23cm PA project

The parts are coming together.  Nice isolators direct from RFCI in CA, peak 1kW, 200W continuous, built in 100W dummy load



FETs. At six quid each, the MRFE6S9160HS makes a nice cheap 23cm PA, cellphone technology instead of broadcast devicesfets

Hybrids. OK, I could use a printed Wilkinson as the input splitter, but hey…


Relays. The R570113030 is rather over-specified for this, but has 85dB isolation up to 1GHz and 80dB up to 2GHz and should handle up to 500W at 23cm very easily.


Tinplate boxes to screen the individual DF9IC PA boards


27V 18.6A switchmode PSU



More KE5FX phase noise tests

Three runs with a VBW/RBW ratio of 0.01, total run time about four minutes.  HP E4421B at 1296.200 MHz 0dBm into HP8652A using KE5FX PN.  Brings out the 50kHz interval spurs very nicely.  Now I need to find if they are from the 4421 or the 8652!


KE5FX PN settings1

Elad SW2 with DUO and CW Skimmer


Quick test with CW Skimmer using the MEM setting for DX cluster to local, using Softrock IF 48kHz with RX1 set to 48kHz IF and using the FT-897 CAT set.

I set the CW tone to 600Hz and the IF offset to -600Hz

Works OK to click in the spots window in SW2 and on the Skimmer display.  Ideally I want to use this on RX3/VRX3 so I con’t have to change to use 48kHz IF instead of audio for when I’m using every other bit of software, and don’t mess things up when I’m using split with RX2. Needs a lot more work!

Quick phase noise tests with KE5FX PN


This is a first shot at using the KE5FX suite of GPIB applications with my HP 8562A spectrum analyser.  Input was from my DB6NT TR1296H transverter driven by my Elad FDM-DUO.  50dB of attenuation of the 10 watt output gave me -10dBm into the 8562A. This is three runs at 1kHz to 1MHz using mostly default settings just to get a feel for the process.

I also ran similar tests with my E4421B signal generator and found it was a few dB better from 10kHz to 100kHz, but a few dB worse from 1kHz to 10kHz.

Now I need to try with some longer plot averaging runs and see if I can understand what goes wrong closer in than 1kHz.   I guess I need a really pure crystal oscillator to use as a reference tone.

Somewhere I have a KK7B LO board, maybe that would drive a doubler and give me a clean phase-noise-free source?

Ongoing xDSL noise problems

Since we got “proper” broadband in the village, I have been unable to use the 10MHz (30 metre) band for anything other than contacts with extremely loud stations.  There is a wall of noise right across the HF band, particularly bad between 8MHz and 12MHz.  The worst part is that if I transmit, the noise is replaced with some very loud training tones at 4kHz intervals.  Not content with wrecking my enjoyment of the band, the broadband service then screams in agony when I spoil its fun.

In the first video, check the band noise in the first few seconds.  It is way louder than it should be.  The the video freezes for 6 seocnd while I transmit.  Soon as I go back to receive, all hell breaks loose, with loud carriers at 4kHz spacing for about 100kHz around my transmit frequency.  It them retrains and goes back to the wall of noise, but leaves the training tones at a high level for 15-30 mins.

They call is progress.  Fibre service would be progress, this is just stone-age cheapo interference-making.

The training tones fire up a comb of carriers at 4kHz spacing.  Note the dip in diginoise close to where I was transmitting.

This is the noise on 10MHz before I transmit.  About 20-25dB over what the noise floor should be. 30m dipole up 30ft in a tree about 40m from the nearest phone pole.  Elad FDM-DUO and SW2

The waterfall shows more clearly that the diginoise spectrum looks like a ploughed field at 4kHz spacing

10MHz ploughed field

On the higher bands, the xDSL is notched out.  There is a sharp edge at 12MHz where the nature of the noise changes from solid mush to a pulsating 25Hz comb that sounds like a buzz on AM.  Here’s the 12MHz edge and guard-band

The notched-out bands from 20m up are well-defined:

and the pumping buzzsaw noises stop at just over 27.7MHz


for info, the screen recordings were made using the Windows 10 built-in recorder, accessed using Win-G to load the games toolbar, then hit record.  Much nicer than the freeware screen recorders with all the nasties they tend to install.

SW2 manual notch filters and UK VDSL tones

Thanks heavens for the notch filters in SW2! I’ve been trying to work V79XF on 10124.000 with a -90dBm VDSL noise floor which he is just about level with, but with a -74dBm VDSL training tone on 10123.991, that is NINE HERTZ LF of him, and another 4Hz LF of that which is -71dBm.

Without the filters, there is nothing apart from the tones, but a 10Hz wide manual notch set on 10123.987 kills the tones completely and lets me pull him out when the noise floor drops a little.

OK, I feel silly giving a 279 or 389 report, but my neighbours’ VDSL mush makes 30m a solid wall of noise with training tones every 4kHz.

Prologix GPIB-USB interface

I ordered the Prologix GPIB-USB Controller from SGLabs in Italy on Friday 11th March via ebay and it arrived today (Monday 14th March).  Impressively fast.

Bought it mostly because it was yellow.

Works really well with my 8562A spectrum analyser, but I haven’t received the GPIB cables I ordered from a UK supplier before placing the order with SGLabs, so only one device at a time for now…

Here’s a quick test of an FSK441 signal on 1296.200 MHz as received off-air by the 8562A, programmed using the KE5FX utilities:

talk 18 "IP;FA 1296.210 MHz;FB 1296.230 MHZ;LG 10 DB;RL -20 DB;RB 100 HZ;AT 0 DB;KSb;KSS;"
start ssm 18 -M -f -cs -w -t






More detailed testing to follow….