Televes Pro-75 UHF array used for UHF EME.


Encouraged by Anthony Mann's recent UHF EME television DX, I decided to also try for UHF moonbounce DX. This web page is thus still a work in progress while further results are obtained.

Spectrum analyzer

The first step was to download DL4YHF's excellent Spectrum Lab V2.0 b7 (Audio Signal Analyzer) FFT program. Some simple adjustments were needed including the FFT averaging setting, FFT input size, audio sample rate, lower and upper frequency scale, amplitude scale, and record line-in level.

Spectrum Lab has both line plot and waterfall displays. The waterfall display is especially useful for visual display of EME doppler shift.

Spectrum Lab also has a very useful function called FFT internal averaging. This parameter will be important when you try to see very weak signals out of the noise. This system is similar to long-term Integration for improving the apparent signal-to-noise ratio. A value of "10" will already reduce the random noise and will make the spectrum look smoother so you may better spot the "peaks" of weak signals with constant frequencies. I found a setting of somewhere between 20-50 gave the best results on extremely weak signals.

The FFT input size was set to 16384, and the audio sample rate was 11025 Hz. This corresponds to a bandwidth resolution of 0.6729 Hz. This is the default setting on Spectrum Lab.

Other FFT/sample rate combinations are possible. For example:

FFT input size 32768, & 11025 Hz sample rate = 0.3365 Hz bandwidth.
FFT input size 16384, & 8000 Hz sample rate = 0.4883 Hz bandwidth.

To adjust the FFT, sample rate, and averaging settings, click options, FFT settings, audio settings.

After repeatedly playing back UHF EME audio recordings into Spectrum Lab, I've found the best settings:

FFT input size: 16384.
Sample rate: 22050 Hz.
Average 30.
Resolution bandwidth = 1.3 Hz.

With the above settings, Spectrum Lab is able to detect signals down to 2dB above the noise. This translates to a sensitivity of roughly -145 to -160 dBM.

This narrow band FFT detection approach provides a significant boost in the signal-to-noise ratio which compensates for the relatively low gain wide-band UHF TV yagi.

Receiving antenna and preamplifier

A Televes Pro-75 wide-band 75 element yagi covering 470-860 MHz was pole mounted at 3 meters above ground. This is adequate height, given that the yagi simply seeing the moon is sufficient. Also, having the receiving aerial at low elevation, helps minimize any interference from local UHF transmitters.

The Pro-75 has a 40 degree tilt mechanism which was adequate for pointing up to a large area of the sky. However, at certain times, the moon is above 40 degrees elevation. To enable tilt elevation up to 80 degrees, the UHF yagi was mounted on a horizontal cross boom. A short length of PVC pipe was bolted to the top of the vertical mast.

The Pro-75 is fitted with a 14dB gain MRD (minimum rising device) 170K (2db noise figure) wide-band UHF masthead preamplifier. A 170K noise figure is of course high by EME standards, but at least is sufficient to detect the stronger UHF stations. A lower noise 75K GaAsFET preamplifier will be added at a later date.

Because of the relatively low gain of the masthead preamp, a second indoor Alcad BR-105 modified balun 14dB UHF preamp was used to compensate for coax cable loss. DSC2.1 coax (similar to RG-11) was used for connection to a Icom R-8500 synthesized VHF/UHF receiver.

The R-8500 was set to USB mode. The audio line-out of the R-8500 is connected to the PC's line-in socket. The audio levels need to be adjusted, in order that the input levels to the computer's sound card are optimum.

The input levels to Spectrum Lab are relatively critical. Excessive signal will result in signal clipping. The PC's record input level should be adjusted for a noise level of approximately 60dB on Spectrum Lab.

By looking at the various Spectrum Lab screen shots included on this page, it will give you a idea regarding correct input levels. Signal clipping will result in a green colored display. Insufficient level will result in either a dark blue or black display. Light blue will indicate a optimum input level.

Reference source for measuring EME TV video carriers

ABN2 was used as a reference source for a 15625 Hz TV-derived reference unit. The variation is as follows (R-8500 tuned to 501.248300 MHz):

ABN2: 2874-2876 Hz variation.
ATN7: 2871-2882 Hz variation.

ABN2 clearly provides better frequency stability. A frequency variation of +/- 1 Hz is adequate at UHF.

The reference signal was combined with the DX signal. By connecting the center pin of the reference coax lead to one input of a Tandy high isolation switch, enough leakage was there to produce a weak, but constant reference signal. In fact, the reference signal is not audible, but strong enough to be seen as a trace on Spectrum Lab.

Blocked UHF TV channels

Due to the extremely high sensitivity of Spectrum Lab, very few UHF channels are clear at my location for EME DX. For example, any tx above 5KW within 100 miles, will display sideband traces. For this reason, most of my EME listening is concentrated below 525 MHz. 470-521 MHz is the most productive area, because no Australian UHF TV channels are allocated below channel 28.

First attempt at receiving KWBT-19

Anthony sent me the moonset schedules for KWBT-19 (see below), including reception time window (UTC), azimuth, elevation, doppler shift, polarization, and receiver offset.

The first attempt was unsuccessful mainly because I was confusing reference signals for DX signals. The following morning Anthony rang 15 minutes before the calculated moonset time window for KWBT. Rain and heavy cloud cover made it very difficult for me to aim the UHF yagi at the moon. I was fortunate to briefly spot some light through the cloud cover, thus was able to at least aim in the general direction. At 1500, 22nd March, 2003, Anthony first started to note traces from KWBT. At 1506 UTC (2:06AM local time) I also started to note traces from KWBT! As predicted, the carrier was drifting down at a rate of 1.8Hz/per minute. My reception lasted from 1506-1515 UTC.

The following evenings were more successful. With a clear view of the moon, and thus more accurate aiming, reception of KWBT was now slightly stronger and of longer duration. On 3 subsequent mornings KWBT appeared right on schedule and lasted for 20 minutes. The last 2 minutes usually provided the strongest reception. This later peak is typical for moonset EME.

A screen shot from Spectrum Lab is shown below. Notice the signal peak for KWBT-19 at 631 Hz (left of screen). The 3518 Hz reference signal can be weakly seen at the far right of the screen.

Fig.1: 501.247713 MHz KWBT-19 Muskogee, Oklahoma, USA.

Another screen shot from Spectrum Lab is shown below. Notice the 12dB signal peak for WNDU-16 at 1495 Hz. The weaker 1275 Hz reference signal can be weakly seen at the left of WNDU's trace.

Fig.2: 483.25 MHz WNDU-16 Southbend, Indiana, USA.

First attempt at receiving chE24 Kuwait City

After the KWBT-19 reception we now knew that my UHF system was at least sensitive enough for the stronger EME TV signals. Given the current unstable situation in the Middle East, I was keen to quickly try for ch24 Kuwait City. If I left it too long, ch24 might be a smoking off-air tx!!

Anthony rang 10 minutes before the schedule moonrise at Kuwait. Sure enough Anthony started to received traces at 2340 UTC. I first noted traces at 2343 UTC (1043AM local time, 27th March). Ch24 peaked at 2345-2347 UTC near the commencement of reception. This early peak is typical for moonrise EME.

A screen shot from Spectrum Lab is shown below. Notice the signal peak for ch24 Kuwait at 1211 Hz. The stronger 914 Hz reference signal can be seen to the left of the ch24 Kuwait signal trace.

When time permits, simultaneous EME reception of channels 24 and 39 Kuwait City will be tried. Anthony reports that the ch39 video carrier is currently around 615.2494 MHz.

Fig.3: 495.250297 MHz chE24 Kuwait City.

Frequency calculations of received video carriers

I use the Doppler at the "baseline" time where the TX elevation is either +1 or 0 degree, whatever is the last entry supplied by MoonSked, and I compute the rate (dividing by 10 to get a Hz per minute rate). Then I multiply both by the frequency ratio (for example, 501/432 MHz) and use those numbers for the calculation. I then correct the measured offset from the frequency reference back to the "baseline" time given in the program. Finally I subtract the (corrected) Doppler and this then gives the correct TX frequency. Example:

Doppler drift rate = 1.8 Hz/min @ 432 MHz (141-123Hz=18Hz).
Doppler drift rare = 2.1 Hz/min @ 501.26 MHz (501.26/432).
1240 UTC = -176 Hz doppler @ 432 MHz.
1248 UTC = -190.4 Hz doppler @ 432 MHz.
501.26/432 MHz = 1.1603 (frequency ratio).
190.4 Hz x 1.1603 = 221 Hz.
Ref signal=1345 Hz, DX trace=1137 Hz @ 1248 UTC.
1345 - 1137 = 208.
501.259 792 + 221 Hz = 519.260013 MHz (WLTX-19 Columbia, Sth Carolina).


One of the most important aspects of EME DX is the advance preparation and organization required. Local weather, time of day, local interference (TV or utility services), TX power, audio input levels, spectrum analyzer parameters, combining reference and DX signals, EME time window, polarization, antenna alignment, etc, are just some of the aspects that need to be carefully considered in advance before any EME opening.

Potential improvements

Since most of my EME listening is concentrated in the 470-521 MHz area, a group A UHF yagi would provide more gain. Two stacked group A yagis, such as the Fuba XC391A, would provide about 16dBd gain at 500 MHz. A low loss (0.3dB) stacking coupler could be used to combine the two yagis. The output of the coupler would be fed to a 0.5dB NF GaAsFET UHF preamplifier.

A Channel Master 4251 7 ft screened dish would be ideal for minimizing earth noise arriving from the rear of the antenna. A TVRO dish with an adjustable feed point would be ideal for variable polarization.

Effect of local UHF TX terrain on EME signal strengths

Television transmitters concentrate most of their power at the local horizon. This explains why EME signals are only heard when the moon is no higher than 2 degrees above the horizon at the transmitter site.

Peak signal strengths are usually noted between 4-8 minutes after the 0 degree TX elevation point.

Any local obstructions, including man-made objects or mountains will attenuate signal beamed toward the horizon. Certain USA UHF TV tx sites, notably California, have a good take-off to the west. Californian UHF TV transmitters are located on high mountains, thus there is minimal obstructions to the Pacific Ocean.

Kuwait is another example of a TX site were there is minimal obstructions to the horizon. Kuwait UHF tx's are heard at relatively good levels because of the excellent eastern take-off toward the Persian Gulf.

Optimum Spectrum Lab parameters

EME signals can be recorded on to various mediums including audio tape, CD, mini disc, PC, etc. These recordings can later be played back for the purpose of finding the optimum bandwidth and average settings on Spectrum Laboratory.

Due to doppler shift, the apparent received frequency of EME signals change with time. At my location, the typical drift rate is -2 Hz/min. The 2 Hz frequency drift means that very narrow bandwidths may offer no improvement. This is because the received frequency may be outside the Spectrum analyzer's bin width. I hope to make some recordings and later determine what settings are optimum on Spectrum Lab.


Subsequent tests on extremely weak UHF TV tropospheric scatter signals indicate that reception is possible out to approximately 500 miles any time of the day. For example, 534.224 MHz SBS channel 29, Bendigo, Victoria (700 km) is about 4dB above the noise on Spectrum Lab. At greater antenna height, these figures would of course be higher.

My recent EME reception confirms that the UHF system is adequate for terrestrial tropospheric TV DX. The results were encouraging, especially considering the foward gain spec of the Pro-75 is only around 12dBd at 500 MHz.

Special thanks to Anthony Mann for his help and suggestions.

Dat-75, 12-16dBd gain UHF TV aerial (replaces Pro-75).

Additional notes on using Spectrum Lab

Go to

The following files were saved by right clicking the links. I created a SpecLab folder in C:/program files

Download and unzip all parts into one folder (about 1.5 MB). Once downloaded, all these zip files need to be extracted using WinZIP.

Then simply run the InstallSpecLab.exe file.

One important step is to first optimize the input level from your radio to the line-input at the back of the PC. Go to Options, then click volume control for "record" (audio in). Make sure the line-in box is ticked.

With the radio connected to an external aerial, and with no signals present, the rec line-in gain control should be adjusted so the noise floor is around -60dB. Typical setting is somewhere between -50 to -65dB. If there is insufficient signal, the noise floor could be as low as -100dB. Excessive signal levels will result in clipping (distortion).

The colour of the spectrum graph (waterfall) background provide an indication of input levels to the sound card. Black or dark blue indicates insufficient signal. light blue and/or bluish green indicated optimum levels. Bright pale green or yellow indicate excessive signal.

The frequency span is set to cover 500-2500 Hz.

The average setting needs to be set higher than 0. I found somewhere between 20 and 40 were ideal for stable weak carriers.

The sensitivity is quite remarkable. Providing there are no obvious high external noise levels or other interference, video carrier signals can be detected out to around 700-900 km via troposcatter.

D layer ionoscatter can also be detected. For example, New Zealand 45.25 MHz TV (1,300 miles) is about 2-5 dB above the noise on most days. Shorter distance signals are even stronger. This is because 1,300 miles is at the extreme upper limit for ionoscatter. 800 miles is about optimum. The peak time for ionoscatter is around midday.

Since my PC radiates high level hash on band I frequencies, I prefer to initially make a audio recording, and then later feed the audio into Spectrum Lab. For example, tuning your scanner in USB mode 1KHz below 48.25 MHz, and then leave the tape running for at least 10 minutes. Of course, if your PC is ultra-quiet, you can monitor real time.

Stations received via EME

WMPT-22 519.26, 39.01 N, 76.61 W, Annapolis, Maryland, USA.
WDCA-20 507.26, 38.96 N, 77.11 W, Washington, D.C, USA.
WGPX-16 483.25, 36.25 N, 79.66 W, Burlington, N. Carolina, USA.
WNCN-17 489.24, 35.68 N, 78.53 W, Goldsboro, N. Carolina, USA.
WLTX-19 501.26, 34.10 N, 80.77 W, Columbia, S. Carolina, USA.
WKCF-17 495.24, 28.58 N, 81.08 W, Clermont, Florida, USA.
WBBH-20 507.26, 26.83 N, 81.77 W, Fort Myers, Florida, USA.
WBXX-20 507.26, 36.11 N, 84.34 W, Crossville, Tennessee, USA.
WNDU-16 483.25, 41.61 N, 86.21 W, South Bend, Indiana, USA.
KXVO-15 477.25, 41.07 N, 96.23 W, Omaha, Nebraska, USA.
KWBT-19 501.247, 35.75 N, 95.81 W, Muskogee, Oklahoma, USA.
KWKB-20 507.24, 41.73 N, 91.35 W, Iowa City, Iowa, USA.
WMPV-21 513.26, 30.59 N, 87.56 W, Mobile, Alabama, USA.
KPPX-51 693.25, 33.3 N, 112.05 W, Arizona, USA.
KXTX-39 621.2494, 32.6 N, 96.97 W, Dallas, Texas, USA.
KFTV-21 513.25, 37.07 N, 119.43 W, Hanford, California, USA.
KUVS-19 501.24, 38.12 N, 120.73 W, Modesto, California, USA.
KDTV-14 471.26, 37.50 N, 121.87 W, San Francisco, Calif, USA.
495.2500, 62.0 N, 14.4 E, Sveg, Sweden.
495.2502/615.2494, 29.33 N, 48.00 E, Kuwait.

Moonsked data 501.247 MHz KWBT-19 Muskogee, Oklahoma, USA

NB: receiver offset (Doppler shift) is calculated for 432MHz: need to multiply by 1.16 for 501 MHz.

date    UTC  rx az  rx el  pol  tx qtf  tx el  range

20-May-03  15:10  090  +48  +17  236   2  377098  -230 Hz
20-May-03  15:20  089  +50  +16  238   0  377141  -252 Hz
21-May-03  16:20  083  +50  +12  241   2  383633  -280 Hz
21-May-03  16:30  082  +52  +11  242   0  383675  -303 Hz
22-May-03  17:30  074  +51  +5  247   1  389799  -336 Hz


DX TV via EME (Moonbounce) propagation A report on Tony Mann and Ian Roberts' recent EME experiments.

Software DSP Solutions for Weak Signal Communications by Roger Rehr, W3SZ.

DSP-10 Weak-Signal EME.

David Anderson, GM4JJJ's Moonsked. "The Complete Moonbounce Scheduling and Tracking solution for Macintosh and Windows".

Spectrum Lab. DL4YHF's Amateur Radio Software: Audio Spectrum Analyzer.




Copyright © 2005 Todd Emslie