Try to find the algorithm that gives the best estimate of the externally measured PPS offset from the internally measured (serial port logged) thunderbolt data. Greetings, Chris

The learning power of these models would be very great, and they could deliver a real-time estimate of the other markets that would be hundreds of milliseconds better that neutrinos. The long term veracity of these computers could not be doubted, the noise in short terms would be continually reduced.

As far as I can tell, there is no way of even >> roughly estimating how tight they should be by saying "hand tight + x >> turns". > > Yes.. if you're talking about finger tight, there's a lot of > variability. With an inexpensive open end wrench, though you can get > pretty consistent.. perhaps it's the "Calibrated thumb" on the little > wrench?

My estimate cost from the Mouser catalog is less than $20 per unit.  For me, cheap insurance. If anyone wants a copy of my list contact me off list and I’ll email you a copy.   Regards,   Perrier

This allows both tight correlation among different instruments in the lab and also allows ADEV-like tools to estimate, and then gradually refine to high levels of precision, the actual data rate, during data collection.

Small factors don't change the practical result. 2mm/yr is a small and very conservative estimate. A better one that reflects current conditions is nearly 4mm/yr. But the worry is that the rate is not constant and there is some positive feedback. That seems to be the case but it will take 20 years to really know for sure. BACK ON TOPIC...

On 4/12/2016 3:16 PM, Michael Wouters wrote: > > evaluations (for accuracy) to finally get its best estimate of what one > second is. Each participating lab is then told what the difference between > its clock and UTC is. > > USNO participates in UTC and keeps within tens of ns of UTC.

The Microchip datasheets lack detailed performance specifications but my conservative estimate is that resolution down to 500ps over 200ns using the built in 10 bit ADC should be possible without complex calibration. (The PIC series ADCs with more than 10 bits are pretty horrible.)

. > To estimate clock frequency offset, well, just send a repetitive pattern. > Amplitude is easy as well. > My application is more like an HF interferometer/radio telescope, so I don't have any control over the transmitted signal.

In my estimation that seemed like the best return on effort. -- Chris Caudle