I have built a circuit using the Si5351A-B (MSOP-10, 3 outputs). I have I2C communications working fine with a Raspberry Pi 2. The crystal is 25 MHz and appears to be oscillating. If I program the fanout, and clock output source, I get a clean 25MHz square wave on each output. Next I tried to program an output of 2 MHz on CLK0, CLK1, and CLK2. I observe an output of roughly 2 MHz on the scope; however, there are sporadic periods of ~1ms where the output goes to 0 volts (this is observed on the scope and is difficult to determine what exactly happens) and the Loss-Of-Lock-A and Loss-Of-Lock-B sticky bits are always 1; this makes the scope output appear with drastic jitter. Also, when I change the Multisynth parameters to try and get 1 MHz or 4 MHz, the output is still 2MHz. I am looking for advice on how to narrow down the problem and achieve a stable output.
I don't know if any of the following questions are pertinent to the problem, but perhaps you could also address these concerns. I have read AN619 and I have generated a register set with ClockBuilder, but there are some discrepancies:
1. ClockBuilder generates a value for register #7, but there is no description in AN619. Is this register R/W and should it be written? and if so, at what point in the procedure?
2. Register #9 deals with OEB pin, but my part (MSOP-10) does not have an OEB pin. Can I ignore this register? i.e. does it need to be written?
3. ClockBuilder generates a value for Registers 10, 12, and 13, but there is no description in AN619. Same questions as 1) above.
4. I am not using Spread Spectrum and my part has no SSEN pin. Clockbuilder generates all zeros for register 149 thru 161. Do these registers need to be written?
5. My part does not have a CLKIN put. Clockbuilder generates all zeros for register 162 thru 164. Do these registers need to be written?
Thanks for any suggestions,
Did you try the Adafruit S5351A breakout board, as a low cost working example ?
That would help isolate data content/link issues, or PCB layout issues..