Just ordered a BG7TBL FA-2, was this a good move?

[apblog]

Hi everyone.

I just got a little impulsive and, after reading eevblog all afternoon, ordered a BG7TBL FA-2 12 GHz frequency counter, and a BG7TBL GPSDO.

My application is measuring a variable frequency oscillator in a digital audio DAC/ADC box.

I am hoping for 0.01 Hz precision at 25 MHZ.  That will allow me to watch the oscillator changing as it adjusts to changing incoming digital audio.
That's 10 digits which is a little expensive to buy new.

In some ways I am less interested in accuracy than in relative movement.

My biggest question is: how long does it take for it take a reading? I forgot to actually check this before ordering. 

Also, does this gear seem generally appropriate for my use case?

And how difficult/inconvenient is it going to be to actually use this setup to get my work done?

Thank you.

[ArthurDent]

There are 3 gate times, 0.1, 1.0, and 10 seconds. Accuracy and stability are two different things. I use an external GPSDO which gives me high accuracy. I believe one user said theirs was within 0.1Hz as received. It is more stable than that if you leave it on for a long time or 24/7. You won't know how accurate it is unless you have a more accurate standard to compare it to.

[Lemonizer]

It seems we are in the same boat ! I'm kinda new to the time-nut game, and ordered myself a FA-2 but the 6Ghz version (I think there's no much differences, besides the max frequency).

I should get it in a few days, hopefully ! I also have a BG7TBL GPSDO, which locks in a few minutes, and seems to be great, according to many people here. I need a second GPSDO, if I want to measure it. It never ends ! 

If you manage to get a clear sky view for your GPSDO to grab a strong signal, I'm pretty sure that you'll get the accuracy you are aiming at, I can't tell for the minimum time you'll need for a reading, but usually, from what I get, the longer the better resolution you'll get.

It seems to be a neat little device, I can't wait to get mine ! 

[DimitriP]

Also, does this gear seem generally appropriate for my use case?

And how difficult/inconvenient is it going to be to actually use this setup to get my work done?

A) No
B) Kinda

Are you trying to measure oscillator jitter while the DAC is processing audio?
I'm sure someone else will jump in but a DSO will be more useful for that.

[Electro Fan]

FA-2 being fed 10,000,000 Hz signal from 33522B.  Both are sync'd to a 2018-06-05 LCD BG7TBL GPSDO that has it's 10 MHz reference signal distributed by a BG7TBL 2018-08-16 distribution amplifier.  You can see the Ext-Ref indicator on the FA-2 display and the Ext Lock indicator on the 33522B display.  Signal on the GPSDO is kinda weak, usually has 2-3 bars - might need to re-position the antenna.  The pwr measurement is off 0.1 dBm.

edited to correct model # typo

[apblog]

FA-2 being fed 10,000,000 Hz signal from 33522B.  Both are sync'd to a 2018-06-05 LCD BG7TBL GPSDO that has it's 10 MHz reference signal distributed by a BG7TBL 2018-08-16 distribution amplifier.  You can see the Ext-Ref indicator on the FA-2 display and the Ext Lock indicator on the 33533B display.  Signal on the GPSDO is kinda weak, usually has 2-3 bars - might need to re-position the antenna.  The pwr measurement is off 0.1 dBm.

Thanks for posting those pictures.  I wish I'd known about the LCD version of the GPSDO before I ordered.

I'll admit to being a bit dense.  What conclusion should I draw from the pictures?  I would expect that if a generator and counter were both synced to the same reference, that of course the frequency readout would be dead on. 

What am I missing?

[Electro Fan]

Fair question/comment.  Others here can probably provide more and much better insight, but fwiw on my bench I have several devices that are locked to the GPSDO (you can see the cables connected to the distribution amp) and they don’t all agree to the same level as the GPSDO, 33522B, and the FA-2. Two other counters (HPs), an oscilloscope (Tektronix), a spectrum analyzer (HP), and a sig gen (Agilent) are also driven by the BG7TBL GPSDO but they are less well aligned - so it’s my experience that just because devices share a 10 MHz reference source they don’t necessarily measure a shared input (DUT) signal as having the same value.

I think the first big assumption is that the GPSDO’s 10 MHz ref is accurate.  I don’t have a good way to know other than it seems to compare reasonably well to another (Star) GPSDO I occasionally use.

As for the FA-2 I think it agrees more closely with the 33522B than two HP counters I have.  Given the pedigree and the cost of the 33522B (purchased new from Agilent/Keysight) I’m inclined to think it’s the most high quality piece of gear on my bench - but that’s just an impression.  So with the 33522B as my “yardstick” but without being able to prove exactly what’s what, I’m leaning toward the BG7TBL GPSDO and FA-2 as being pretty (very) high bang for the buck devices.

Sort of in summary, after accumulating test gear over a period of about seven years I’m still a rookie compared to the pros here but I’m inclined to think/guess that a double oven GPSDO 10 MHz reference is likely to outperform the internal oscillators of most other gear (at least on my bench) and to the extent the other gear has a 10 MHz reference input not only is accuracy likely to be improved with the GPSDO 10 MHz reference but the synchronizing of the gear across the bench brings additional mental peace even if it’s a bit of a placebo

I’m not an official card carrying TEA and Volt/Time Nut member but I sort of aspire to be
YMMV

Edit:  correction only 1 of the 2 HP counters has a 10 MHz ref input (and it is less aligned than the other HP counter but after a lengthy warm up it makes a reasonable showing); I’ve run lots of tests with and without the Ref inputs and I’m convinced the GPSDOs bring not just synchronicity but also improved accuracy - and in turn I’m convinced the FA-2 performs way beyond it’s price class.  I think it’s all a big tribute to BG7TBL.

[FransW]

What am I missing?

A personal GPS system independant from NIST.

Success, Frans

[jpb]

FA-2 being fed 10,000,000 Hz signal from 33522B.  Both are sync'd to a 2018-06-05 LCD BG7TBL GPSDO that has it's 10 MHz reference signal distributed by a BG7TBL 2018-08-16 distribution amplifier.  You can see the Ext-Ref indicator on the FA-2 display and the Ext Lock indicator on the 33533B display.  Signal on the GPSDO is kinda weak, usually has 2-3 bars - might need to re-position the antenna.  The pwr measurement is off 0.1 dBm.

Thanks for posting those pictures.  I wish I'd known about the LCD version of the GPSDO before I ordered.

I'll admit to being a bit dense.  What conclusion should I draw from the pictures?  I would expect that if a generator and counter were both synced to the same reference, that of course the frequency readout would be dead on. 

What am I missing?

There is always noise and a basic resolution with counters. I have the Tektronix FCA3100 counter which is a Pendulum 95(?) also sold by Fluke I think. The minimum one shot resolution is 50psecs with a resolution in terms of measurement of around 30 psecs I think.
It also has two modes of measuring frequency. One is Back-to-back where there is no dead zones. The other it uses linear fitting to get better resolution but this leads to small gaps (I think).

What I find is that with back-to-back I'll get exact agreement on the external reference as an average but with the linear fitting I get a very small difference perhaps in the last digit.

[ArthurDent]

You won't know how accurate it is unless you have a more accurate standard to compare it to.

My assumption was that the GPSDO would be the more accurate standard or am I missing something.

To be clearer, in the above quote I was just talking about the FA-2 using its internal oscillator. I also said I use an external GPSDO which gives me high accuracy. If you use a GPSDO with a strong signal and no reflections from nearby tall buildings, your accuracy will be close to the GPS accuracy.

[Electro Fan]

Does Anybody Really Know What Time It Is?
Kind of a metaphor for a metaphor.

Not the original reference source but a pretty good/great approximation:


The original (actually something along the line of a digitization uploaded to the web probably made from a vinyl record made from a master disc probably made from a tape of the original, probably after it was mixed):


But the question remains:  Does anybody really know what time it is?

At some point we just have to appreciate and enjoy however close we can get, as we keep trying figure out how to get closer.

[apblog]

HAHA!  great metaphor, and great song. 

I didn't recognize it right away until they started singing. 

That cover band was fantastic too!  I might actually prefer that to the original.

Thanks everyone for helping me understand.

[Electro Fan]

FA-2 being fed 10,000,000 Hz signal from 33522B.  Both are sync'd to a 2018-06-05 LCD BG7TBL GPSDO that has it's 10 MHz reference signal distributed by a BG7TBL 2018-08-16 distribution amplifier.  You can see the Ext-Ref indicator on the FA-2 display and the Ext Lock indicator on the 33522B display.  Signal on the GPSDO is kinda weak, usually has 2-3 bars - might need to re-position the antenna.  The pwr measurement is off 0.1 dBm.

edited to correct model # typo

Making a follow-on to the previous post:

Below is a photo of the FA-2 set to 10s gate while measuring the same DUT (Agilent 33522B set for 10 MHz) but in this case the 10MHz reference is coming from a Star GPSDO with an Oscilloquartz 8663 rather than from the BG7TBL GPSDO shown above.  There are some differences in the tests:  1) not as many samples on the Star as the BG7TBL (l'll see if I can catch the Star when it reaches a comparable number of samples), 2) the warm up periods are ballpark similar (about a day) but not same, and 3) they (the BG7TBL and the Star) both have similar but not the same "hockey puck" antennas sitting a couple inches apart from each other looking at the same view of the sky.

Long story short, I think the BG7TBL GPSDO is holding it's own vs. the Star GPSDO but we'll see if anything notable changes as we get more samples.

Edit:  Added photo of Star GPSDO with more (5278 samples) - didn't dramatically change results

[jpb]

I can't comment on the BG7TBL vs the Star as I don't have the BG7TBL, but my experience with my two Star 4+s is that they are "better" over periods up to 1000 seconds if the time constant is made much longer than the default. The default is 200 seconds while I think a setting of 3600 secs (1 hour) gives a much better ADEV curve.
I posted some early comparisons here (I've done some better measurements since to get down to 1 second for all the cases but they are not yet complete):
https://www.eevblog.com/forum/metrology/some-adev-measurements-for-star4-gpsdo-at-different-time-constants/msg2854130/

[Electro Fan]

jpb, your adjustments on the GPSDO no doubt are an enhancement, I just haven't figured out how to do those yet - but what got me going on these tests was this post from EV

https://www.eevblog.com/forum/metrology/bg7tbl-fa1-frequency-analyzer/msg2894686/#msg2894686

EV's newer (12.4 GHz) FA-2 seems to show roughly an order of magnitude improvement - whether that's due to the counter or maybe his GPSDO (better Ref output?), or maybe his DUT, or some combination I don't know.  I just liked the result as shown in the link to his post

[jpb]

jpb, your adjustments on the GPSDO no doubt are an enhancement, I just haven't figured out how to do those yet - but what got me going on these tests was this post from EV

https://www.eevblog.com/forum/metrology/bg7tbl-fa1-frequency-analyzer/msg2894686/#msg2894686

EV's newer (12.4 GHz) FA-2 seems to show roughly an order of magnitude improvement - whether that's due to the counter or maybe his GPSDO (better Ref output?), or maybe his DUT, or some combination I don't know.  I just liked the result as shown in the link to his post

The FA-2 does seem to be very good value for money - I'm tempted to buy one but already have two counters including the FCA3100 which cost £1500 when I bought it!
The time setting on the Star 4 is very easy with Lady Heather - you just use &t and you can set values in a range up to 5000 seconds. I can't remember what the lower end of the range is (10 seconds?) I found that anything below 100 seconds gave rather erratic behaviour.

[Electro Fan]

I can't comment on the BG7TBL vs the Star as I don't have the BG7TBL, but my experience with my two Star 4+s is that they are "better" over periods up to 1000 seconds if the time constant is made much longer than the default. The default is 200 seconds while I think a setting of 3600 secs (1 hour) gives a much better ADEV curve.
I posted some early comparisons here (I've done some better measurements since to get down to 1 second for all the cases but they are not yet complete):
https://www.eevblog.com/forum/metrology/some-adev-measurements-for-star4-gpsdo-at-different-time-constants/msg2854130/

Hi jpb,

Just to clarify, when setting &t (which displays as User time constant and Real time constant) to 3600 seconds - is this going to impact the ADEV info displayed by Lady Heather, or is it also going to impact the stability or accuracy of the 10 MHz ref output?  Thx

[jpb]

It changes the time constant used by GPSDO. It shouldn't affect the ADEV measurement process (though obviously it changes the behaviour of the GPSDO so the ADEV measurements will be different).

I don't use lady heather for ADEV measurements, I use Stable32 - I have C code which talks to my counter and generates files and then I feed the file to Stable32 so I have never actually done an ADEV measurement directly on lady heather though I'm aware it is able to do so.

To clarify a bit more. The &t command sends a command to the GPSDO which changes the parameter value on the GPSDO itself. The value will stay in place while the GPSDO is turned on. You can disconnect lady heather and restart and the time constant should be the same as was set. If the power to the GPSDO is turned off then the time constant will revert to 200 seconds. The limited range of up to 5000 seconds is a limitation of the GPSDO commands and not a lady heather thing (lady heather lets you set a higher value but the GPSDO won't accept it).

[notfaded1]

I have an FA-2 as well.  It just doesn't show enough data.  I'm not sure if there's a way to read the data from the FA-2 so you can graph ADEV and MDEV with timelab?  I've found that until you get the data into timelab or stable32 it's really hard to tell what you've got.  I'm using HP 53132A and SRS SR620 to capture data.  Has anyone figured out how to get data like this from an FA-2 into timelab?

Regards,

Bill

[Electro Fan]

It changes the time constant used by GPSDO. It shouldn't affect the ADEV measurement process (though obviously it changes the behaviour of the GPSDO so the ADEV measurements will be different).

I don't use lady heather for ADEV measurements, I use Stable32 - I have C code which talks to my counter and generates files and then I feed the file to Stable32 so I have never actually done an ADEV measurement directly on lady heather though I'm aware it is able to do so.

To clarify a bit more. The &t command sends a command to the GPSDO which changes the parameter value on the GPSDO itself. The value will stay in place while the GPSDO is turned on. You can disconnect lady heather and restart and the time constant should be the same as was set. If the power to the GPSDO is turned off then the time constant will revert to 200 seconds. The limited range of up to 5000 seconds is a limitation of the GPSDO commands and not a lady heather thing (lady heather lets you set a higher value but the GPSDO won't accept it).

Roger all that, but does changing the &t command result in any impact on the 10 MHz reference output from the Star 4?

[jpb]

It changes the time constant used by GPSDO. It shouldn't affect the ADEV measurement process (though obviously it changes the behaviour of the GPSDO so the ADEV measurements will be different).

I don't use lady heather for ADEV measurements, I use Stable32 - I have C code which talks to my counter and generates files and then I feed the file to Stable32 so I have never actually done an ADEV measurement directly on lady heather though I'm aware it is able to do so.

To clarify a bit more. The &t command sends a command to the GPSDO which changes the parameter value on the GPSDO itself. The value will stay in place while the GPSDO is turned on. You can disconnect lady heather and restart and the time constant should be the same as was set. If the power to the GPSDO is turned off then the time constant will revert to 200 seconds. The limited range of up to 5000 seconds is a limitation of the GPSDO commands and not a lady heather thing (lady heather lets you set a higher value but the GPSDO won't accept it).

Roger all that, but does changing the &t command result in any impact on the 10 MHz reference output from the Star 4?

Yes. That is the whole point of changing the time constant. The OCXO in the Star 4 has very good short term stability (say 1 to 100 secs) - certainly much better than GPS over that range because of the GPS jitter.
With a short (100 second) time constant the noisy GPS data is pulling the OCXO one way and another so its short term ADEV at 1second say goes from a good 2 x 10^-12 to something around 2 x 10^-10 which is 100 times more noisy. The 10MHz output with t = 3600 is a much better reference particularly over the time spans it is being used (10 second gate say).

I've found this in practice, when I set the time constant very low (10 secs) it was so noisy I found it difficult to take measurements with it as a reference.

It seems, from my various experiments, that setting a time constant of t = T affects the behaviour for  tau values much less than T so the default of 200 seconds has a major effect even down at 1 to 10 seconds.

The only negative effect of setting t to a few thousand rather than a couple of hundred seconds is that the stability over periods of a day or so is a bit worse but it is still of the order of 10^-13 (ADEV) so I don't think this is important unless you are trying to measure the long term behaviour of say a Caesium source.

[jpb]

I have an FA-2 as well.  It just doesn't show enough data.  I'm not sure if there's a way to read the data from the FA-2 so you can graph ADEV and MDEV with timelab?  I've found that until you get the data into timelab or stable32 it's really hard to tell what you've got.  I'm using HP 53132A and SRS SR620 to capture data.  Has anyone figured out how to get data like this from an FA-2 into timelab?

Regards,

Bill

I don't have an FA-2 but I thought from what Texaspyro has posted it spews out ascii measurements.

if it's like the FA1, it outputs an ASCII stream of the freq measurements.

The only problem might be the need to filter out spurious other messages if they appear.
The frequency values might need to be converted. For Stable32 you need to make them relative i.e.
(Fmeas - Fnom)/Fnom
Where Fnom = 10MHz (probably) and Fmeas is the value spewed out by the FA-2.

The first step is to open up a terminal (Putty if you're using windows) and see what comes out - I presume the USB is just converted serial output.

[Electro Fan]

A question on the Star 4+

KEFX says that it can be invoked with /rx4 as follows:

http://www.ke5fx.com/heather/heather.pdf
/rx4 - Oscilloquartz STAR-4 GPSDO (9600:8:N:2)

Are we/you pretty sure that it really wants 2 stop bits?  (I have to admit, for a long time I was connecting to it with 1 stop bit and didn't notice any issues.) 

Thx

[jpb]

I just use rx4 I think - I set it up a while ago and haven't changed the settings.

If it is talking to you then I wouldn't worry, if it fails to respond properly to commands then you'll need to investigate things like stop bits.

[Kibabalu]

I also have a BG/TBL FA-2 frequency counter.

But I do not recommend it.

The Channel 2 input, the one for the higher frequencies, doesn't work reliably. During my tests with frequencies of 500 MHz and 1 GHz Sometimes it measures the correct value, sometimes not. Not predictable. Such a behaviour is useless for real measurements.