Download presentation
Presentation is loading. Please wait.
Published byEugene Jackson Modified over 9 years ago
1
Tx “bad” lags and range data gaps Pasha Ponomarenko 10/10/2014STELab discussion1
2
Sampling rate problem Problem: for the single-pulse technique, the sampling rate should satisfy two mutually exclusive requirements: |V| 2000 m/s: f 200 Hz d 3500 km: f 40 Hz Range/time targets pulses 10/10/2014STELab discussion2
3
Solution Solution: measuring autocorrelation function (ACF) with sampling rate > 200 Hz and cancelling unwanted echoes from other ranges through coherent averaging. It can be measured using a sequence of unevenly spaced pulses which provides a set of time lags. 10/10/2014STELab discussion3
4
SD pulse sequence Blanking pulses 10/10/2014STELab discussion4
5
Sampling ACF Each pulse in the sequence generates its own echo profile, e.g. like this: Power Time/group range 10/10/2014STELab discussion5
6
ACF sampling and Tx overlap1 10/10/2014STELab discussion6
7
Sampling ACF (cont.) For different pulses, these profiles are very similar but shifted in time domain according to the pulse separation in the sequence. As a result, the sampled echoes represent superposition of the echoes from different samples. 10/10/2014STELab discussion7
8
ACF sampling and Tx overlap 1 2 3 4 5 6 7 10/10/2014STELab discussion8
9
Sampling ACF (cont.) For any given range, we can calculate when the return from a given pulse should arrive by shifting the “mask” (pulse sequence) along the receiver sample time series. The sampled range gate is determined by a position of the first pulse in the sequence. 10/10/2014STELab discussion9
10
Sampling ACF (cont.) 1 2 3 4 5 6 7 10/10/2014STELab discussion10
11
Sampling ACF (cont.) Then we combine pairs of the receiver samples (pulses) to generate complex ACF values at different time lags. The lag values are determined by the separation of the pulses in each pair. 10/10/2014STELab discussion11
12
1 2 3 4 5 6 7 1010 2020 3030 10/10/2014STELab discussion12
13
Averaging Together with the “wanted” signals coming from the desired ranges, we also receive echoes from “unwanted” ranges generated by other pulses of the sequence. In order to minimise this interference, we need to average ACFs. In this case the coherent components from desired ranges will remain unchanged while the incoherent cross-range interference will be suppressed due to its random phase relation with the desired signal. The signal itself would also become less coherent with increasing so that the averaging also allows to measure ACF power decay time (spectral width), which is impossible with just a single pulse sequence. AB c / 2 cc 10/10/2014STELab discussion13
14
Tx overlap and power gaps At certain locations along the receiver sample sequence some of the “pulses” in the mask coincide with the Tx emission times so that the receiver is blanked and the respective sample has very low (ideally zero) power. 10/10/2014STELab discussion14
15
Tx overlap and power gaps 1 2 3 4 5 6 7 10/10/2014STELab discussion15
16
Tx overlap and power gaps If one of the pulses is “blanked” (i.e. it has very low power) then the respective ACF lags should also have low power. 10/10/2014STELab discussion16
17
If we know which pulse is a low-power one, it is easy to calculate which lags of the respective ACF will have low power too. This combination is unique for each “lost” pulse For example, in the old pulse sequence, for pulse #2 these will be lags 3, 9, 11, 13, 17 and 18. 1 2 3 4 5 6 7 Low-power lags 10/10/2014STELab discussion17
18
It can be predicted The dostribution of low-power lags vs range gate is also unique for each combination of pulse sequence, basic ACF lag (mpinc), spatial resolution (rsep) and initial range (lagfr) so that it can be predicted for the given set of the above parameters. 10/10/2014STELab discussion18
19
On-line tool @ VT: automatic “bad” lag calculator 10/10/2014STELab discussion19
20
On-line tool @ VT 10/10/2014STELab discussion20
21
On-line tool @ VT: Predicted and observed gaps 10/10/2014STELab discussion21
22
Range periodicity The affected range gates appear periodically, every mpinc/rsep gates. old sequence: 2400 µs /300 µs = 8 gates katscan: 1500 µs/300 µs = 5 gates high-resolution katscan: 1500 µs/100 µs = 15 gates 10/10/2014STELab discussion22
23
Range offset estimate Matching observed and predicted range/lag patterns of low-power lags can be used for finding range offsets which are greater or equal to the spatial length of the blanking pulse (1-2 gates). Using the observed position of the Tx-overlap lags as a time-of-flight reference is possible because these blanking pulses are accurately synchronised with the emission regime. Otherwise, the input circuits of the receiver would have been burned, or at least we would observe power peaks instead of the power gaps. 10/10/2014STELab discussion23
24
Range offset estimate (cont.) The “blanking” occurs at the input circuits of the receiver so that any following filtering could shift or smooth the whole pattern but it would not change its temporal structure, i.e. relative positions of the power gaps both in range and ACF lag. 10/10/2014STELab discussion24
25
Another on-line VT tool: ACF plotter 10/10/2014STELab discussion25
26
Example: No offset Tx overlap 10/10/2014STELab discussion26
27
Example: No offset Tx overlap 10/10/2014STELab discussion27
28
Example: No offset Tx overlap 10/10/2014STELab discussion28
29
Example: No offset Tx overlap 10/10/2014STELab discussion29
30
Example: No offset Tx overlap 10/10/2014STELab discussion30
31
Example: No offset Tx overlap 10/10/2014STELab discussion31
32
Example: No offset Tx overlap 10/10/2014STELab discussion32
33
Example with offset 10/10/2014STELab discussion33
34
Example: 3-gate delay Tx overlap 10/10/2014STELab discussion34
35
10/10/2014STELab discussion35 Example: 3-gate delay Tx overlap
36
10/10/2014STELab discussion36 Example: 3-gate delay Tx overlap
37
10/10/2014STELab discussion37 Example: 3-gate delay Tx overlap
38
10/10/2014STELab discussion38 Example: 3-gate delay Tx overlap
39
10/10/2014STELab discussion39 Example: 3-gate delay Tx overlap
40
10/10/2014STELab discussion40 Example: 3-gate delay Tx overlap
41
10/10/2014STELab discussion41 Example: 3-gate delay Tx overlap
42
Yet another on-line tool: offset estimator 10/10/2014STELab discussion42
43
Range offset Automatic way to estimate the offset value and sign is implemented at the VT website under Data Diagnostic / Badlag Finder Just enter the suspect radar, data and time and press “Find”. Wait until the calculation is finished and then click on the image to open a PDF file. Scroll to the second page. 10/10/2014STELab discussion43
44
10/10/2014STELab discussion44
45
Range offset (cont.) The peak on all plots should coincide with the vertical dashed line. If this is not the case, then there is a range offset. Its magnitude and sign are determined by the shift from the dashed line. In the above case we have a 3-gate negative shift so that the sampling starts earlier than expected. In this case the first range gate has to be shifted by 3 gates closer to the radar. i.e. from 120 km to (120 km-3*15 km)=75 km. 10/10/2014STELab discussion45
46
“Normal” offset (time delay) 10/10/2014STELab discussion46
47
Reverse offset (late sampling) 10/10/2014STELab discussion47
48
Possible causes of the time offset Sampling start earlier than required (“normal” offset): – Time delay of the signal inside the receiver (e.g. due to narrow-band filtering). Sampling starts later than expected (reverse offset): – Overcorrection of the “normal” offset – Something else (?) 10/10/2014STELab discussion48
49
Enough for today? 10/10/2014STELab discussion49
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.