Notes to myself
apr 2003
Notes to refresh what's left of my memory
(or:
"an attempt to postpone the losing battle with entropy").
going
from
continuous to discrete fourier transform
sin(x)/x
polyphase
filterbanks.
gaussian
width
after convolving 2 gaussians.
Jon
Hagen's
acf-> spectra memo (.pdf)
rms voltages needed for
2,3,4, and 5 bit sampling.
Radar
decoding.
clipping a sine
wave
Creating
harmonics
(how strong the various freq terms are)
Harmonics created when clipping a
sine
wave.
Side band harmonics do not
always
move farther out than at the fundamental.
Problems
using
a 90 deg hybrid to convert linear to circular (lbw example)
properities of a gaussian
properteis of sin(x)/x
Computing Trcvr.
Telescope gain
pulsarNotes
Beamwidth,
near/far
field transition
Using
idl
to process ao spectral line data
List
of
cpu
compute servers at AO.
Disc
locations
by cpu
Backing
up of
monitor data.
bitfield
storage
format. converting big endian to little endian.
Ao
Network
Info
- side band harmonics do
not
always
move farther out than at the fundamental. (top)
Suppose you have a tone of Amplitude A and
frequency
w and a side band of amplitude B and frequency (w+delta)
then:
(Aexp(iwt)+Bexp(i(w+delta)*t)2= AAexp(i2wt)+ ABexp(i(2wt
+delta))
+ BB*exp(i(2w+2delta)t)
If B << A then the AB term will be much larger than the BB
term
so the +delta signal will be stronger than the 2*delta.
- Gaussian: (top)
A gaussian centered at x=0 can be defined as:
y1= 1./sqrt(2*pi)*e-.5*(x/sig)^2)
integral
normalized to unity
y2=
e-(x/sig)^2)
y2(x=0)=1.
y2(x=1)=e-1= .367879
Properties are:
| property |
y1 |
y2 |
| maxvalue y(0) |
y1(0)=.398942 |
y2(0)=1. |
| y(sig) |
y1(sig)=.2419707 |
y2(sig)=.367879 |
| integral -inf to inf |
1. |
sqrt(pi)=1.77245 |
FWHM=m*sig (see 1*)
sig=n*FWHM |
m=2.35482
n=.424661 |
m=1.66511
n=.60056118 |
1*
y1=A0/2=A0*e(-.5*x/sig)^2)
sqrt(2*ln(2))*sig=x(hwhm)
FWHM=2*x =2*sqrt(2*ln(2))
y2=1/2=e(-(x/sig)^2)
sqrt(ln(2))*sig=x(hwhm)
FWHM=2*x= 2*sqrt(ln(2))*sig
- sin(x)/x (top)
y=sin(x)/x . If x is defined in units of the first null, then use
sin(pi*x)/(pi*x)
| property |
|
| maxvalue : y(0) |
1. |
| minvalue :y |
y=-.2172336, x=4.4934100.. |
| zero crossings |
pi*n n=1.. |
| peaks,min |
when x=tan(x) (more
info)
for large x :close to x=(2n+1)pi/2 |
| x when y=.5 |
x=1.8954950 |
integral between 1st
nulls (-pi,pi) |
integral: 9.8696191 |
rms voltages needed
for
2,3,4,
and 5 bit sampling. (top)
The optimum
voltage
levels
for 2, 3, , and 5 bit sampling were computed using the threshold
levels
from fred schwab. To compute the rmsVolts I just took
PktoPkVolts/Nlevels
* sigmaLevels. This ignores the problem of whether or not the levels
are
centered on 0 volts or not.
Rms Volts vs Nbits
| Nbits |
level threhsold
(in sigmas) |
Sigma (1./level)
(in levels) |
rms (Volts) assuming
2V PkToPk A/D |
rms (Volts) Assuming
5V PkToPk A/D (ri) |
| 2 |
.99568 |
1.004 |
0.502 |
1.255 |
| 3 |
.58601 |
1.706 |
0.427 |
1.067 |
| 4 |
.33520 |
2.983 |
0.373 |
0.932 |
| 5 |
.18814 |
5.315 |
0.332 |
0.830 |
Compute
Trcvr. (top)
The receiver temperature is computed using a hot
and
cold load at the input. let:
- T1 and T2 be the temperatures of these loads.
- gamma is the voltage reflection at the input to the horn. let
G2=(1-gamma^2).
It will be the fraction of the input power that gets into the
device
- alpha is the loss in the omt, Tomt is the temperature of the
omt.
(1-alpha)
of the entering temp will be passed through. (1-alpha)*Tomt will
be
contributed
by the omt temp.
- Tamp is the amplifier temperature. Trcv is normally Tamp
+
alpha*Tomt.
If we measure the output power ratio on load1 and load2:
Y=Pwr2/Pwr1
y=(Tamp + alpha*Tomt + (1-gamma^2)(1-alfa)T2)/(Tamp + alpha*Tomt + (1-gamma^2)(1-alpha)T1)
Solving for Tamp gives:
Tamp(1-Y)=Y(alpha*Tomt + (1-alpha)(1-gamma^2)*T1) - (alpha*Tomt + (1-alpha)(1-gamma^2)T2)
Tamp= (1-alpha)(1-gamma^2)*(T2-Y*T1)/(Y-1) - alpha*Tomt
If you assume that alpha,gamma are equal to zero, then you will get
a
Tamp
that is higher than it really is.
Telescope gain
(top)
Units:
k - boltzman's constant
K - deg Kelvin
Jy - Jansky
m - meters
w = watts
J - Joules
Hz- hertz
Ae - effective area of telescope
G - gain
T - Tsys
item
|
value
|
notes
|
k: Boltsmans' Constant
|
1.38e-23 Joule/degK
-228.6 db/degK
-198.6 dbm/degK
|
convert TempK to energy
Joules
|
Jy: Jansky
|
1e-26 W/(m^2*Hz)
-260 db/(m^2*Hz)
-230 dbm/(m^2*Hz)
|
flux density per hz
|
k/Jy:boltzmans constant/Jy
(J/K)/(w/(m^2*Hz)
|
1380 (m^2/K) for 1 Jy
dual
pol?
2760 (m^2/K) for 1 Jy single pol?
|
A telescope with Ae=2760 m^2
give:
Ae/2760 = 1K for a 1 Jy source
|
Gain 1K/Jy = 2760m^2
|
|
|
G = K/Jy = Ae/2760.
|
|
|
G/T = (Ae/T)/2760.
|
|
SEFD=T/G = (T/Ae) * 2760.
|
|
|
nrao
Brightnes and Flux density
|
|
|
- 1 K/Jy is 2760 m^2
- for a uniformilly illuminated circular appeture:
hpbw=1.02*lambda/D
- if plane wave with energy power density S hit telescope. The
telescope
extracts Pe energy then the effective area of the telescope is
Ae=Pe/S
- The appeture efficiency is Na=Ae/Ag where Ag is the
geometric
area.
- Gmax=4*pi*Ae/lambda^2
- Ae*OmegaA=lambda^2 where OmegaA is the beam solid angle
(integrated
over
4pi steradian).
- OmegaMb is the main beam solid angle. Defined to be the solid
angle
that
has the same area as the mainbeam down to the first nulls.
Pulsars
- dm smearing across a band
- tmSmMillisecs= (202/cfrMhz)^3 * dm *BwMhz
- optimum channel width for incoherent processing
- 1/channelwidth = dmsmearing across the channel..
- or
- 1./(bw/nchan) = dmSmearingbw/nchan
- nchan/dmSmearingBw = bw/nchan
- nchan= sqrt(bw*dmSmearingBw) optimum.. or
- nchan=sqrt(bwMHz^2 * (202/cfrMhz)^3 * dm * 1e3)
svn
repository structure
- How directories are used:
- trunk
- main developement goes on here
- branches
- Different versions stores with possible updates needed to
fix
a version
- tags:
- snapshots of a particular version. It is frozen. Might be
easier to role back
- /share/megs/phil/svnrepos
:
the repository
- Cima
- branches
- tags
- V3.0.10 (normal)
- V3.1.03 (smart)
- trunk
- idl - holds phil's idl code
- branches
- tags
- trunk
- idl directories under /pkg/rsi/local/libao/phil
- pdev - the datataking system for pdev. excludes jeff's code
- branches
- tags
- rev-1026/ before parallactic angle changes 16feb09
- rev-879/ before psrfits changes.
- trunk
- Readme
- aosoft/
- linux.sh
- linuxfiles/
- pdev/
- vw - vxWorks datataking code from /home/phil/vw/
- branches
- tags
- trunk
- directory under /home/phil/vw/
- vwTcl - tcl procedures for vxWorks. from /home/online/vw/Tcl
- branches
- tags
- trunk
- code from /home/online/vw/Tcl
svn commands
- list contents of repository:
- svn list file:///share/megs/phil/svnrepos
- copy a new set of files into the repository
svn notes CIMA
- Locations:
- file:///share/megs/phil/svnrepos/Cima ..
repository
- Online cima versions:
- /home/cima/Software/XX XX= Mocksp, Svnwork
- Working directories
- /home/cima/svn/Cima/XX XX =
svnwork,mocksp, etc..
- svn list file:///share/megs/phil/svnrepos/Cima
- checkout to a working directory
- cd /home/cima/svn/Cima/ ; mkdir mocksp
- svn -r revision checkout
file:///share/megs/phil/svnrepos/Cima/trunk
/home/cima/svn/Cima/mocksp
- Make Svnwork a new cima version release (eg. Mocksp)
- make sure svn repository uptodate for svn/Cima/svnwork svn
-u
status
- record svn version number
- cd /home/cima/svn/Cima/svnwork
- make install
- moves most recent updates to Svnwork
- does a cmake install aon Frontend,Exectuive,
tcl_utilities,
then copies the version files back to
Cima/svnwork/Software.
- add /home/cima/bin to front of path variable (to find cmake)
- Copy online version Svnwork to new version
- cima_copy Svnwork Mocksp
- answer Questions: mocksp 3.2.00
- Make a branch for a new version in the svn repository.
- svn copy file:///share/megs/phil/svnrepos/Cima/trunk
file:///share/megs/phil/svnrepos/cima/branches/mocksp -m
"creating mocksp branch from trunk revision (revision
number from
status above)"
- checkout the mocksp version
- cd /home/cima/svn/
- svn checkout
file:///share/megs/phil/svnrepos/Cima/trunk/Software
mocksp
SVN notes pdev
home_~phil