As of 26.09.12, this page becomes the home page for the 3cm Mow Cop  .....................................receiver running via WebSDR

Best dx so far:  HB9G at 952km

Latest news (25.05.15) - Head end failure.

The receiver had seemed a little deaf for the best part of a year, and it finally failed at the beginning of May. Inevitably, the cause was water ingress. For a Full report, please click here.

Also, a second front-end unit has been built so that we can have something actually on the tower, with full 360 degree coverage of the far horizon. Details here.

* The beacon receiver is currently running 1.14 kHz low (16.02.15)*
(use GB3XGH as your frequency reference - it is GPS locked)

The Bawdsey remote SDR installation was removed off site in 2009, when permission to operate there was effectively revoked. No internet connection was ever realised from that site, though Dave G4HUP did record some decent dx, and it has now found a very good new home at Mow Cop and is about to be made accessible via WebSDR thanks to the efforts of Martin, G7CKX.

antenna location

To date, the antenna has remained firmly attached to the site cabin roof - at some point, either it or a duplicate antenna should make it onto the tower, adding at least 10m to the height. This additional height above ground would ensure that the current shadowing to the North and North-West will be eliminated.

A single192 kHz window centred on 10 368.855 MHz is currently being used. The following UK beacons are therefore covered:

10 368.810  GB3XGH   Rochdale   (heard) 
10 368.830  GB3MHX  Martlesham  (heard - though only rarely)
10 368.850  GB3SEE   Reigate   
10 368.870  GB3KBQ  Taunton   
10 368.895  GB3NGI   Antrim
10 368.900  GB3AZA  Scarborough (currently off-air)
10 368.905  GB3SCX  Bell Hill   
10 368.940  GB3CCX  Cleeve Common   
10 368.945  GB3PKT  Clacton   
Plus, just outside of +-96 kHz, but still available as alias* signals are:

10 368.752 GB3FNY Sandtoft (alias appears at 10 368.944)    (heard)
10 368.755 GB3CAM Wyton (alias appears at 10 368.947)    (heard)  
10 368.955 GB3LEX Leicester (alias appears at 10 368.763)   

Plus well over a dozen mainland Europe beacons.

Here are some screen shots:  Flat band   GB3LEX/GB3CCX    GB3SCX   
  GB3LEX out of lock   GB3CAM   PI7ALK    GB3KBQ    HB9G    F5ZBA    F1ZAI   GB3FNY   GB3PKT
* alias signal reception:
There have been a few comments/queries via the receiver text screen regarding the reception of GB3LEX and GB3CAM. The former is 100 kHz higher in frequency than the sound card sampling frequency whilst the latter, by co-incidence, is 100 kHz below, so both are at first thought, out of range of this receiver. However, if you consider sound card sampling as being a mixing process, you would expect an image response. This ought to be removed at source by the sound card anti-alias filter. In practice though, it takes the filter some 10 kHz beyond the 96 kHz sampling frequency to provide a lot of attenuation. This is easy enough to show by feeding a signal into the receiver, initially at the centre frequency (ie, DC as far as the sound card is concerned) and then moving away higher, as the picture shows below

This particular signal was 30dB above noise floor, 1 kHz above centre frequency. As its frequency was increased, an eye was kept on this signal to noise ratio. It remained within a dB of this right up to 88 kHz above centre frequency. At 96 kHz it was 3 dB down. Increasing beyond the 96 kHz sampling frequency, the trace disappears off screen on the rhs, only to reappear on the lhs. It is the phasing nature of the receiver that causes this effect - if you detect using only the I or the Q audio signal, you see a more familiar sight:
I or Q audio 
ie, the image appears both sides. At 100 kHz offset, the signal to noise ratio has decreased by 10 dB, so when you see either GB3LEX or GB3CAM, the signal is actually 10 dB stronger than it appears.                                                                                                                                                                                                                                                                 


IF section
Receiver description

A Connor-Winfield ASOV5S3E 10 MHz OCXO should ensure that the frequency stability of the receiver remains better than +- 200Hz.

The same antenna and LNB combination that was used for the Bawdsey receiver is used here, but a different IF is employed, based on an '80m Finningley SDR' board. In a bit of a Heath Robinson arrangement, the on-board 15 MHz crystal oscillator is replaced with an oscillator/multiplier board using a 12.570 MHz crystal multiplied by a factor of 6 to 75.420 MHz (in fact, it's another of the same multiplier boards used in the front-end 'Berniebox').  This results in an IF coverage of 18.855 MHz +- 96 kHz.

Output from the SDR feeds an Asus Xonar D1 sound card fitted to the server.

Notes on the original Bawdsey 3cm receiver

This receiver uses the LNB based converter described elsewhere on the index page.

To obtain a nominally even azimuth gain, a slotted waveguide antenna with side wings has been used. This has been produced by David Wrigley G6GXK, and is shown below. Multiple slots provide elevation gain.

There are 12 pairs of slots resulting in a gain
of 13.5dBi. Extensions, or wings, added to the
narrow waveguide face reduce the variation
in azimuthal gain to about +-2.3dB.

(view radiation pattern picture in a seperate)

(antenna pics/plot courtesy of David Wrigley)

slot antenna
slot antenna with radome
radiation pattern

Waveguide input LNB

Although both WR-20 and C120 waveguide input LNBs are available, they are not very common (particularly with 9.75GHz LO and extended lower band edge of 10.75GHz). It was discovered that on a couple of brands of mini-dish LNB that the first choke ring provided a tight fit for a 22mm straight copper coupler, providing a suitable interface for 22mm copper pipe to be used as waveguide.

LNB 22mm coupler LNB front-on
This particular LNB is manufactured by WNG, and
the 22mm coupler is a good tight intereference fit,
which means that the mating copper tube wave-
guide can be rotated without having to worry about
the coupler itself rotating or coming loose, which is
actually quite a useful feature.

Probe skew (relative to the eliptical feed horn)
ensures that it is not obvious what the correct
alignment is for the copper tube waveguide, so
being able to rotate this for best signal-to-noise
ratio is very useful.

The next requirement is a transition from 22mm circular waveguide to WR-16.

transition lnbfeed

Amazing what can be done with a rubber mallet...

Tuning screws were added to correct the inevitable mismatch, though this turned out not to be too bad, the un-tweaked return loss being about 10dB. Originally, a seperate coupler was envisaged to insert the reference (LO) signal whilst providing some directivity, but has not been entirely satisfactory, so a simple coupling probe has been added to the transition - this explains the short sma flying lead.

Some noise figure measurements

An HP 346A noise source was used to feed an sma-to-WG16 adaptor, which was bolted to the end of the transition section shown above, and the tuning screws adjusted for best sensitivity (measured via the Bernie box at 18MHz). Repeatedly switching the noise source on and off gave the following result:

                set up


The LNB  housing is a 1m length of Wickes soil pipe. This provides a moderately good fit to the 4" mast. Note that the reference feed tail on the transition has been replaced by a bulkhead sma connector. Not only does this give a mechanically beter arrangement, but it also allos the probe length to be reliably set (and adjusted).




LNB, with David's antenna fitted:


A final tweak of the matching screws on the transition piece was then made - it did make half a dB difference, which is a slight supprise.
The following set-up was used for the tweak.


A 30 dB amplifier was added to the HP noise source, which fed the antenna fastened to the ladder on the lhs of the picture (probe fitted within an old Grundig LNB).  The receive antenna/LNB combination (rhs of picture atop three wooden pallets) then fed the converter via 8m of twin coax. Output from the converter was split between the noise figure meter input and a spectrum analyser, used as a monitor to check all was well. Using the noise figure meter for this adjustment is good, because the meter can be set for heavy averaging.