Hi,
I'm using a Spectran HF 60105 X with MCS 1.7.8. I would like to know when and how long a signal is sampled.
After reading all posts of this forum, I found 2 interesting topics:
viewtopic.php?f=31&t=1303&p=3193&hilit=SampleTime#p3193
viewtopic.php?f=6&t=1243&start=0
However, my understanding of some parameters is still not perfect and I would like to share it with you.
1) Bandwidth and video bandwidth: common parameters present in a lot of Spectrum. Ok.
2) Samples: the detection filter is applied "Samples" time. "Samples" is specified by the user but when MCS detects that some part of the Spectrum can
be missed, it increases this number and indicates it on the "Display Measurement Settings". The condition that must be fulfilled to avoid
missing samples and that is used by MCS is: Samples * Bandwidth > 2 * Span. Ok.
3) When the measurement performed: as I understand and observe, the measurement is performed and once the computed result is available,
it is displayed on the screen (asap) at the cursor position. Right?
4) How long the measurement is performed is not clear at all for me. Initially I thought that "SampleTime" defines the measurement time at one
frequency ("Sample"). However, this interpretation seems not to be confirmed. For example, let us connect a 21 MHz -50 dBm signal. Let us set "SampleTime"
at 30000 ms (max value) and the detector at Min/Max. Let us run a scan between 20 and 22 MHz. The signal is correctly displayed.
Let us now switch off the 22 MHz signal just 1 s before the cursor reaches 21 MHz. With 30000 ms sampling, I expect the Min/max detector to have
recorded a peak at 21 MHz during about 29 s and to display it even if the signal has just disappeared 1 s before the sweep reaches 21 MHz. However,
the signal is not displayed. This experiment shows me that my understanding of "SampleTime" is not correct. Could you comment?
5) When setting "SampleTime" at a too low value for a given filter bandwidth, one can miss or underestimate some signals. In my opinion,
the "(Uncal)" signal should warn the user for that situation. However, the (Uncal) signal seems to be independent of the filter bandwidth and
"SampleTime" parameters. Uncal seems to be active for SampleTime < 200 ms.
Any comment would help me to do better measurements with your very nice instrument and software.
Thanks in advance,
Fabrice
Calculating the real measurement time is not simple with the SPECTRAN since we use a unique mesurement methode (e.g. "sample" is not one pixel or a RBW span on the display, its much more complex), but you can see the real sweep time (calculated with the system clock after the first sweep) at the right hand bottom of the MCS.
The "uncal" indeed shows up at a fixed sample time value because this will switch to an other measurement mode which can have some aliases at the beginning of the sweep (try the EN55011 setup and you should see a quite high noise level at the beginning of the sweep, if uncall is poping up).
Snip
