Agilent DSO X2012A Frequency Counter Stability

[Terabyte2007]

Does anyone know how I might improve the stability of the frequency counter on the Agilent DSO X2012A Oscilliscope? I know it's a software based counter, but the stability lately has been an irritation. It's accurate, but not very stable even when I feed it a stable sine wave from its own internal Gen or my Rigol DSG. If I feed these signals into my B&K Freq counter, its ultra stable. Yes, the B&K is hardware based and it will be better but if I can improve the Agilent counter even some that would be useful. If I increase the time base on the signal I can get some improvement but this is not always an acceptable solution. The waveform on the screen is triggered and very stable, cursors measurements are spot on and drift is minimal. Any suggestions would be appreciated.

[jpb]

I don't know specifically about the DSO X2012A as I have a WaveJet but as this has both a software and a hardware counter perhaps some of my experience might be useful.

The software counter (on the WaveJet) is much less accurate than the hardware one and seems to work on edge timing.  This means if you're zoomed in so there is only about one wavelength it is easily thrown by losing edges. The other thing that throws it is any noise which it mistakes for an edge. So I get better readings with averaging on and with the trigger level set about half way up the wave form and with the timebase set to allow at least a few peaks with corresponding edges. It also helps to set the bandwidth as low as feasible and perhaps to use HFreject (though I think this only affects the trigger - I'm not sure).

All the above is sort of obvious and you've probably already considered it - there are no good answers as software frequency measurement on a scope is never very accurate.

[echen1024]

The Agilent only has a Software counter.

[Terabyte2007]

I don't know specifically about the DSO X2012A as I have a WaveJet but as this has both a software and a hardware counter perhaps some of my experience might be useful.

The software counter (on the WaveJet) is much less accurate than the hardware one and seems to work on edge timing.  This means if you're zoomed in so there is only about one wavelength it is easily thrown by losing edges. The other thing that throws it is any noise which it mistakes for an edge. So I get better readings with averaging on and with the trigger level set about half way up the wave form and with the timebase set to allow at least a few peaks with corresponding edges. It also helps to set the bandwidth as low as feasible and perhaps to use HFreject (though I think this only affects the trigger - I'm not sure).

All the above is sort of obvious and you've probably already considered it - there are no good answers as software frequency measurement on a scope is never very accurate.

Thanks, I have tried everything you mentioned already but averaging. I am beginning to think it is what it is!

[georges80]

I don't know specifically about the DSO X2012A as I have a WaveJet but as this has both a software and a hardware counter perhaps some of my experience might be useful.

The software counter (on the WaveJet) is much less accurate than the hardware one and seems to work on edge timing.  This means if you're zoomed in so there is only about one wavelength it is easily thrown by losing edges. The other thing that throws it is any noise which it mistakes for an edge. So I get better readings with averaging on and with the trigger level set about half way up the wave form and with the timebase set to allow at least a few peaks with corresponding edges. It also helps to set the bandwidth as low as feasible and perhaps to use HFreject (though I think this only affects the trigger - I'm not sure).

All the above is sort of obvious and you've probably already considered it - there are no good answers as software frequency measurement on a scope is never very accurate.

Thanks, I have tried everything you mentioned already but averaging. I am beginning to think it is what it is!

Averaging definitely helps - 1000's of averages. Also, amplitude makes a difference, increase the sensitivity to get your waveform to full screen (vertically) and the 'jitter' also drops.

This is from some tests I just did with my dsox2000 series scope and an Agilent 33250A as the source. Shape makes a difference too (i.e. signal rise time). The frequency readings are much more stable with a square wave versus sine or ramp etc.

I presume the threshold decision which the counter software uses has a resolution based on the front end ADC (which averaging increases the effective # of bits and also if you 'fill the screen' vertically with the waveform.

I guess they must have run out of free gates to have to rely on software to count frequency (i.e. some marketing decision to sell more frequency counters....).

cheers,
george.

[grenert]

I always thought the DSOX-2000 scopes had a hardware counter.  I get rock-steady reading (5 digits worth) from a function generator regardless of signal type, or whether you even have a complete cycle on the screen.
I think that you need the DVM option to make this available.  Clearly the basic measurements function depends on the data shown on the screen.

[AndyC_772]

"Frequency" is a different measurement from "Counter".

"Counter" seems better.

[Someone]

I always thought the DSOX-2000 scopes had a hardware counter.

Unlocked in the DVM (multimeter) upgrade option.