[evlatests] C-band Sensitivity Oscillation

[Robert Hayward]

Bob D.,

Although WL Gore still sells radome material (3T20), they really don't 
seem to have any serious EM expertise left in their company anymore. 
Alas they make all their money with their architectural material line. I 
received the following email from Gore to a query I made concerning 
their RA7906 and 3T20 fabrics back in Feb 2005...

"We really do not have much information on the differences of the two 
product throughout different frequencies. Gore a few years back really 
stepped away from the Microwave fabrics business and lost most of our 
technical resources. At that time we decided to become a material based 
business model and not a technical support business model. For that 
reason we have always relied on the customer to determine the 
requirements of their application by sampling them material for testing. 
What we were finding is that each requirement was slightly different and 
it was becoming very costly to have the testing preformed in house or by 
outside firms. In addition because of the slight differences many 
customers were asking for us to preform the testing to their specific 
frequency. Again the business would not support this type of technical 
resource."

In June 2007, when Jim Ruff inquired about buying RA7906 material for 
the Ka & Ku-Band horns, he was informed that Gore were no longer making 
our favorite RA7906 radome material, but were accepting orders for the 
last 44 meters (x 45" wide) that they still had on hand. So we bought 
all they had to complete the remaining feeds on the EVLA for $9000 (18 
meters were needed to handle X, Ku, Ka plus extra for K & Q-Bands). So 
we have a fair bit of spare material, but certainly not enough to 
replace the ESSCOLAM-10 used on the C-Band feeds.

The VLBA uses ESSCOLAM-8 (and maybe even ESSCOLAM-6) on their L, S & 
C-Band. The ESSCOLAM-8 product line is now now obsolete and was replaced 
with ESSCOLAM-10 which has a hydrophobic coating (water on the radome 
surface collects in beads rather than in sheets, and since water has a 
high dielectric constant and a very high loss tangent, even a thin film 
of water on the surface of a radome can cause rather large signal 
attenuation).

As best we can tell, the 24 mil thick ESSCOLAM-10 that we're using on 
the EVLA L, S and C-Band feeds has a dielectic loss of 3.0 (the ESCCOLAM 
8 we use on the VLBA was 2.8) and the loss tangent is 0.009. The 15 mil 
RA7906 used on the EVLA X, Ku, K, Ka and Q-Band (as well as W-Band on 
the VLBA) has a dielectric constant of 1.55 and a loss tangent of 
0.0005. So the RA7906 is much better microwave-wise than 3T20 but we 
might be getting kind of close to its mechanical limits for a big feed.

There is some question as to whether the VLBA C-Band radome has a 24 mil 
thickness or a 15 mil. I've found old purchase orders for 5ft x 5ft x 24 
mil material and 1.5ft x 1.5ft x 15mil material, which sounds 
suspiciously like an L vs. C-Band feed dimension. Vivek says he does not 
believe the VLBA exhibits any ripple similar to the one seen on the 
EVLA. I don't know enough about the size of the linear VLBA feed versus 
the compact EVLA feed, but I suspect the VLBA feed has a somewhat 
smaller path length and thus, if it had a ripple, its period would be 
somewhat longer in period and might be hard to distinguish over a 500 
MHz bandwidth. But, it might be worth investigating whether we can get a 
thinner ESSCOLAM-10 material for the EVLA C-Bands (i.e., 15 vs 24 mil).

It turns out that Ed Szpindor ordered some 3T20 radome material from 
Gore back in the early days of the EVLA Project hoping to use it for the 
high frequency receivers (about 8m x 45" x 18mil). When we started using 
it on some of our early K-Band receivers for the EVLA, I discovered that 
it added a fair bit of loss (about 5K) and was just plain awful at 86 
GHz (added about 15K). It was obviously not prime radome material for 
any receiver above X-Band. According to the Gore salesman (before they 
stopped making RA7906)...

"We have tested the 3T20 in a 4.5 meter antenna at 14.250 Ghz frequency 
and saw no real reduction in it's performance. At higher frequencies I 
would expect higher losses. I always let people know of this when I 
discuss the new product but their are some advantages with the new 
material. One is the price, two is the width, three is the durability, 
fourth is that it can be heat welded to create a seam."

Alas, they do not provide any RF information (dielectric constant or 
loss tangent) and I have had no luck on finding anybody who has tested 
it (the CSIRO were the ones who had measured the RA7906 performance for 
them). But it might be worth trying to use at C-Band. I suggest the next 
C-Band that goes out have a 3T20 radome installed. We might want to try 
some measurements with ESSCOLAM-10, as well as no radome at all, for 
comparative purposes. Obviously, doing this in dry weather would be a 
good idea.

-Bob H.


Bob Dickman wrote:
> I wonder whether Gore makes a similar fabric these days.  When I worked with
> them, there were a variety of essentially transparent fabrics.
> 
> Bob DIckman
> 
> 
> On 7/28/08 5:20 PM, "Robert Hayward" <rhayward at nrao.edu> wrote:
> 
>> I made the following reply on Friday to Rick's evlatest message, only to
>> find he was the only one to get it. It covers some of the same ground as
>> Mert's message today but adds a few more things...
>>
>> -Bob
>>
>> -------- Original Message --------
>> Subject: Re: [evlatests] C-band Sensitivity Oscillation
>> Date: Fri, 25 Jul 2008 10:07:13 -0600
>> From: Robert Hayward <rhayward at aoc.nrao.edu>
>> Organization: NRAO
>> To: Rick Perley <rperley at nrao.edu>
>> References: <48891708.7030906 at aoc.nrao.edu>
>>
>> Look at Figure 11 in the EVLA Memo 95 "Design, Prototyping and
>> Measurement of EVLA C-Band Feed Horn" which you can get from here...
>>
>> http://www.aoc.nrao.edu/evla/geninfo/memoseries/evlamemo95.pdf
>>
>> The return loss ripple with the ESCCOLAM-10 radome in place looks pretty
>> similar - about 11 to 12 cycles between 4.5-5.5 GHz. Note that it is at
>> its worst in the 4-5 GHz region. Still, the return loss is below -25 dB,
>> at least across 5-8 GHz, so the feed is not wildly out of spec. The
>> ripple might get worse if there's moisture on the membrane (or the
>> vacuum window).
>>
>> We might get better results using a radome made from Goretex RA7906
>> material, similar to what we use at K, Ka & Q-Band, and which we intend
>> to use at X & Ku. Unfortunately, the last I heard was that W.L. Gore no
>> longer sells it and we bought the last of their stock (a roll 44 meters
>> long by 45 inches wide - I loved their mix of units - for just under
>> $10,000) to complete the EVLA Project with enough spare material to take
>> care of a reasonable amount of lossage (although a hail storm would
>> certainly put us out of business).
>>
>> Before we get too worried, we should wait and see what the passbands
>> look like when we have a few more wideband OMT's on the Array.
>>
>> The good news is that this ripple is a lot less of a problem then the
>> higher frequency ripple seen from reflections with the subreflector
>> (something like 10 MHz or so) and the old 3 MHz ripple from the circular
>> waveguide system which can cause confusion when looking for your typical
>> narrow, weak spectral line. What we'd need to be assured of is that this
>> 90 MHz ripple is stable. For example, would a radome that isn't
>> stretched very taut cause the ripple frequency to change enough to
>> affect the data with changes in elevation or when it is windy?
>>
>> -Bob
>>
>>
>> Rick Perley wrote:
>>>     We have long noted that IF#1 at C-band had slightly (~5%) poorer
>>> sensitivity than IF#2 for EVLA antennas.   Recent tests showed that this
>>> is not an effect due to the IF -- swapping frequencies caused the
>>> sensitivity difference to change.  Furthermore, the effect is not seen
>>> at X-band, nor for the VLA antennas at C-band.
>>>
>>>     Emmanuel ran a test last night to clarify the picture.  He tuned the
>>> array from 4.80 through 5.05 GHz, in 5 MHz steps, using 6MHz continuum.
>>> The sensitivity for each frequency was derived via the correlation
>>> coefficients. 
>>>
>>>     We find a remarkable sinusoid oscillation in sensitivity, for all
>>> IFs, for all EVLA antennas, over the entire bandwidth surveyed.  The
>>> amplitude of the oscillation is about 10% (in Tsys/effic), the period is
>>> close to 90 MHz.  If due to a standing wave, the length is 1.7 meters --
>>> about the length of the C-band horn?
>>>
>>>     We stumbled upon this because the default frequencies -- 4885 and
>>> 4835 MHz, just happen to lie on the peak and trough of the standing wave...
>>>
>>>    
>>> _______________________________________________
>>> evlatests mailing list
>>> evlatests at listmgr.cv.nrao.edu
>>> http://listmgr.cv.nrao.edu/mailman/listinfo/evlatests
>>>
>>>
>>
>> _______________________________________________
>> evlatests mailing list
>> evlatests at listmgr.cv.nrao.edu
>> http://listmgr.cv.nrao.edu/mailman/listinfo/evlatests
> 
>