Dear All,


I am having some problems In measurements relating to the TIME and frequency.
I have GPS freq Standard(910R) in my lab and i have a 5351B and also a HP53132A as working standards.
If i tie or lock all the time bases together, what would my uncertainty of my best measurment be?
Should i RSS the values of the 910R and the Systematic Uncertainty measurements as evident in the manuals of the HP5351B/53132A?
if yes, what are the paremeters that i should consider for the 910R?
Appreciate the help.

Rgds
Ruben

Certainly if you have a GPS-sync'd frequency standard, you should tie the other units to it. Then your 910R will drive the stability and uncertainty of the time base. You may find that you uncertainty gets down to around the -11 or -12 level as a result.

Hope this helps.

Dear Sir,

Thanks for the reply. But i would be obliged if you could elaborate with some facts and figures?
Or is there any leads or read outs over the internet about this?

Appreciate the help.

Rgds
Ruben

Dear Sir,

Say if we connect the time bases of both the gps standard and a counter, how do we tabulate the uncertainty ?
The BMC so to speak and the possible uncertainty budgets?

Rgds
Ruben

Generally, the GPS has a much lower MU than a counter, and so if the GPS time sync is used to drive the counter, then the MU is effectively the same as the GPS, with a little extra for the connection loss.

Hope this helps.

Dear Sir,

Are there any books specifically touching time and frequency calibration/measurement (if possible, power sensor and meter measurements)topics currently available in the market?
Besides Agilent's website?
Appreciate the help.
Rgds
Ruben

A good resource for the subject of time and frequency calibration is www.leapsecond.com, they have a link at http://www.leapsecond.com/pages/books/ with some good resources.

Dear All,


I am having some problems In measurements relating to the TIME and frequency.
I have GPS freq Standard(910R) in my lab and i have a 5351B and also a HP53132A as working standards.
If i tie or lock all the time bases together, what would my uncertainty of my best measurment be?
Should i RSS the values of the 910R and the Systematic Uncertainty measurements as evident in the manuals of the HP5351B/53132A?
if yes, what are the paremeters that i should consider for the 910R?
Appreciate the help.

Rgds
Ruben

Type A data:
-For your bmc's is going to be pretty much a repeatability study since any tech you have in your lab can slap a bnc cable into the 53132's input.
Normal Distribution, Divisor of 1.0.

Type B data:
-Good link here for the 53132's and other agilent counters;
http://www.home.agilent.com/upload/cmc_upload/All/measurement_error.htm
-Uncertainty of the 910R,
-Uncertainty recieved on last calibration of these units,
-Enviromental, (I dont see much from this number, unless you have a storm outside)
All these will probably be a rectangular distribution with a divisor of 1.732.

-Resolution
Most likely set to 1/2 your minimum resolution with a divisor of 1.732 for rectangular, others will say full resolution with a divisor of SQRT (6) for a triangular distribution.

Add everything together, use the Student-T for distribution at 2 sigma and call it a day.

With the 910R you will most likely come out with something along the lines of -11 to -12.

Hope I helped,
Mark.

Mr.Hershal ,

Hello again.
Greetings and i hope you are in the soundest state of health.
Pertaining this question of the Frequency calibration and the allan deviation...say when we use a counter( 53132A) to measure a 10MHz signal when we say tau(the term that we use in allan dev) = 1 second , does it mean that the gate time of the counter is also to be set at 1 second?
I have read and viewed through some samples involving the allan deviation computation over a period of tau = 1s and i get what they say when there should be no dead time in between measurements.
Does that mean we set the tau= gate time = 1second and take 20 measurements and compute the allan deviation?
That concludes the allan deviation over tau = 1 second and i somehow can derive this but what if we wish to get the allan deviation over a period of tau = 2 seconds?
We set the gate time to 2 seconds and take 20 readings/measurements..then compute the allan deviation? Is it as the same as averaging of 20 readings/measurements every 2 seconds?
Recently there was this PT test set by our national lab to compute the stability of a 10MHz oscillator but i pulled out of it cos my GPS unit was faulty and my universal counter was of the lowest option possible.
I have repaired my GPS and got it calibrated.
Can you suggest what i should look at in deciding the factors of the uncertainty budget of this PT test?
Appreciate a suggestion and i seek your humblest reply.

Rgds
Ruben

If the tau says 1 second, then my interpretation is that you are correct about gate time.

For most measurements on a 10 MHz signal, a 1 second gate is suffient; however, for the most reliable I would likely go to a 10 second gate time, and run 10 or 20 such samples, and average them. If the instructions state 1 second though, then follow the instructions given.

Watch your oscillator aging rate as this may throw you off. The aging rate of your oscillator should be at least one order of magnitude better than the UUT, and more is better there. Resolution error can also affect the uncertainty, in particular if the resolution goes to a lower number than your oscillator aging rate.

In other words, if the UUT has an oscillator with 5x10^-6, then your oscillator should have 1x10^-7, and the resolution should not be any more than 1x10^-8.

Hope this helps.

Mr.H,

Greetings and thank you for the reply.
Say we have 40 readings to start with for tau = 1 second.
We treat each point as a y1 to y 40 and then start the averaging and computation of the Allan deviation but if we take tau = 2 seconds, does it mean we take readings y1 and then y2..averaging it and then rename it as y1 and then proceed to do the same for adjacent readings( reading y3 and y4 averaging it and renaming it y2) to compute the allan deviation? Therefore reducing the 40 earlier mentioned readings or measurements by half..meaning we have 20 readings to compute the allan deviation for tau = 2 seconds?
And if the tau = 4, we take the 40 earlier readings / 4 = 10 readings?

Appreciate the help.
Rgds
Ruben

Mr.H,

Can i use a spectrum analyzer instead of a high end microwave counter to measure microwave and rf frequencies?
If i connect my GPS 10MHz signal to the 10MHz signal in (at the rear of the SA) , how accurate the frequencies be?
In my uncertainty budget, do i have to tabulate the allan deviation as well as what one can do using a counter?

Rgds
Ruben