Coded long pulse test
24sep09
Links to plots:
various
properties of the transmitter samples (.ps) (.pdf)
Spectra
of coded long pulse height data (.ps) (.pdf):
Image
of 1 second of averaged data (with decoding) (.gif):
Image
of 1 second of averaged data (with no decoding) (.gif):
A test was done with the rdev
spectrometer on 24sep09 using the dome 430 transmitter.
The setup:
- gregorian dome 430 txmitter/receiver with 420-442 filter
installed on receive.
- cycle thru coded long pulse, power, and mracf
- codedlong pulse: 40 seconds (2 usec baud)
- mracf, power 10 seconds each
- 10 millisecond ipps were used.
- Rdev setup:
- 160 Mhz clock, decimate by 5: 32 Mhz Bw, 16 bits output,
data rate 32 Mhz i,q.
- txSamples: skip 2, numsamples:20000
- hghtSamples: start: 20000/32e6=625 usecs=93km: numSamples:65000
- Set rms in counts at digitizer to about 80 counts.
- Data file: used
'/share/tdevs0/sasdr_gold.20090924.b0a.02300.pdev'
The transmitter samples:
The plots show various
properties of the transmitter samples (.ps) (.pdf):
- Page 1: When the programs cycled:
- Top: the power in the Tx sample 2000: (62.5 usecs).
- This is beyond the end of the barker code
- You can see the program transitions because the power changes
a little between program types
- There is only 5 secs/20 seconds of data for pwr,mracf / clp.
There should be 10/40 so we are missing 1/2 of the data.
- Bottom: The Voltage in the I digitizer for sample 2000.
- Page 2: Examples of clp,
mrac, power tx samples:
- 75 usecs of tx samples was plotted for the 3 programs. I chose
an ipp where the i,q digitizers had equal power.
- There is a .5 usec oscillation apparent in the data.
- Page 3: ipp to ipp oscillations
in the transmitted power:
- The power was varying between the I and Q samples on an ipp to
ipp basis. To measure this i computed:
- y[i]=mean(txSmp[0:1600,i]^2) and then plotted this versus ipp.
- The power variation is seen in all 3 programs.
- The frequency is doubled since i squared the voltage.
- Page 4: Spectrum of the ipp to
ipp tx oscillations.
- These plots show the spectrum
of the ipp to ipp oscillations.
- In power it is: 5.86 Hz. In voltage it is 2.93 Hz.
- Page 5: Coded long pulse
Average transmitted spectrum:
- 3000 points (93.5usecs)
from the start of each clp ipp was used to compute the spectrum.
- The spectra were then averaged over 19 seconds of data.
- Top: 32 Mhz bandwith of average spectra.
- I've flagged 2Mhz where the previous oscillation within the
ipp was seen.
- Bottom: blowup +/- 1 Mhz about 430.
- The nulls are at the correction locations (.5 Mhz) for a 2
usec baud).
- The main lobe is offset by about 100 Khz.
The data samples:
The plots show Spectra
of coded long pulse height data (.ps) (.pdf):
- 1 second of data was used (100 ipps).
- Page 1:Average all heights
together (linear scale)
- Top: No decoding
- Bottom: with decoding.
- Page
2:Average all heights together (db
scale)
- Top: No decoding
- Bottom: with decoding.
- Page 3:Average spectra 1 height
(123 km) (db scale)
- Top: No decoding
- Bottom: with decoding.
- The decoding is not working.
- I tried decoding a single height using the tx sample offset by
up to +/- 20 ipps from the correct ipp. I did not see any narrowing of
the ion line or the plasma line.
Images where made of 1 second of height data.
- The x axis is frequency, the y axis is range.
- Images were made with and with decoding.
- The images were flattened using the upper 15 km of data.
Image of 1
second of averaged data (with decoding) (.gif):
Image of 1
second of averaged data (with no decoding) (.gif):
Conclusions:
- We are only getting 1/2 of the data output
- the clp data does not decode even after shifting up/down by up to
20 ipps.
- There is an ipp to ipp oscilation of about .34 hz.
- With an ipp there is a 2 Mhz oscillation (30 db down).
- The main lobe of the clp tx spectrum is offset by about 100 Khz
but the nulls are in the correct location.
processing:
x101/rdev/24sep09clp/test_24sep09.pro
<-
page
up
home_~phil