Adev and standard used in measurements of it.

[KE5FX]

Question...  What does TimeLab send to devices?  Does this mean counter setup is automatically done when I set up acquisition dialog?

Very little.  It relies on you to set up the measurement manually, then fetches individual readings.  You can see the different instructions sent to various counters including the 53131/53132 if you look at the makefile in the drivers\GPIB_TIC subdirectory:



This is why I usually suggest using talk-only mode when possible.  That leaves everything up to the operator to set up, resulting in less room for confusion.

[tkamiya]

I see.  I would imagine with number of counters in the market, it would be impossible to send commands for setup.....

One of the issues I experienced in the past with talk-only setting is, some counters can readings out of order.  TICC is a good example of this.  I know there are walk-around for this but unexpected outcome can happen.  I had a lengthy email conversation with TICC developers and apparently, it cannot be helped.

Right now, I have 17 serial ports setup on my PC, and all ports are taken.  It comes in very handy for use with TimeLab and LadyHeather. 

[tkamiya]

The measurement has few hours to go. 

Again, I have a stable 10MHz from Rubidium fed to both counters input and external reference.  Gate time 1s, full 12 digit mode.  Measurement is frequency.

As I watched the counter, it mostly stayed at 9.999..... occasionally showing 10.000...  I didn't think anything of it as last digit on any counter is always iffy by a count or two.  Then I looked at frequency graphs on TimeLab.  It's actually centered on -1.00E-11 with almost equal jump up and down by one count.  (-2.00E-11 to 0.00E-12)  Why is this?  Since the signal was tee'd off to both, it should center on exactly the 10MHz, doesn't it? 

Shouldn't reference = DUT = 10MHz?

[jpb]

The ADEV noise floor looks too low to me.
Given that the resolution of the 53132A is 150 psecs I'd expect a noise floor at 1 second of around 10^-10. My FCA3100 has a resolution of 50psecs and has a noise floor (back-to-back measurements) or around 5 x 10^-11.

Are you sure that the measurements are each separate (it is not acquiring a running average of some sort)?

I don't think that the 53132A can do back-to-back measurements, and I don't think it does time-stamping which would allow for some linear regression though neither of these is the issue here.

[tkamiya]

I do believe 53132A's internal interpolation mechanism does use some kind of running total or averaging.  I did not intentionally turn on anything.  It is at default state other than gate time.  Other than that, I'm not sure of anything.  In fact, how unsure I am is the topic of this thread.

Either that or I have a super deluxe version option 999 on my counter.

[tkamiya]

OK, 2 day run is complete!  Again, setup is HP53132A in frequency mode, both external reference port and port A is being fed by a stand-alone PRS-10 10MHz output.  Method of dividing signal is just a T connector.  Right at the output port of PRS10, there is a T connector, and identical 1m length of coax runs to the counter.  Both Ext Ref and Port A gets the signal.

[tkamiya]

With this done, I resumed measuring of my project.

I saw odd things.  I changed the counter input and started a new measurement.  I didn't change a thing on measurement setup screen.  Multiplication factor is 10e6.  Now, frequency difference screen tells me I'm measuring GHz!  Stop and view the data.  Sure enough, decimal place is in the wrong place.  Remove 10e6 and put 1.  It's correct on TimeLab. 

I guess HP53132A behaves in odd ways.