EEVblog #459 - Counter Shenanigans

[JackOfVA]

Dave, how accurate are the Rigol 10Mhz out compared to your Rubidium standard. ?

Without even testing it, I can tell you it's crap. Same as the internal crystal in the Agilent counter.

There's another issue with the Rigol time base that surfaces in the function generator. I have a DG-4102 and found that there's a lot of phase noise and/or jitter on the output when using the internal time base. But the output cleans up very nicely when the DG-4102 is clocked with a high quality external time base.

The two plots below show the difference. Test setup is to put the DG-4102 to Sine output at 10.050 MHz, and connect that to a frequency mixer (in this case, I used a Mini-Circuits mixer) and the second mixer input to a known high quality clean crystal oscillator (in this case, an HP oven 10 MHz oscillator). The difference signal, centered at 50 KHz, is piped into an HP 3562A dynamic signal analyzer, which for the purposes of this discussion can be thought of as a very low noise FFT spectrum analyzer operating over the range 0-100 KHz. Mixing the DG-4102 output with a clean 10 MHz signal lets you magnify, in effect, the phase noise by an instrument with very narrow  bandwidth operating at 50 KHz. The 3562A frequency span is set at 20 Hz, so each horizontal division is 2 Hz.

The first plot shows phase noise with the DG-4102's stock time base. Second plot shows the phase noise with the DG-4102 external time base input connected to an HP 3816A GPSDO. (I say phase noise, could be considered jitter as well.)

Quite the difference.

Rigol should have taken a leaf from Agilent's book and made an extra cost option of a high quality time base.

[vk6zgo]

suggestion for what video amp choose?

Some TV VDAs go out to 10Mhz or so,but others roll off a bit more sharply.
Don't get a VCA,as the clamp circuit might mangle your signal-----it relies on the composite video signal for its timing.

[ddavidebor]

Thanks.

And another thing, my clock give me a square wave, because a lot of instrument accept square wave (i'm right?) is ok if i use digital buffer of rthe right speed for it?

[Joules]

Thanks JackOfVA, that was just the data I was looking for.

[JackOfVA]

Out of curiosity, I connected a Racal 1992 frequency counter to the 10 MHz time  base output of a Rigol DG-4102 function generator and looked at the frequency change from a cold start. The 1992 counter time base is supplied by an HP Z3816A GPSDO. 1992 set to read frequency to 0.01 Hz increments, with a 1 second gate. (Some clever work by the lads at Racal.) Data collected via the IEEE-488 bus.

Plot below shows results from a cold start (last time it was powered up was several days ago) with approximately 3.5 hours of data collected. I'll let it run overnight and see how it looks tomorrow morning.

Leaving aside the questionable short term jitter, the overall frequency accuracy and stability isn't all that bad. After a rather short warm up period, it's stabilized about 0.5 Hz high at 10 MHz or 0.05 ppm.

The DG-4102 is in my basement shop, so it's not exposed to large temperature swings, which should help considerably.

It's about a year old and I have not adjusted the time base, by the way.

[Teneyes]

Rigol DG-4102 function generator .
collected. I'll let it run overnight and see how it looks tomorrow morning.

Leaving aside the questionable short term jitter,
the overall frequency accuracy and stability isn't all that bad. After a rather short warm up period,
it's stabilized about 0.5 Hz high at 10 MHz or 0.05 ppm.

It is not correct to say that the error is 0.05 PPM
And the combined jigger and short term stabilty within +- 0.003 PPM.

It would be nice to see your same setup with the Rigol DS4102 generating  100MHz when also externally locked to the GPSDO.  ( to reudce the phase jitter), Please 

[JackOfVA]

Rigol DG-4102 function generator .
collected. I'll let it run overnight and see how it looks tomorrow morning.

Leaving aside the questionable short term jitter,
the overall frequency accuracy and stability isn't all that bad. After a rather short warm up period,
it's stabilized about 0.5 Hz high at 10 MHz or 0.05 ppm.

It is not correct to say that the error is 0.05 PPM
And the combined jigger and short term stabilty within +- 0.003 PPM.

It would be nice to see your same setup with the Rigol DS4102 generating  100MHz when also externally locked to the GPSDO.  ( to reudce the phase jitter), Please 

Plot below shows 17 hours of data.

By an error of 0.05 ppm, I meant the offset error.  Short term excursions are considerably less as you note.

Don't understand which data set you want to see with the Rigol at 100 MHz - a frequency count or the spectrum display showing close in noise and jitter?

Frequency count is simple enough, but in order to measure close in noise over a few Hz span requires a clean, stable, source of 100 MHz to mix with the Rigol. At 10 MHz I can do this easily with an HP precision crystal oscillator with excellent phase noise performance. Unfortunately I don't have such a reference source for 100 MHz. All my 100 MHz signal generators are synthesized so they will be the limiting factor.  (I suppose my elderly HP 8640B would be the cleanest of the various signal generators I have, but even with the frequency lock enabled, it moves around a few Hz over the short term at 100 MHz.)