[evlatests] The 128 MHz comb zoo ...

[Dan Mertely]

In search of the not so elusive 128 harmonic
source, Ed Menne and I brought a spectrum analyzer
(SA) up into ea10 @ BN8 this morning.  We chose as a
target the 23rd harmonic of 128 @ 2944 MHz in
S-band.  (The 128*23 harmonic was very strong in the
RFI sweep, and, being an odd harmonic, eliminates the
512 MHz reference as a source.)

We had the operator (JC) set-up the system to
"default S-band", which sets the L301-1 (for AC)
to 13184 MHz, and L301-2 (for BD) to 13440 MHz.
These 2 LO frequencies were verified at the antenna
using the spectrum analyzer.

We 1st checked at the output of the Front End, to see
if the 2944 MHz harmonic was being radiated out from
some hardware then making it into the feed.  The
answer is no.  There was NO 2944 signal coming out
of the FE.  We set the spectrum analyzer to a very
narrow RBW in order to maximize sensitivity (reduce
noise).  The excess gain of the FE put the viewed
noise level (NL) at -100 dBm (the  SA NL was intrinsically
-120 dBm at this RBW).  Just as a sanity check, we
looked for the XM radio signal down in the 2320 - 2345
MHz range, and were able to see it.  So, the 128*N
signal is not coming out of the FE.

We then checked the L305 reference generator module
output and saw the 2944 MHz harmonic @ -87 dBm on the
128 MHz output (we tapped in on the spare output of
the rack-mounted splitter).  (We didn't check the power
level of the 128 MHz fundamental.)

We then checked the L301-2 output, verifying the set-up
frequency of 13440 MHz (as mentioned earlier), and
finding the 2944 MHz harmonic in the output @ -59 dBm.

Using 13440 - 2944 = 10496 MHz for the conversion, we
then checked the output of the T304B LSC converter, and
saw the 128*23 signal at the expected 10496 MHz.  The
power level at this point was -87 dBm.  As a reference
check, we tuned "up" (reverse spectrum) around 600 MHz
to look for the XM radio signal in the 1st IF passband,
and saw it clearly around 11104 MHz at -80 dBm.

Finally, we checked the output of the L301-1 LO.  The
main output was at 13184 MHz @ +7 dBm, and the leaky/
sneaky 128*23 (2944 MHz) harmonic was also seen in the
output at -60 dBm.


So:

2944 MHz is seen at the L305 output, but low.

2944 MHz is seen at the L301 outputs, but down 67 dBc.

2944 MHz is seen in the IF passband at the T302 output,
within 7 dB of the strength of the XM satellite (one
of the strongest sky sources in S-band).

Since in the correlated RFI sweep data of 18 March, 2011
the harmonic isn't clearly seen above C-band (except for
one occurrence in Ka band at 128*230=29440), I (with
input from Rob Long) would suspect some problem in the
T302 mixer scheme.  (With some other cause for the single,
strong harmonic detection in Ka band?)

I leave it to LO/IF to find the mechanism.

-Mert




Rick Perley wrote:
>     I've labored hard to try and understand the characteristics of the 
> 128 MHz comb -- largely without success.  The characteristics vary by 
> band and even by frequency within a band. 
>     But a few general characteristics seem clear. 
> 
>     A)  Harmonics of 128 MHz are seen in L, S, and C bands.  Only a 
> single harmonic is seen in X-band -- at 8192 MHz.  This one probably has 
> a separate origin within the T304. 
> 
>     B)  The odd harmonics are usually, but not always, the strongest.  
> Not all odd harmonics are visible, and some even harmonics are very 
> strong. 
> 
>     C) There is no spatial correlations for these harmonics -- the 
> 'loudest' antennas (which vary by band -- see below) are randomly 
> distributed.  This clearly indicates an internal origin. 
> 
>     D) At L-band, antennas 9, 10, 14, 20, 24 and 26 appear to have no 
> 128 MHz harmonics.  All the others have varying degrees of the  tone. 
> 
>     E)  As S-band, the harmonics are generally much stronger than in 
> either L or C bands.  But the situation varies remarkably (or 
> depressingly, depending on your POV) depending on which harmonic we look 
> at.  For example, for the 21st harmonic (2688), antennas 6 and 7 are 
> 'clean'.  But for the 23rd harmonic (2944), antennas 16 and 26 are 
> 'clean', while 7, 8, 9, 24, and 28 are strong.  For the 25th harmonic 
> (3200 MHz), the situation is again different:  antennas 14 and 26 are 
> completely clean, antennas 16 and 24 give only very weak birdies.  For 
> the 27th harmonic, antenna 10 and 26 are now 'clean', while 8 is 
> especially bad.  (I gave up trying to characterize the harmonics  higher 
> than this at this stage ...)
> 
> 
>     F)  At C-band, the situation is again different ... <audible 
> sigh>.     The 32nd harmonic (4096 MHz) is generally quite weak, but is 
> strong on antennas 10 and 12.   The same pair gives a strong harmonic at 
> 4992 (39th).   But at 5504 MHz, (43rd harmonic), only antenna 10 is 
> bad.   But (sadly), the situation deteriorates at 5888 MHz, where 
> antennas 10, 14, 20 and 24 are all very bad, while 15, 27 and 28 are good. 
> 
>     In summary, multiples of 128 MHz show up everywhere at the lowest 
> three bands, but it is very difficult to identify particular 
> commonalities. 
> _______________________________________________
> evlatests mailing list
> evlatests at listmgr.cv.nrao.edu
> http://listmgr.cv.nrao.edu/mailman/listinfo/evlatests