Hi Graham

Way back in the mists of time, one of the Si5351A Synth kit batches was polluted with 3rd overtone crystals instead of the 27MHz fundamental mode crystals. These oscillated at around 9MHz in the circuit and the U3S etc did work, but with 1/3 the output frequency. There were some people on the group who tried injecting non-25MHz or 27MHz frequencies into the Si5351A. The chip does work, but it appears that the spectral purity of the output is somewhat impaired. One member published a web page about using a 30MHz reference and showed some spectrum analyzer traces to demonstrate it.?

Incidentally, that forum thread you provided, wonders about why 25MHz OR 27MHz and nothing else... but there is an SiLabs forum post by an SiLabs engineer which confirms that the RANGE 25-27MHz is Ok.?

I also can confirm that it does work, injecting a reference signal into the XA pin. It is already done in the QRP Labs OCXO kit but at 27MHz.

Recently I had cause to be trying this out again. Attached shows the output of the Si5351A set to 14MHz, using the usual 27MHz crystal on one hand; and on the other hand an 8MHz crystal oscillator. The standard crystal version has a considerably cleaner spectrum. Whether that is because the frequency is out of spec, or whether it is due to amplitude levels, waveform shape, etc. remains to be seen. How important the degradation in spectral purity is, would depend on the application.?

73 Hans G0UPL

On Fri, Jan 19, 2018 at 4:36 PM, Graham <planophore@...> wrote:

Users of Hans' U3s, si5351 synthesizer etc may find this interesting, particularly those who might be interested in disciplining these devices to GPSDO or other high precision reference having only a 10MHz output by driving them directly at 10MHz



Of course, the current firmware will not work as it does currently BUT at least does give other options for driving these synthesizer chips.

I haven't tried this yet but it looks interesting and? is something I am planning on trying.

cheers, Graham ve3gtc

Graham (and Hans),

Based on Hans' emails regarding locking an Si5351A synth module to a 10 MHz external reference (Rubidium, in my case), I can confirm that I was successful, based on the Nov. 2017 ARRL Frequency Measuring Test results. I invested in a VFO/SigGen kit with relay-switched LPF kit and got it operational about 1 minute before the FMT started.? I set the reference frequency to 10 MHz (vs. 27 MHz) and was ready to go with a frequency difference measuring technique and Spectrum Lab software.? My first measurement was a bit over 1 Hz error; the remaining 3 measurements were less than 1 Hz error.? With a bit of practice, I hope to get all measurements < 0.1 Hz error.

My Rb standard power output is a little low- about 3-4 dBm into 50 Ohms.? The 10 MHz reference signal goes directly into the Si5351A pin 2 (crystal pad) and ground on the synth. module board.? I plan on adding a 10 MHz CMOS crystal oscillator to the VFO/SigGen kit soon with an external BNC-BNC jumper cable so the siggen doesn't require the Rb standard for basic operation.? The Rb oscillator is an LPRO-101.? The LPRO AC-coupled output RF is split using a Minicircuits LRPS-2-1 in-phase power splitter.? The LPRS DC resistance of an output measures close to zero Ohms.? Then the output is wired directly to the BNC output connector.? So not external DC biases are applied; similar to what a crystal would do.