for Alan de G1FXB
If you are going to quote my text, please don't edit it to make me sound like an idiot.
What is the purpose for all the "?????" that you added?
"I stand by a belief a OCXO reference is more preferable than applying correction to a less stable source.
If a normal SI5351 module is used then best case construction applies, give the controller the easiest possible job.
With thermal masses (others use the term heatsinks ;-)) & avoiding draughts....
With GPS correction there needs to be a trade off between the choice of continuous discipline (reloading Reg2 every sec) or allowing an determinable amount of drift but maintaining a stable 50/50 duty cycle in the mean time. Ideally you don't want to have to recalculate Register 2, ever........"
I agree that stabilizing the xtal temp would allow for less frequent corrections. But that is not the issue, the issue to me is that the controller is not making the necessary corrections. Please look at my Photo 2. I can open the loop by removing the 1PPS and insert my self at the keyboard to close the loop by loading Reg 2 with whatever value is required to bring the CLK0 freq back to 10 MHz to within about 0.37 Hz. (the Reg 2 resolution is 1 Hz at 27 MHz which is 10 / 27 = 0.37 Hz at 10 MHz).
But I don't have the patience to have to do this every few minutes, or hours, or days depending on how well I am controlling the xtal temp (or using an OCXO or whatever). I want to let the controller use the 1PPS to do this for me...it's called disciplining an oscillator.
"Maybe phase disturbance caused in the discipline process has more than an influence in the counter readings than it first appears / more so if clashing with the gate period?
(It's something that often gets omitted in many discussions.... )"
Yes, thank you, I am aware that when Reg 2 changes (either from the ATtiny84 or from the keyboard) that the current 10-sec count is to be disregarded. Further, the visual info given by the technique of Photo 3 is as good or perhaps better than the counter value.
for geoff M0ORE
"I think that it should be borne in mind that some builders are attempting to achieve stability from a unit costing a few dollars, in some cases the cost of a mug of coffee, that professionals spend thousands of dollars to achieve."
This builder was/is expecting to "...achieve stability from a unit costing a few dollars..." that was given on the web site.
"The crystals supplied are standard computer grade units which are not intended to be used as a frequency standard, just to give timing signals for a micro-processor."
Nowhere in the ProgRock description does is say, nor do I expect, that the 27 MHz xtal is a "frequency standard". The Si5351 does not measure the xtal freq. It wants only two numbers (aside from some house-keeping stuff) from the controller... the desired CLK0 freq and the measured (by using the 1PPS) xtal freq.
The "... just to give timing signals for a micro-processor." is a non sequitur, the only microprocessor involved uses its internal RC oscillator.
"If you want to measure the stability of anything, you need to have the test equipment better than the unit you are testing to get any meaningful results. In the case of frequency, what standard are you going to use??"
I have twice given this information, but here comes number three...
"...my frequency standard is a Trimble ICM-SMT GPSDO and my ultimate 'sanity check' is the 10 MHz xmtr at WWV in Colorado."