Sharing the MClk across several boards
Thread Summary
The user inquires about sharing a 12 MHz MClk across multiple devices on different boards. The final answer recommends using a clock oscillator module and a clock buffer chip to ensure signal integrity, especially for longer runs. The ADAU1701 lacks a CLKOUT pin, complicating direct crystal usage. For distances over 4-5 inches, coaxial cables and clock buffer/drivers are advised to maintain signal integrity.AI Generated Content
I'm thinking worst case here but has anyone succeeded in sharing the MClk (from a 12.MHz clock module) across multiple devices on different boards? Surely not so much of an issue on a single board but what happens if we're going through a pogo pin or some sort of PCB interconnect pin.
This sounds like a recipe for a reflection party.
Or has anyone managed to do this?
Audio Signal Processors
hardware
SigmaDSP Audio Processors
pcb layout
adau1701
Hello Marcdraco,
Well,, it has been done successfully and not successfully! One little problem with the ADAU1701 is that it does not have a CLKOUT pin! So if you are using an actual crystal then there are some difficult issues to ensure the crystal will oscillate and not be loaded down. The OSCO pin is part of the feedback loop for the crystal so you cannot load it down. I have seen some people try to use a FET input OpAmp to buffer the signal but I really would not recommend that.
What I would recommend is to use a Clock Oscillator module not just a pure crystal. So then you can take the output of that module and send it to a clock buffer chip. Those chips will have several buffered and isolated outputs and have more drive capabilities. So you feed one output to the 1701 and one to go out of the board to other parts. Since they have better drive you could get away with longer runs. But, you may need to have a series damping resistor and I usually do not like any caps but it depends on things like reflections etc. It can get messy. If you run it to another board I like to use a 2-pin piece of ribbon cable so I can run the clock and a ground right next to each other to keep the signal integrity better and give a return path for parasitic currents.
But, I would NOT try to go very far. A few inches. Longer that 4 or 5 inches will start to get iffy. I have done as long as 12 inches in the lab and was successful. In the lab with the Audio Precision we have cables that are around 3 ft long but they are coaxial with a defined impedance and the AP is designed to drive them!
So if you have a long way to go then go to coax and for sure use good clock buffer/drivers.
Dave T
Please log in to reply.