I bought the tinyPFA to experiment with the stability of the Bodnar and several GPSDO's I build the last years.

I started with the Bodnar and wanted to measure the noise floor of the tinyPFA. I did this using the method given by Erik Kaashoek. Look for his presentation here:  a good reference to both inputs of the PFA. I took the Bodnar mini as this reference to both inputs.

The result surprised me !!  sigma(1) = 2.5E-13

                                               sigma(10) = 2.5E-14

                                               sigma (100) = 9.8E-15

 

 

This makes very accurate measurements possible in terms of phase, frequency and specifically differences between two sources.

I was curious what the plot would be when I connected my GPSDO-1 (35 uHz DAC-resolution)  to both inputs. This is the result:

 

In the trace field it says Bodnar, but it is GPSDO-1. It is more or less comparable to the plot of the Bodnar to both inputs of the PFA.

 

In the doc of Bodnar I found this plot. The miniref is compared with a rubidium and a Tico2.

Bodnar claims Sigma(1)  = 5*10e-11

                          Sigma(10) = 2*10e-11

These measurements show that the Bodnar should be able to act as a "reference signal" for my GPSDO's.

 

So, now we put the Bodnar in channel 1, and the GPSDO-1 in the second channel.

We measure with tau = 0.1 sec. The GPSDO-1 is running for over a month now, so it should be quite stable.

 

The GPSDO-1 is basically a build of what Lars and Paul described in their webpages.

This is what I measured against the Bodnar miniref:

 

 

My first GPSDO has a resolution of 35 uHz per DAC-bit. In a "LOCK" situation of the GPSDO the lsb of the DAC is going up and down very slowly, about two or three steps per 10 seconds. So I expect a ADEV of about 100uHz. This is basically what I measure also. 10E-11 (at 10MHz) is 100uHz. The Bodnar is ok to measure this value, because it is doing 10E-12 itself.

Conclusion is that this GPSDO is working properly and very useful to i.e. stabilizing my Oscilloscope and my Spectrum Analyzer.

 

But, because the resolution can be increased fairly easy, and the temperature control can be increased a lot!, we should be able to do better.

 

So I build a second GPSDO, with the same Bliley VCO. Using better temp controllers (TMP117), a 16 bits DAC, and a better ADC.

look here for details.

 

I expect a better performance than my first GPSDO, but measuring is knowing !! The question is whether the Bodnar is good enough to get a good image from the GPSDO-2. Let's put the Bodnar on channel 1 of the PFA and the GPSDO on channel 2.

Let it run for a few hours to settle down everything, and to get used to the temperature of my measuring room (17 degrees Celsius).

I want to run the test with Timelab for an hour and look what happens during the test.

Chosen configuration:

for the GPSDO:

TC = 275

Gain = 250

Damping = 5(00)

The resolution of the DAC is 4,4 uHz per DAC lsb (so, almost 10x the resolution of the first build GPSDO).

for the PFA:

Sample interval = 1, 2, 3, 5, 7, 10 sec, just to check out what happens

Duration = hour per measurement

Output = unwrapped phase to usb