Monday, August 28, 2017

Monitoring the "CT" MedFER beacon from "Eclipse land"


Figure 1:
The MedFER beacon and vertical, tophatted
antenna on the metal roof of my house, attached
to an evaporative ("swamp") cooler.
Click on the image for a larger version.
I must admit that I was "part of the problem" - that is, one of the hordes of people that went north to view the August 21, 2017 eclipse along its line of totality.  In my case I left my home near Salt Lake City, Utah on the Friday before at about 4AM, arriving 4 hours and 10 minutes later - this, after a couple of rest and fuel stops.  On the return trip I waited until 9:30 AM on the Wednesday after, a trip that also took almost exactly 4 hours and 10 minutes, including a stop or two - and I had no traffic in either case.

This post isn't about my eclipse experiences, though, but rather the receiving of my "MedFER" beacon at a distance of about 230 miles (approx. 370km) as a crow flies.

What's a MedFER beacon?

In a previous post I described a stand-alone PSK31 beacon operating just below 1705 kHz at the very top of the AM broadcast ("Mediumwave") band under FCC Part 15 §219 (read those rules here).  This portion of the FCC rules allow the operation of a transmitter on any frequency (barring interference) between 510 and 1705 kHz with an input power of 100 milliwatts using an antenna that is no longer than 3 meters, "including ground lead."  By operating just below the very top of the allowed frequency range I could maximize my antenna's efficiency and place my signal as far away from the sidebands and splatter of the few stations (seven in the U.S. and Mexico) that operate on 1700 kHz.
Figure 2:
Inside the loading coil, showing the variometer, used to fine-
tune the inductance to bring the antenna system to
resonance.  This coil is mounted in a plastic 5-gallon
bucket, inverted, to protect it from weather.

As described in the article linked above, this beacon uses a Class-E output amplifier which allows more than 90% of its DC input power to be delivered as RF, making the most of the 100 milliwatt restriction of the input power.  To maximize the efficiency of the antenna system a large loading coil with a variometer is used, wound using copper tubing, to counteract the reactance of the antenna.  The antenna itself is two pieces:  A section, 1 meter long, mounted to the evaporative cooler sitting on and connected to the metal roof of my house and above that, isolated from the bottom section is an additional 2-meter long section that is tophatted to increase the capacitance and reduce the required amount of loading inductance to improve overall efficiency.

As it happens, the antenna is mounted in almost exactly the center of the metal roof of my house so one of the main sources of loss - the ground - is significantly reduced, but even with all of this effort the measured feedpoint resistance is between 13 and 17 ohms implying an overall antenna efficiency of just a few percent at most.

Figure 3:
The tophatted vertical antenna, loading coil and transmitter, looking up
from the base.  In the extreme foreground along the left side of the
picture can be part of the weather-resistant metal box that
contains the transmitter.
Click on the image for a larger version.
Originally intended only as a PSK31 beacon, I later added the capability of operating on 1700 kHz using AM and being able to do on/off keying of the carrier at the original "1705" kHz PSK31 frequency, permitting the transmission of Morse code messages.  For the purpose of maximizing the likelihood of the signal being detected, this last mode - Morse - I operate using "QRSS3", a "Slow" Morse sending speed where the "dit" length of the characters is being transmitted is 3 seconds - as is the space between character elements - and a "dah" and the space between characters themselves is 9 seconds.

Sending Morse code at such a low speed allows sub-Hz detection bandwidths to be used, greatly improving the rejection of other signals and increasing the probability that the possibly-minute amount of energy reaching the receive antenna may be detected.

Detecting it from afar:

Even though this beacon had been "received" as far away as Vancouver, BC (about 800 miles, or 1300 km) using QRSS during deep, winter nights, I was curious if I could hear it during a summer night near Moore, ID at that 230 mile (370km) distance.  Because we were "camping" in a friend's yard, we (Ron, K7RJ and I) had to put up an antenna to receive the signal.

The first first antenna that we put up received strong AC mains-related noise - likely because it paralleled the power line along the road.  Re-stringing the same 125-ish feet (about 37 meters) of antenna wire at a right angle to the power line and stretching out a counterpoise along the ground got better results:  Somewhat less power line noise.  It was quickly discovered that I needed to run both the receiver and the laptop on battery as any connection to the power line seemed to conduct noise into the receiver - probably a combination of noise already on the power line as well as the low-level harmonics of the computer's switching power supply.

I'd originally tried using my SDR-14 receiver, but I soon realized that between the rather low signal levels being intercepted by the wire - which was only about 10 feet (3 meters) off the ground - and the relative insensitivity of this device, I wasn't able to "drive" its A/D converter very hard, resulting in considerable "dilution" of the received signals due to quantization noise.  In other words, it was probably only using 2-4  bits of the device's 14 bit A/D converter!

I then switched to my FT-817 (with a TCXO known to be accurate to better than one part-per-million) which had no troubling "hearing" the background noise.  Feeding the output of the '817 into an external 24 bit USB sound card (the sound card input of my fairly high-end laptop - as with most laptops - is really "sucky") I did a "sanity check" of the frequency calibration of the FT-817 and the sound card's sample rate using the 10 MHz WWV signal and found it to be within a Hertz of the correct frequency and then re-tuned the receiver to 1704.00 kHz using upper-sideband.  It had been several years since I'd measured the precise frequency of my MedFER beacon's carrier, last being observed at 1704.966 kHz, so I knew that it would be "pretty close" to that value - but I wasn't sure how much its crystal might have drifted over time.

For the signal analysis I used both "Spectrum Lab" by DL4YHF (link here) and the "Argo" program by I2PHD (link here).  Spectrum Lab is a general-purpose spectral analysis program with a lot of configurability which means that there are a lot of "knobs" to tweak, but Argo is purposely designed for modes like QRSS using optimized, built-in presets and it was via Argo that I first spotted some suspiciously coherent signals at an audio frequency of between 978 and 980 Hz, corresponding to an RF carrier frequency of 1704.978 to 1704.980 kHz - a bit higher than I'd expected.

As we watched the screen we could see a line appear and disappear with the QSB (fading) and we finally got a segment that was strong enough to discern the callsign that I was sending - my initials "CT".

Figure 4
An annotated screen capture of a brief reception, about 45 minutes after local sunset, of the "CT" beacon using QRSS3 with the "oldest" signals at the left.  As can be seen, the signal fades in so that the "T" of a previous ID, a complete "CT" and a partial "C" and a final "T" can be seen on the far right.  Along the top of the screen we see that ARGO is reporting the peak signals to be at an audio frequency of 978.82 Hz which, assuming that the FT-817 is accurately tuned to 1704.00 kHz indicates an actual transmit frequency of about 1704.979 kHz.

As we continued to watch the ARGO display now and again we could see the signal fade in and out and be occasionally clobbered by the sidebands of an AM radio station on 1700 kHz - at least until something was turned on in a nearby house that put interference everywhere around the receive frequency.

The original plan:

The main reason for leaving the MedFER beacon on the air during the eclipse and going through the trouble of setting up an antenna was to see if, during the depth of the eclipse, its signal popped up, out of the noise - the idea being that the ionospheric "D" layer would disassociate in the temporary darkness along the path between my home where the eclipse would attain about 91% totality and the receive location within the path of totality, hoping that its signal would emerge.  In preparation for this I set up the receiver and the ARGO program to automatically capture - and then re-checked it about 5 minutes before totality.

Unfortunately, while I'd properly set up ARGO to capture, I'd not noticed that I'd failed to click on the "Start Capturing" button in ARGO and the computer happily ran unattended until, perhaps, 20 minutes after totality, so I have no way of knowing if the signal did pop up during that time.  I do know that when I'd checked on it a few minutes before totality there was no sign of the "CT" beacon on the display.

In retrospect, I should have done several things differently:
  • Brought a shielded "H" loop that would offer a bit of receive signal directionality and the ability to reject some of the locally-generated noise and would have saved us the hassle of stringing hundreds of feet of wire through trees.  Some amplification with this loop would also have helped the SDR-14 work properly.  Alternatively, a simple active whip (such as a PA0RDT "mini-whip") could have been built and used, its location chosen for lowest noise pick-up.
  • Actually checked to make certain that the screen capture was activated!
  • Record the entire event to an uncompressed audio (e.g. ".WAV") file so that it could be re-analyzed later.
 Oh well, you live and learn!

P.S.  After I returned I measured the carrier frequency of the MedFER beacon using a GPS-locked frequency reference and found it to be 1704.979 kHz - just what was measured from afar!

[End]

This information stolen from ka7oei.blogspot.com

Tuesday, August 15, 2017

Analyzing "fake" solar eclipse viewing glasses - how good/bad are they?

Note:  Please read and heed the warnings in this article.

About a month and a half ago I ordered some "Eclipse Viewing Glasses" from Amazon - these being those cardboard things with plastic filters.  When I got them, I looked through them and saw that they were very dark - and in looking briefly at the sun through them they seemed OK.
Figure 1:
The suspect eclipse viewing glasses.
These are the typical cardboard frame glasses with very dark plastic lenses.
Click on the image for a slightly larger version.

I was surprised and chagrined when, a few days ago, I got an email from Amazon saying that they were unable to verify to their satisfaction that the supplier of these glasses had, in fact, used proper ISO rated filters and were refunding the purchase price. This didn't mean that they were defective - it's just that they couldn't "guarantee" that they weren't.

I was somewhat annoyed, of course, that this had happened too soon prior to the event to be able to get some "proper" glasses, but I then started thinking:  These glasses look dark - how good - or bad - are they?

I decided to analyze them.

WARNING - PLEASE READ!

What follows is my own, personal analysis of "potentially defective" products that, even when used properly, may result in permanent eye damage.  This analysis was done using equipment at hand and should not considered to be scientifically rigorous or precise.

DO NOT take what follows as a recommendation - or even an inference - that the glasses that I tested are safe, or that if you have similar-looking glasses, that they, too, are safe to use!

Figure 2:
The 60 watt LED light used for testing.  This "flashlight" consists of
a 60 watt Luminus white LED with a "secondary" lens placed in front of it.
The "primary" lens (a 7" diameter Fresnel) used to collimate the beam
was removed for this testing.
Click on the image for a larger version.
This analysis is relevant only the glasses that I have and there is no guarantee that glasses that you have may be similar.  If you choose to use similar glasses that you might have, you are doing so at your own risk and I cannot be held liable for your actions!


YOU HAVE BEEN WARNED!

White Light transmission test:

I happen to have on hand a homemade flashlight that uses a 60 watt white LED that, when viewed up close, would certainly be capable of causing eye damage when operating at full power - and this seemed to be a good, repeatable candidate for testing.  For measuring the brightness I used a PIN photodiode (a Hammatsu S1223-01) and relative measurements in intensity could be ascertained by measuring the photon-induced currents by measuring that current with and without the filter in place.

Using my trusty Fluke 87V multimeter, when placed 1/4" (about 6mm) away from the light's secondary lens I consistently measured a current of about 53 milliamps - a significantly higher current than I can get from exposing this same photodiode to the noonday sun.  In the darkened room I then had the challenge of measuring far smaller current.

Switching the Fluke to its "Hi Resolution" mode, I had, at the lowest range, a resolution of 10 nanoamps - but I was getting a consistent reading of several hundred nanoamps even when I covered the photodiode completely.  It finally occurred to me that the photodiode - being a diode - might be picking up stray RF from radio and TV stations as well as the ever-present electromagnetic field from the wires within our houses so I placed a 0.0022uF capacitor across it and now had a reading of -30 nanoamps, or -0.03 microamps.  Reversing the leads on the meter did not change this reading so I figured that this was due to an offset in the meter itself so I "zeroed" it out using the meter's "relative reading" function.  Just to make sure that the all of the current that I was measuring was from the front of the photodiode I covered the back side with black electrical tape.
Figure 3:
A close up of the S1223-01 photodiode and capacitor in front of the LED.
The bypass capacitor was added to minimize rectification of stray RF
and EM fields which caused a slight "bias" in the low-current readings.
Click on the image for a lager version.

I then placed the plastic film lens of the glasses in front of the LED, atop the flashlights secondary lens - and...

It melted.

Drat!

Moving to a still-intact "unmelted" portion of the lens I held it against the photodiode this time, placing it about 1/4" away from the LED as well and got a consistent reading of 0.03-0.04 microamps, or 30-40 nanoamps.  Re-doing this measurement several times, I verified the consistency of these numbers.

Because the intensity of the light is proportional to the photodiode current, we can be reasonably assured that the ratio of the "with glasses" and "without glasses" currents are indicative of the amount of attenuation afforded by these glasses, so:

53mA = 5.3*10E-2 amps - direct LED, no glasses
40nA = 4.0*10E-8 amps - through the glasses

The ratio is therefore:

5.3*10E-2 / 4.8*10E-8 = 1325000

What this implies is that there is a 1.325 million-fold reduction in the brightness of the light. Compare this with #12 welding glass which has about a 30000 (30k)-fold reduction of visible light and the absolute minimum that is considered to be "safe" for direct viewing while #14 offers about a 300000 (300k)-fold reduction.  According to various sources (NASA, etc.) a reduction of 100000 (100k)-fold will yield safe direct viewing.  The commonly available #10 welding glass offers only "about" a 10000 (10k)-fold reduction at best and is not considered to be safe for direct solar viewing.
Figure 4:
The typical spectral output of a "white" LED (blue line) and
a typical silicon PIN photiode (black line.)  The distinct peak
is from the internal blue LED while the "yellow" Ce:YAG
phosphors emit longer wavelengths to produce a "white" light.
As can be seen, the sensitivity of the photodiode increases
with longer wavelengths while the spectral output of a white
LED drops.
Click on the image for a larger version.

This reading can't be taken entirely at face value as this assumes that the solar glasses have an even color response over the visible range - but in looking through them, they are distinctly red-orange.  What this means is that the spectrum of the white LED - which is mostly red-yellow and some blue (because white LEDs use blue LEDs and a phosphor to produce the rest of the spectrum) and very little infrared - means that we are doing a bit of apples-oranges comparison.

In addition to this, the response of the photodiode itself is not "flat" over the visible spectrum, peaking in the near-infrared and trailing off with shorter wavelengths - that is, toward the blue end.  Figure 4, above, shows the relative peak light outputs of a typical "white" LED overlaid with the response of the photodiode and once can see that they are somewhat complimentary.

To a limited degree, these two different curves will negate each other in that the sensitivity of the photodiode is a tilted toward the "red" end of the spectrum.  With the inference being that these glasses may be "dark enough", I wanted to make some more measurements.

Photographing the sun:

As it happens I have a Baader ND 5.0 solar film filter for my 8" telescope to allow direct, safe viewing of the sun via the telescope.  Because I'd melted a pair of glasses in front of the LED, I wasn't willing to make the same measurement with this (expensive!) filter so I decided to place each filter in front of the camera lens and photograph the sun using identical exposure settings as seen in Figure 5, below.

Figure 5:
The Baader filter on the left and the suspect glasses on the right.
These pictures were taken through a 200mm zoom lens using a Sigma SD-1 camera set to ISO 200 at F8 and 1/320th of a second.  Both use identical, fixed "Daylight" white balance.
Click on the image for a lager version.

What is very apparent is that the Baader filter is pretty much neutral in tone while the glasses are quite red.  To get a more meaningful measurement, I used an image manipulation program to determine the relative brightness of the R, G and B channels with their values rescaled to 8 bits:  Because the camera that I used - a Sigma SD-1 actually has RGB channels with its Foveon sensor rather than the more typical Bayer CMY matrix, these levels are reasonably accurate.  Note that the numbers below do not take "gamma" (discussed later) into account.

For the Baader filter:
  • Red = 163
  • Green = 167
  • Blue = 162
For the glasses:
  • Red = 211
  • Green = 67
  • Blue = 0 
Again, this seems to confirm that the glasses are quite red - with a bit of yellow and thrown in, which explains the orange-ish color.  Clearly, the glasses let in more red than the Baader, but the visible energy overall would appear to be roughly comparable using this method.

What the eye cannot see:

It is not just the visible light that can damage the eye's retina, but also ultraviolet and infrared and these wavelengths are a problem because their invisibility will not trigger the normal, protective pupilary response.  I have no easy way to measure the attenuation of ultraviolet of these glasses, but the complete lack of blue - and the fact that many plastics do a pretty good job of blocking UV - I wasn't particularly worried about it.  If one was worried, ordinary glasses or a piece of polycarbonate plastic would likely block much of the UV that managed to get through.

Infrared is another concern - and the sun puts out a lot of it!  What's more is that many plastics - even strongly tinted - will transmit near infrared quite easily even though they may block visible light.  An example of this are "theater gels" that are used to color stage lighting:  These gels can have a deep hue, but most are nearly transparent to infrared - and this also helps prevent them from instantly  bursting into flame when placed in front of hot lights.

Because of this I decided to include near-infrared in my measurements.  In addition to my Sigma SD-1, I also have an older SD-14 and a property of both of these cameras is that they have easily-removable "hot mirrors" which double as dust protectors.  What this means is that in a matter of seconds, one can adapt the camera to "see" infrared.  Using my SD-14 (that camera is mostly retired, and I didn't want to get dust on the SD-1's sensor) I repeated the same test with the hot mirror removed as can be seen in Figure 6.

Figure 6:
The Baader filter on the left and the glasses on the right showing the relative brightness when photographed in visible light + near infrared.
This camera, a Sigma SD-14, was set to ISO 100 at F25 and 1/400th of a second using the same 200mm lens as Figure 5.
Click on the image for a larger version.

According to published specifications (see this link) the response of the red channel of the Foveon sensor is fairly flat from about 575 to 775 nanometers and useful out a bit past 900 nanometers while the other channels - particularly the blue - have a bit of overlapping response while the hot mirror itself very strongly attenuates wavelengths longer than 675 nanometers.  What this means is that by analyzing the pictures in Figure 5, we can get an idea as to how much infrared the respective filters pass by noting the 8-bit converted RGB levels:

For the Baader filter:
  • Red = 111
  • Green = 0
  • Blue = 62
For the glasses:
  • Red = 224
  • Green = 0
  • Blue = 84 
While the camera used for figures 5 and 6 aren't the same, they use the same technology of imager which is known to have the same spectral response.  Taking into account the ISO differences, there is an approximate 3-4 F-stop difference between the two exposures (some of this is due to the fact that the morning sun was higher when the infrared pictures were taken) indicating that there is a significant amount of infrared energy - particularly manifest by the fact that the exposure had to be reduced such that the green channel no longer shows any readings when using the Baader filter.   

(Follow this link for a comparison of the transmission spectra of common filter media and follow this link for a discussion about the Baader filter in particular.)

What is clear is that the glasses let in a significant amount more infrared than the Baader filter within the response curve of the sensor - but by how much?

The data indicates that the pixel brightness of the "Red+IR" channel of the glasses is twice that of that of the Baader filter, but if one accounts for the gamma correction applied to photographic images (read about that here - link) - and presume this gamma value to be 2 - we can determine that the actual differences between the two is closer to 4:1.

What does all of this mean?

In terms of visible light, these particular "fake" glasses appear to transmit about the same amount of visible light as the known-safe Baader filter - although the glasses aren't offering true color rendition, putting a distinct red-orange cast on the solar disk.  In the infrared range - likely between 675 and 950nM - the glasses seem to permit about 4 times the light of the Baader filter.

At this point is is worth reminding the reader that this Baader filter is considered to be "safe" when placed over a telescope - in this case, my 8" telescope, as the various glass/plastic lenses along the optical path (e.g. corrector lens, eyepiece, etc.) will adequately block any stray UV.  What this means is that despite the tremendous light-gathering advantage of this telescope over the naked eye, the Baader filter still has a generous safety margin.  (It should be noted that this Baader film is not advertised to be "safe for direct viewing".  Their direct-viewing film has a stronger blue/UV and IR blocking.)

What may be inferred from this is that, based solely on the measurements that obtained with these glasses it would seem that they may let in about 4 times the amount of infrared (e.g. >675nm) light as the Baader filter.

Again, I did not have the facility to determine if these glasses adequately block UVA/B radiation - but the combination of these glasses and good-quality sunglasses will block UV A/B - and provide additional light reduction overall.

Will I use them?

Based on my testing, these particular glasses seem to be reasonably safe in most of the way that matter, but whatever "direct viewing" method that I choose (e.g. these glasses or other alternatives) I will be conservative:  Taking only occasional glances.

(I will acquire some "bona-fide" glasses and analyze them when I get a chance.)

* * *
Once again:

WARNING - PLEASE READ!
 
What preceded was my own, personal analysis of potentially defective products that, even when used properly, may result in permanent eye damage.  This analysis was done using equipment at hand and should not considered to be scientifically rigorous or precise.

DO NOT take what follows as a recommendation - or even an inference - that the glasses that I tested are safe, or that if you have similar-looking glasses, that they, too, are safe to use!

This analysis is relevant only the glasses that I have and there no guarantee that glasses that you have may be similar.  If you choose to use similar glasses that you might have, you are doing so at your own risk and I cannot be held liable for your actions!

YOU HAVE BEEN WARNED! 




After the eclipse:

As it turns out, I did use these glasses - at least part time.  There turned out to be an excess of eclipse glasses on-hand, many of them being verifiably "genuine".  With these on-hand I'll do an update and make a comparison between the two.


     [End]

This page stolen from "ka7oei.blogspot.com".