Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://www.naic.edu/~phil/hardware/pdev/newmix_19sep07_1digSpc.html
Äàòà èçìåíåíèÿ: Fri Sep 21 18:07:21 2007
Äàòà èíäåêñèðîâàíèÿ: Sat Dec 22 23:22:01 2007
Êîäèðîâêà:
Ïîèñêîâûå ñëîâà: ï ï ï ï ï ï ï ï ï ï ï ï ï ï ï ï ï ï ï ï ï ï ï ï ï ï

pdev_birdies_25jun07

Noise spectra with and without the mixer chassis.

19sep07

The red noise source was used to take spectra with and without the new mixture chassis.
The setup was:

Box 103 polB high band was used. All spectra were 30 seconds of integration, 170 Mhz sampling and 8192 channels.
The red noise source was used.
The spectra through the mixer were centered at 325 Mhz in the noise spectra.
The non mixer spectra used a 60 Mhz low pass filter between the noise src and the digitizer. Only a single digitizer was used. The unused digitizer (of the complex pair) was set to 0.

the single digitizer spectra are symetric. We can check if the different digitizers are acting correctly.

The two measurements used different portions of the noise source spectrum (325 Mhz and baseband).
On the spectrum analyzer we saw a 1db slope in the noise src output from 0 to 70Mhz. The power was flat (on 1db/division scale) below 2 Mhz.

The plots show the spectra with and without the mixers (.ps) (.pdf):

Page 1: The top plot shows the full bandpass. The vertical scale is in db's.

The black plot is from the mixer chassis. It uses both I and Q digitizers.
Red/Green are the spectra without the mixer. They have the other digitizer of the complex pair set to 0.
The bottom plot is a blowup in freq,amplitude. The two spectra without the mixer track each other pretty well.
The large scale difference between the red/green and the black trace are probably the differences in the noise source output (at the different frequencies).
The 15 Mhz birdies in the black plot is the LO leaking through the mixer (325Mhz - 170*2=15Mhz). These are not seen in the red/green plots since they don't go through the mixer chassis.

Page 2: Is a further blowup showing +/- 10 Mhz about the center of each spectra.

The mixer spectra shows an increase in the spectra below 2Mhz. This is not seen in the spectra that do not go through the mixer chassis.
The red/green spectra have dips/bumps around 2 Mhz. These were not seen in the noise source spectrum on the spectrum analyzer (at least at 1db/division...)

Conclusions:

The mixer spectra increases below 2 Mhz. It is not seen when we bypass the mixer chassis.

We looked at the output of the mixer chassis on the spectrum analyzer and saw a similar increase.

There are a number of isolation amplifiers after the mixer to allow us to split the baseband signal between ealfa and palfa. They could be causing the problem.
The rise below 2 Mhz is going to be in the center of the bandpass. If we can't get rid of it, we need to check to see that it is stable so that divides out when doing on/offs.
Now i remember why we wanted to go to digital downconversion with the interrim correlator and the wapps....

The single digitizer spectrum (no mixer) have some unexplained dips around 2Mhz. The dual spikes below 1 Mhz (about 170 Khz) have been seen before and are probably coming from some components inside the spectrometer.

processing: x101/070819/testpdev.pro

<- page up
home_~phil