Remove 10ns jitter from a 1PPS signal from a GPS receiver

Thread Starter

Adrienboub

Joined Jun 15, 2020
29
Hello everyone,

I'm trying to make a very accurate frequency counter. A GPS receiver allows me to have a 1PPS signal that I use as a reference to do my frequency measurement. According to the datasheet this 1PPS has a jitter of 10ns.

In short, the more stable this signal, the more accurate my frequency measurement will be, right?

So I would simply like to remove or reduce this jitter.

Internet friends, do you have any ideas?
 

Attachments

AnalogKid

Joined Aug 1, 2013
10,065
In short, the more stable this signal, the more accurate my frequency measurement will be, right?
No.

Since you have not given us the schematic, or even a block diagram of the circuit, everything is just a guess.

You don't say if your receiver is a disciplined oscillator. That matters.

IF IF IF you are using the 1 pps signal to gate a counter that is clocked by the signal being measured, then that circuit's best case ambiguity is +/- 1 count. For a 1 kHz signal, that is +/-1 ms. Having a gate width that is *only* 100,000 times more accurate should be good enough. At 1 MHz it still is 100 x better.

Also, know that the propagation delay through a string of counters is significantly greater than 10 ns, and varies with temperature. How this affects your circuit is a mystery.

AND - read up on how GPS timing works. That spec probably is for the receiver itself, not the total error from the satellite atomic clock through the transmitter and the atmosphere. The actual jitter in the receiver output signal will be greater than 10 ns, and it still will be a near-miraculous thing of beauty. The only way to have a timing signal with less jitter is to have a rubidium or cesium standard on your desk.

What is the signal whose frequency you are trying to count, and what is your required precision / accuracy / whatever?

ak
 
Last edited:

MrChips

Joined Oct 2, 2009
26,133
You need to state the accuracy you are trying to achieve.
There are two parameters to state, error and long term drift.
Also, what are the range of frequencies you wish to measure?
 

Thread Starter

Adrienboub

Joined Jun 15, 2020
29
Since you have not given us the schematic, or even a block diagram of the circuit, everything is just a guess.

You don't say if your receiver is a disciplined oscillator. That matters.
I gave the datasheet of the component that I use as a GPS receiver.

For the counting and frequency deduction stage I do this on an FPGA:
Only rather than using the external 10MHz oscillator as a source, I use my GPS receiver to make my measurement window.

Also, know that the propagation delay through a string of counters is significantly greater than 10 ns, and varies with temperature. How this affects your circuit is a mystery.

AND - read up on how GPS timing works. That spec probably is for the receiver itself, not the total error from the satellite atomic clock through the transmitter and the atmosphere. The actual jitter in the receiver output signal will be greater than 10 ns, and it still will be a near-miraculous thing of beauty. The only way to have a timing signal with less jitter is to have a rubidium or cesium standard on your desk.

What is the signal whose frequency you are trying to count, and what is your required precision / accuracy / whatever?
1644585132592.png
My objective is to measure at least with an accuracy of 0.1Hz a frequency of maximum 100MHz
 
Top