to be happy.
EDIT: just recalled that JL Sounds makes an oscillator board that has pads to install a buffer chip. However, I had to reverse engineer it to find out what the unused pads were for. They don't provide any documentation about how to modify it. Also, they use ferrites on the power rails for the clocks and buffer. Personally I would not use them for clocks unless there was a proven need and no other practical choice, but other people use them and seem to be happy.
Last edited:
Thanks for the info!
Those boards seem to be more expensive that what I am willing to spend on this project.
Should I try to wire the 2nd DSP to the crystal on the 1st DSP? If it is too much of a long shot I would probably just sell the boards and call it a day.
ADAU1701 datasheet, page 12, says pin-31, the oscillator output, is not for driving another chip. If someone wanted to clock two of them together then an external clock module seems like a good way to go. Also, looks like they don't directly operate from the clock; instead the clock input acts as a reference for a PLL, which personally I would rather avoid. So if it were me I would probably search for a different device, one that is a better fit for my application. That above having been said, I haven't studied the datasheet enough to figure out whether or not two of the chips could be reliably used together. Could be there is a way.
Last edited: