CHASING THE ILLUSIVE MILLIVOLT & NANOSECOND – PART II
....
Letting the scope automatically measure the 462 MHz frequency via. the red HELP button, indicated 474.7 MHz, which is 2.3% high.
...
Hey Old-E, was rereading some older posts of this thread and came across your post. I wonder if you are still around and ever where able to draw any causes/solutions/improvements or conclusions in reference to the inaccuracy of the parametric measurements in comparison to using cursors on the waveform. I also found that in particular the frequency measurements where way off for my taste and expectation, specially for a top of the line scope as the 2465B. But to be fair and realistic, of course it deserves many merits, but frequency counting is definitively not one of the fortes of this model.
Just for kicks recently I tried to measure the frequency of a 10MHz output from my recently calibrated Rubidium oscillator, and after a few seconds after pushing the Measure and then selecting CH1 for Frequency, the scope just presented a scant measurement result of "10.16MHz"
. If I use the cursors to measure frequency, they seem dead on. If I measure that same signal on any of my counters or with the 2247A scope, it shows of course 10-dot plus a whole lot of zero decimal places all the way into the lower Hz region.
I also did some testing with my R2008HS service monitor that can generate up to 1GHz and has the High-Stability time base option and is calibrated within a few hertz of the expected frequency output, and got similar disappointing results while trying to measure frequency on the 2465B. The error seems to get larger as frequency goes up. At 400MHz it was showing an error of several MHz high (I seem to remember it was showing 416MHz). The frequency measurement stopped responding around 650MHz, but the scope would still trigger without any issues even at 800MHz, although the waveform was mighty fuzzy due to using the lower 10mV input range, but rock solid stable.
The semi conclusion that I have reached is that at least frequency parametric measurements on the 2465B are not very accurate, maybe unless you have the CT options things might get better (?), so don't expect much in that sense. Also, as it was explained to me recently, the way the scope actually tries to measure the frequency of a signal is not at all like a traditional frequency counter would, but it rather uses a very convoluted procedure to get to the results. Only to start with, it involves a Miller integrator, and the CPU and DAC interacting to get a result. In case anyone is interested, still do have the detailed explanation of the measuring procedure by someone that knows these scopes very well in and out, and has been working on them and calibrating/selling them for many years.
BTW: I just recently got this 2465B, and so far have not done any calibration to it. But after some testing it seems still to be quite well calibrated in both time/div and volts/div. At least with the equipment I have available which does not allow me to do a full verification/calibration.