Links:
setup
setup files
running the experiment
results
summary

Other links:
using the 12m,fs,rdbe at AO: an overview

Intro:

A fringe test was done with the evn only 2gbit/sec setup on 5nov21. Arecibo took part while tracking 3C345. There were 5 scans (no0009 to no0013).

no0009 - 600 secs
no0010 - 600 secs

2 secs of data ftp'd to jive after scan

no0011 - 600 secs
no0012 - 600 secs

2 secs of data ftp'd to jive after scan

no0013 - 480 secs

A 2 secs of data ftp'd to jive after scan

Setup:

ao iflo:

vlbisetup 8433

this put 8433 at 750 Mhz,
8415 is at 768 Mhz (center of the rdbe a/d sampler)

attn if2 was -6 -4
if2mp gave -30,-30 dbm

fs setups:

on vlbis1, under /usr2/sched/n21x2.....
summary file: n21x2.sum
vex file n21x2.vex
snap file: n21x2ac.snp
proc file: n21x2ac.prc
fs logfile: n21x2ac.log
pwradj output: adjpwr.txt

Running the experiment

It's best to start this at least 20 minutes prior to first timestamp so you have time to check things.
in tclsh session
# get the telescope moving to the source

getsrc 3C345
pnt tr $ra $dec
pnt mode tr
check the monpnt12 display to see that telescope is in track mode and moving to source.

# setup the ao if/lo

vlbisetup 8433
udc query

1 8433.200000 7358.300000 15.5 15.5 Locked
1-2 ghz is offset by .2 Mhz.. this is fixed when mixing to 750 MHz

attn
attenuation:
udc 15.5 15.5 db 0 to 15.5
if2 -6 -4 db 11 to -30
if2mp

1 -30.59 -30.16

if2 .. lot of lines of output

synfrq 2249800000 1175000000 1175000000 ...
2498000000 corrects for udc .2 MHz offset

check that the ao if/lo is setup correctly

xmit a 8450 birdie for the synth/antenna in wilson's office (make sure external ref is hooked up to synth). make amp 15dbm
bring the agilent spectrum analyzer (SA) to the control room (it's heavy:)
turn on, hook up 10Mhz external reference, go to system and enable external ref
Plug the750 IF signal into the SA (front panel output).
set SA freq to

733 Mhz
.. 750 had 8433, 8450 is 17 Mhz away. and band is flipped to 750-17 =733

set SA rbw to manual, 1 hz
set SA span to 200 hz
set amplitude to linear. adjust ref level
you should see a spike at 733 mhz
When tested, go turn off the synthesizer xmitting the 8450 birdie.

edit getit.sc to ftp n21x2.vex to . (may need to change from user phil to yourself)

./getit.sc

edit doit.sc

exp=n21x2
use the vex2snap line with station Ac and no --shift
./doit.sc

schedule=n21x2ac
This starts the schedule.
wait until the setup is done..

2021.308.13:34:49.31:source=3c345,164258.8099666,394836.993986,2000

in an xterm window on vlbis1
cd ~/bin

./ddc_quantize_read_adj_set.py

This sets the rdbe attens and quantization levels

./ddc_quantize_read_adj_set.py

run a 2nd time since 1st time thresholds were set to 0.

vlbish ..

use this to query the rdbe's
rdbe2,rdbe4/dbe_dot?;

check that both rdbe's are synced to the same value

back to fs window look to make sure things start at the correct time.
at end of scan 12 you should see an xterm window popup showing the ftp xfer of 2 secs of data

--> we saw an ftp rate of 680KBytes/sec .. it should have been much higher than that.

RESULTS:

the first 100 headers from scan 12 (before vmuxing) (.txt)
With the 2 sec data sent during the experiment:

We got fringes with effelsberg when correlating our polB with their polA (we probably have our rcp,lcp reversed).
There was no correlation between our polA and Ef polB (or Ef polA)

The first plots show spectra and histograms for 2 secs of data from scan 12 (.ps) (.pdf)

page 1,2 spectra for the 4 64Mhz bands

black is polA (dbe0,rdbe2,our rcp)
red is polB (dbe1,rdbe4,our lcp)
the spectra have been normalized to a mean of 1.
some of the rdbe params are listed on the plots
polA,dbe0, our rcp only has data over 32Mhz (1/2 the band)

Page 3,4 histograms of the voltage samples

they look ok ..

Spectra output using m5spec.py

the files were first vmuxed with: vmux 5032 6400 "0,1,2,3,4,5,6,7"
the spectra were then made with: m5spec.py name VDIF_40000_8_2 1000 16384

1 second of data from scan12 (.png)

chan 1,3,5,7 are the 4 freq bands from dbe0,polA,ourRcp
chan 2,4,6,8 are the 4 freq bands from dbe1,polB, ourLcp
dbe0 only has power in the 1st 32Mhz of the 64Mhz band

This channel was correlated with Ef lcp and there was no correlation

dbe1 has power across the entire 64 Mhz

this pol correlated with Ef RCP.

We checked the field system log files, and all of the n21x2ac rdbe configurations completed correctly. there were no errors.

500 ms of data after n21x2 had finished (.png)

We took 500 ms of data after n21x2 had finished.
We did this with some routines of Louis' that talked directly to the rdbe and the mk6

the rdbe was not reinitialized.

You can see that the channel 1,3,5,7 still only have 1/2 the power.

500ms of data after n21x2 with channel 1 reconfigured (.png)

How could you keep the same bandwidth but only have 1/2 the band with data?
If the ddc mixing freq was set to 0,then 1/2 the band would be cutoff by the 128Mhz pfb channel filtering.
from the field system we re-issued the command to configure dbe0, channel 0

rdbe_dc_cfg=dbe0,0:2:96.00:0;

We then took 500 ms more of data (without re-initializing anything else.
The plot now shows that channel 1 has power across the entire 64 MHz..the channels 3,5,7 still have 1/2 the power since we didn't touch them.

Cross correlating the 8 bands

160000 samples from each of the 8 bands (4 freq by 2 pols) was used to check the xcorrelation between the 8 bands.
128 lags were computed.

the plots show the correlations between the bands (.ps) (.pdf)

Each frame is a correlation of a single band with all 8 bands,
There are 128 lags/correlation
The left column shows correlations of aoPolA with all bands
The right column shows the correlations of aoPolB with all bands
the dotted red lines separate the correlations
The polA bands correlate with all of the other polA freq bands
the polB bands only correlate with themselves.
the polA bands are all the same frequency band.

Summary:

correlation was seen between ao polB (our lcp,rdbe4) with Ef rcp

we probably have our polarization lines reversed,.

There was no correlation between AO polA (our rcp,rdbe2) and Ef lcp
The spectra for polA only showed power in the first 32 Mhz of each 64 Mhz band
Resending the rdbe_dc_cfg command to a single channel caused power to now be across the entire 64 Mhz band.
the rdbe_dc_cfg=dbe0,0:2:96.00:0; does 2 things

:2: set the decimation to 2. That takes the band from 128Mhz bw to 64 Mhz bw
the ddc mixing freq is set to 96 Mhz
dbe0 is using xbar=7 so this is the 64 Mhz of the lowest freq.
if the mixing freq was set to 0 then the lower 32 Mhz of the 64Mhz band would be truncated giving a 32Mhz output
if the decimation had also been set to 0, then i think you would still get 64 Mhz out.

When doing previous testing, it looks like the default parameters for the rdbe is 0.
correlating the polA band with all of the other bands showed that all of the 4 polA freq bands come from the same frequency range.
so it may be possible to correlate the 32Mhz in rdbe2 .. all you have to do is figure out which of the 32Mhz bands in Ef to use (and the correct sideband :)

My first guess would be to use the 1st or last 32 Mhz of the 256 Mhz band.

processing: x101/211105/fstest.pro

evn fringe test 05nov21