3 Spectrum Analyzer Mode
3.2 Swept SA Measurement
Instead, we can choose the frequency for the marker to higher resolution than the
bucket resolution by interpolating from the trace data. This technique does not
require another“zoom sweep” and thus will not misbehave with nonstationary
signals like pulsed RF. Therefore, there should be no need to defeat the zoom
function under user control.
In EMC measure-at-marker operations, we assume a CW signal, because fine-
tuning the frequency of a CW signal can have a significant effect on accuracy. For
other signal statistics, such as noise and pulsed-RF, fine-tuning is neither possible
nor required. So, we can assume CW.
This algorithm will address two issues:
1.
When to position the zero-span acquisition at a new frequency
2.
What that new frequency is
To determine 1) we need to first perform the calculation in 2).
First we specify that the zoom will always be performed with the peak detector. The
math is different (simpler but much different) with the sample detector and
unreasonably complicated with other detectors, so it is not reasonable to assume
anything other than a peak detector. The algorithm can still operate for other
detectors, but it is not as accurate for those.
From the marker amplitude and the amplitudes at the adjacent buckets, we can
compute the peak frequency for a CW signal. In fact, with the peak detector, we can
compute that frequency from the marker bucket and the next lower bucket, and we
can semiredundantly compute it symmetrically from the marker bucket and the next
higher bucket. If these two agree, it means the signal is CW; if they disagree, it is
not. If it is CW, we will change the frequency at which we acquire data for the
measure at marker from the frequency of the marker; if the signal is not CW, we will
not change the frequency (not zoom).
Step 1. Compute the frequency from the lower pair of buckets. Compute
fLowerCandidate = fCenterOfBucket – 0.5*BucketWidth + RBW3dB*sqrt(deltaA/
(3dB))
fLowerCandidate = the estimated CW frequency computed from the lower bucket
and marker bucket
fCenterOfBucket = the frequency at the center of the bucket. This is computed from
fStart + BucketNumber * Span / (NumBuckets – 1). Of course, this expression in not
computable if NumBuckets = 1, which is a nonsensical case, but the instrument
should not crash when set up this way. When NumBuckets = 1, use 2 for
NumBuckets.
BucketWidth = Span / (NumBuckets – 1)...as above, to protect against the one-
bucket case, use 2 for NumBuckets when NumBuckets = 1.
Spectrum Analyzer Mode User's &Programmer's Reference 365
To Next Page
Loading...
Need help?
General
