[time-nuts] FA2 notes

Sat Oct 5 21:49:33 UTC 2019

Hi Jerry,

Thanks for that update on the FA-2. I had the opportunity to test one 
last week. Comments:

 > the unit isn’t competitive with the TICC2 and TADD combo

They are quite different instruments: one is a timestamping or time 
interval counter, and the other is a frequency counter. Using a 
frequency counter as a LF time interval counter or using a time interval 
counter as a RF frequency counter will not often yield best results. 
Note that both instruments can achieve on the order of 11 digits/second 
of resolution in ideal cases.

One significant difference is that the FA-2 has a large dead time, the 
tau is not actually 0.1 or 1 or 10 s, and it reports only frequency and 
not phase. This may have implications for using it for long-term 
monitoring. I'm doing some tests to quantify this effect. It wouldn't be 
the first time a BG7TBL product struggled with the concept of phase vs. 
frequency measurement.

 > the returned values are (around 1.000 000 000) 1.000 000 005 and 
0.999 999 995

Correct. The FA-2 does not do well with LF signals; for that its best to 
use a time interval counter, such as the TAPR/TICC (which has ~0.1 ns 
single-shot resolution). The FA-2 manual implies the core counter has 
only 10 ns resolution so for slow signals (such as 1PPS) you're limited 
to 8 digits/second resolution. The PLL-based (?) enhanced resolution 
kicks in above 100 kHz.

 > never seen anything come back from it other than 10:9 or 10.9 format

Right. Counter resolution is often quoted as "digits per second". This 
counts *all* significant digits, not just *decimal places* (digits to 
the right of the decimal point). For example, a reading of 9 999 999.976 
has ~10 digits of resolution. And a reading like 10 000 000.000 32 has 
~12 digits of resolution. I can explain more if you want and how this 
ties to tau and ADEV, etc.

 > the 1PPS out of an HP 5061B Cs

The 5061 is known for an unusually powerful 1PPS output. Beware. It can 
fry the front-end of some counters.

 > someone should start a group to move this off list

Probably not. Discussion of frequency counters is about as on-topic as 
one can get here on this list. If we started a new group every time a 
shiny new GPSDO or counter showed up on eBay the list would have a 
hundred subs by now. What is helpful are postings like yours that 
combine multiple issues and discuss them in depth. By contrast, short 
stream of consciousness postings tend to fill up the list and push busy 
people away.

/tvb

On 10/4/2019 10:38 AM, Jerry Hancock wrote:
> I noticed you can change the baud rate to for instance, 115200, but I can’t get it to stay persistent after power off.  You send it a command of $B11520* to change it to 115,200, but since it only sends like once every 300msecs, it doesn’t really matter the baud.
>
> I found the unit isn’t competitive with the TICC2 and TADD combo.  When measuring a 1PPS signal, the returned values are (around 1.000 000 000) 1.000 000 005 and 0.999 999 995.  Note the ‘5’ as the last digit.  When measuring signals greater than a certain frequency that escapes me, the last digit will then add more resolution switching from units of 5 to 1.
>
> Also, in the doc they talk about digits returned from 8 - 12.  I’ve never seen anything come back from it other than 10:9 or 10.9 format depending on your coding skills.  I tried putting it into ‘High precision” mode by sending it a $E3434* command.
>
> You can change the gate on the fly using the $G* command.  It cycles through the gate times opposed to being able to set it directly.  There are a number of persistent commands that enable default startups, etc.
>
> And one other issue, though it has a somewhat high input level, I noticed I had to put an attenuator on it for measuring the 1PPS out of an HP 5061B Cs.  Otherwise it returned a time of 3hz instead of 1hz for some reason.  All done in 50 ohm mode.
>
> Otherwise a nice unit for the money.  I’m thinking someone should start a group to move this off list.
>
> Regards,
>
> Jerry
>
>
>> On Oct 3, 2019, at 8:59 PM, Mark Sims <holrum at hotmail.com> wrote:
>>
>> I did a test of the FA2 internal OCXO warm up time.  I powered mine down for 4 fours and then powered it up and measured the frequency as it warmed up.  It looks like it takes 45 minutes to stabilize.
>>
>> Note that it took a week of constant power for the oscillator drift rate to settle down.  The drift on mine seems to be around 0.00025 Hz/day.  I need to do a longer test to better weed out temperature sensitivity.
>>
>> BG7TBL internal oscillator warm up time:
>>
>> Minutes   Freq error
>> 0         221 Hz
>> 1         17 HZ
>> 2         0.2760 Hz
>> 3         0.0500 Hz
>> 4         0.0350 Hz
>> 5         0.0240 Hz
>> 10        0.0130 Hz
>> 15        0.0080 Hz
>> 20        0.0060 Hz
>> 25        0.0050 Hz
>> 30        0.0042 Hz
>> 35        0.0040 Hz
>> 40        0.0034 Hz
>> 45        0.0021 Hz <--- stable
>> 50        0.0025 Hz
>> 55        0.0022 Hz
>> 60        0.0020 Hz
>>
>> _______________________________________________
>> time-nuts mailing list -- time-nuts at lists.febo.com
>> To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
>> and follow the instructions there.
> _______________________________________________
> time-nuts mailing list -- time-nuts at lists.febo.com
> To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
> and follow the instructions there.