Author Topic: Software to analyze frequencies (Read 2248 times)

Picuino · « **on:** April 10, 2020, 11:22:12 am »

Hi,
I would like to know software programs that I can use to analyze frequencies.
And standars for communicate frequency meters with computer and this kind of software.

Regards.

Theboel · « **Reply #1 on:** April 10, 2020, 11:41:08 am »

try "Timelab", the author also very active and respectable Gentleman member of this forum.
http://www.ke5fx.com/timelab/readme.htm

or You can also use
"stable32"
https://ieee-uffc.org/frequency-control/frequency-control-software/

Picuino · « **Reply #2 on:** April 10, 2020, 12:28:11 pm »

Thank you very much.

Here, the manuals:
The PhaseStation 53100A User's Manual serves as a general-purpose guide to TimeLab.
Tutorial: Learn Frequency Stability Analysis Using Stable32 by William Riley

Regards.

Picuino · « **Reply #3 on:** April 10, 2020, 12:36:51 pm »

I can't found the ASCII or binary protocol to send data from frequency meter to this analysis programs.
¿Are there any standards in frequency counters via serial COM ports?

Regards.

Dr. Frank · « **Reply #4 on:** April 10, 2020, 01:18:02 pm »

What do you mean?
There is probably no such standardized protocol.
There are counters which can be set to TALK ONLY, and then dump ASCII format to the GPIB or RS232 bus, every time they made a measurement.
TimeLab obviously can fetch this ASCII transfer w/o any handshaking.

You need to know, whether your counter supports this mode.
As far as I know, old HP counters do this. (e.g. 5345A, 5335A, 5370B)

Frank

Picuino · « **Reply #5 on:** April 10, 2020, 03:08:35 pm »

Well, my purpose is very simple.
I want to emulate some of those protocols with my home made frequency counter in order to use Analysis software with it.

JohnPi · « **Reply #6 on:** April 10, 2020, 03:31:00 pm »

Look at the SCPI features for the HP/Agilent 53131A models - https://www.keysight.com/main/editorial.jspx?ckey=1000000328:epsg:man&id=1000000328:epsg:man&nid=-34972.0.00&lc=eng&cc=US

Picuino · « **Reply #7 on:** April 10, 2020, 04:03:00 pm »

It seems that the SCPI standard is a bit complex to implement.
https://en.wikipedia.org/wiki/Standard_Commands_for_Programmable_Instruments

I would prefer a old standard in which the frequency meter sends the readings without handshaking.

Does anyone have a captured RS232 reading of a frequency counter?

notfaded1 · « **Reply #8 on:** April 10, 2020, 06:25:46 pm »

It's just a text file with a list of single readings of the frequency one per line in as many digits as you can provide... that's it. Timelab or stable32 can read a text file like this. Not too complicated.

Dr. Frank · « **Reply #9 on:** April 10, 2020, 11:59:42 pm »

Sorry, I only have counters with GPIB interface, 5335A and 5370B. The data output formats, always ending with <cr> <lf>, is described in their manuals.
If you have a 34401A DMM, it has a TALK ONLY mode over its RS232 mode also (set address 31), intended for printer dump, but this can also be sent to a PC with serial interface and PROCOMM -alike monitor program. And the 34401A also measures frequency. That would be the same thing as with the 53131/132 universal counters, which also had an RS232. Modern counters probably lack that old serial interface, as they have USB instead.

Again, there is no real standard for the output string.

Frank

PS: maybe the screen dump from TALK ONLY over GPIB is what you are looking for.

Picuino · « **Reply #10 on:** April 11, 2020, 06:53:03 am »

Thanks, that's what I was looking for:

First: FRQ= (maybe a space or a tab)
Second: 1.00000563E+7 (frequency in scientific format)

Now my frequency meter sends readings in engineering format ending with Hz string (10.0000563 MHz)
I will need to reprogram the output format.

Thank you.

Picuino · « **Reply #11 on:** April 11, 2020, 08:17:27 am »

Now I have another question about the frequency readings.
My frequency meter is of the continuous counting type, so simple frequency measurements lose some information.

I can also send the number of counted pulses and the exact time of the sample interval. This is much more accurate than the frequency in the long run.

Is there a way to take advantage of continuous counting mode with analysis software?

Dr. Frank · « **Reply #12 on:** April 12, 2020, 12:54:05 pm »

That depends on your method used.
If your counter is of the time stamping method, that makes sure that you get seamless samples, i.e. without a dead time in between That's good, because the ADEV will receive the correct interval data.
If you added some averaging methods, like the 53131/132, then the Allan Deviation will not calculate the correct stability figures.

The best method is a Time Interval measurement, e.g. on two 1 pps signals. I don't know if your setup can handle this.

Frank

Picuino · « **Reply #13 on:** April 12, 2020, 01:43:46 pm »

Yes, my homemade counter is of the time stamping method.

What I don't know is how to display the time readings. Currently the readings are displayed as an integer number of clock cycles between samples. For example one second between samples is shown as 8000000 tick counts.
Do I have to represent it in seconds?
Interval seconds between samples or accumulated seconds since first sample?

notfaded1 · « **Reply #14 on:** April 12, 2020, 09:29:18 pm »

Time lab can handle either the frequency in Hz or the time interval in fractions of seconds between counts. It'll work with data in decimal or in scientific notation.

Bill

Picuino · « **Reply #15 on:** April 15, 2020, 09:36:23 pm »

I have made a test with 2 sources of 1PPM.
The frequency meter gets timestamps every 2 input pulses (every 2 seconds)

But I don't know how to enter the data in TimeLab.

Attached raw data list.
Time A comes from 1PPS signal of a OCXO
Time B comes from 1PPS signal of a GPS

Picuino · « **Reply #16 on:** April 15, 2020, 09:41:22 pm »

I tried with TimeLab: File... Import ASCII phase or frequency data...

but it doesn't work. The error is: "Insufficient data to render the selected deviation trace"

Dr. Frank · « **Reply #17 on:** April 15, 2020, 10:04:24 pm »

Sorry, I'm a bit tired of C...a hospitalization.. so I won't analyze your whole problem.
Please investigate by yourself what the different counters spill out as ASCII messages.

At first sight it's clear, that TimeLab (or any other English) program will NOT accept comma instead of a DOT.
Then you have to strictly compare, which direct output fromat from an instrument worked already.
I've given an example already.

Picuino · « **Reply #18 on:** April 15, 2020, 10:22:43 pm »

Thank you. Now it works.

Edit: The graph is wrong!! See the correct one 2 post later.

Dr. Frank · « **Reply #19 on:** April 15, 2020, 10:58:31 pm »

Really looks great, as intended. Thank you!

I just finished the long-awaited full calibration of my 5370B.
Jitter went down from 35ps to 11ps, and I now finally, after many many years begin to understand the culprits of this counter.

I probably will take the equivalent Adev diagram the next days.

Picuino · « **Reply #20 on:** April 16, 2020, 06:46:09 am »

The graph is wrong.
My DIY frequency meter has only 7 decimals of precision (resolution of 200ns) in this measure (I hope to get 10ns resolution later).
And the measured time is about 6600 seconds, not days.

The graph is shifted towards more measured time and resolution than in reality.

Picuino · « **Reply #21 on:** April 16, 2020, 09:53:17 am »

Ok. After play with TimeLab options I have got the correct graph.
Much more modest than the previous one, but I am satisfied with my $3 frequency meter.


EEVblog Main Site	EEVblog on Youtube	EEVblog on Twitter	EEVblog on Facebook	EEVblog on Odysee

Author Topic: Software to analyze frequencies (Read 2248 times)

Share me