GW4ALG's 136 kHz Pages

[ GW4ALG went QRT in February 2007 ]

Home ] Antennas ] Equipment ] For Sale ] Links ] Logbook ] Introduction ] Station Summary ]

Up ] 100 W P.A. ] 400 W P.A. ] Audio Filter ] Components ] Loop ATU ] [ Noise Canceller ] QRP TX ] Receive Pre-selector ] Remote Controlled Tuner ] SWR Bridge ] Transverter ]


Noise Canceller for 136 kHz

Operation on LF using small antennas from an urban or sub-urban location is a challenge in itself.  But just one local source of noise can make it impossible to hear any amateur signals on the band.  It was local noise that prompted me to build my first noise canceller.

The principle is quite easy to understand: Obtain a sample of the interference signal; amplify it; and then couple the 'noise' signal into the receiver so that it exactly cancels the noise signal picked up by the main antenna.

Initial results
Initial results with a noise canceller on local sources of interference were most encouraging.  Based on an idea by VK5BR (RadCom, March 1993, p34), I constructed the circuit shown in Figure 1.  In particular, I found that:
- the RF loss through the canceller (between the Main Antenna port and the Receiver port), was about 8 dB at 136 kHz;
- under test conditions, a sine wave can be nulled by greater than 50 dB;
- my strongest noise source (possibly a light dimmer), which, since mid-1998, has prevented me operating in the evenings can now be reduced from S9 to less than S5;
- the number of significant noise sources at my QTH has increased from 1 to 3 over the past 4 months;
- an untuned length of wire, about 10 m long, and located in the loft [attic] in proximity to mains wiring seems to be make a good noise sense antenna for nulling out interference received on my vertical antenna.
- the 10 m wire does not always provide sufficient noise to null out interference heard while using my 60 m single-turn delta loop antenna;
- a combination of phase/amplitude settings, which take about 30 seconds to optimise, can only null out one interfering signal at a time, for a given receive antenna.


FIGURE 1 - Circuit diagram of basic noise canceller

Basic noise canceller

Basic Noise Canceller

Figure 1 shows my version of the VK5BR noise canceller, adapted for use on 136 kHz.   L1; L2; and T1 were made using 25 mm OD 3C85 ring cores and wound with wire obtained from stripped-down internal telephone cable (the wire is plastic coated with a conductor size of 0.020 inch (25 SWG, or about 24 AWG).  L1 & L2 are wound on separate ring cores and each is wound with 28 turns (see 'Alignment Procedure' below).  The hybrid transformer, T1 uses 13 (+/- 2) turns, quadfiliar wound at about one twist every 20 mm.  (If you are new to this type of transformer construction, see below.)

The variable capacitor provides a variable phase shift of 0-180 degrees, and SW1 provides a further 180 degree phase shift, permitting an overall phase change of 0-360 degrees.  The potentiometer is used to adjust the amplitude of the noise signal.   In use, getting the exact settings optimised can take several seconds - but the results are well worth the effort!

Alignment Procedure
The values for L1 and L2, together with the 570 pF of fixed tuning capacitance, will provide the required phase shift of  0 to 180 degrees.  (If using a 200 + 200 pF variable capactor, L1 and L2 should be wound with 34 turns; and the two 570 pF capacitors should each be changed to 400 pF.)   But variation between different cores will mean that some adjustment will be required to ensure that the required phase shift range is obtained.

Firstly, disconnect the variable capacitor wire to L2, connect the noise antenna and monitor the noise output on the station receiver.  Noise on 137 kHz should peak with the variable capacitor (now tuning L1 only) at about the centre position.  (Note that the 'Q' of the circuit is quite low, so the tuning will be very 'flat', with a broad-band frequency response.)  If the noise gets stronger at maximum capacitance: increase the value of fixed capacitance.  If the noise gets stronger at minimum capacitance: decrease the number of turns on L1, one turn at a time, until resonance occurs near the centre position of the variable capacitor.  You may need to remove up to 6 turns from L1. 

When this has been done, re-connect the variable capacitor wire to L2; then disconnect the variable capacitor wire from L1.  As before, if the noise gets stronger at maximum capacitance: increase the value of fixed capacitance.  If the noise gets stronger at minimum capacitance: decrease the number of turns on L2.  Again, you may need to remove up to 6 turns from L2. 

Using the noise canceller
The useable bandwidth at the optimum cancellation settings often seems to be about 600 Hz, depending upon the strength (level of annoyance) of the QRM.  The apparent bandwidth over which the cancellation is optimum decreases for an increase in the level of interfering signal.  My log for the 20/21st November 1999 reads:

Saturday, 20th November 1999:
06:00 HB9ASB (12 m vertical; sent 579, received 3/4 69) Our first QSO with only my 12 m vertical
07:36 G3YMC (12; 559, 599)
09:33 G4GVC (12; 599, 599)
11:08 GD3XTZ/P (12; 589, 579) First GD-GW
11:48 G6NB (12; 589, 589)
19:30 G6RO (12; 579, 569)
20:20 OH1TN (12; 579, 3/4 39)

Sunday, 21st November 1999:
04:55 SM6PXJ (12; 579, 439)
07:02 PA0BWL (12; 349, 449)
10:40 GD3YXM/P (12; 599, 579)
11:07 GD0MRF (12; 599, 579)
11:50 GB2CPM (12; 599, 589)
17:17 GI3PDN (12; 569, 449) Our first QSO
19:17 HB9ASB (20; 589, 579)
20:00 OH1TN (20; 579, 549)
20:08 OH3LYG (20; 359, 559) Our first QSO
20:24 G8RW (20; 569, 559)
20:45 G3KEV (20; 599, 579)
21:17 ON7YD (20; 569, 549)
22:02 SM6PXJ (20; 579, 559)

None of the evening QSOs on these two days would have been possible without the use of my noise canceller - the noise level was in the range S8 - S9 throughout.

A pair of switched noise cancellers

Because the basic noise canceller requires different settings for each of my two receive antennas (loop antenna/experimental vertical), I found myself having to adjust the noise canceller quite frequently.  My current arrangement involves two switched noise cancellers in one box.  I use a sense signal from my antenna switch to automatically select the appropriate noise canceller for the chosen antenna.  Also, I have now introduced a 'mix' facility so that I can try to cancel two different QRM sources simultaneously.

Front View

Front View

Rear View

Rear View


FIGURE 2 - Circuit diagram of switched noise cancellers
Dual Noise Canceller Circuit

Note that, in the case of T1B, an extra winding has been added to couple the output of the second canceller.  Thus, for T1B, it is necessary to twist six wires together (at about one complete twist per 20 mm) before winding the transformer.  To make it easier to wind T1B, I used a 42 mm OD 3C85 ring core.   (But a 25 mm core would probably have worked just as well.)  To prevent interaction between the phase and amplitude settings of NC1 and NC2, avoid selecting the same noise antenna for both noise cancellers.  Also, during construction, use 'point-to-point' wiring around SW1 and SW2 to reduce capacitance between the two circuits.

In the 'Auto' mode, noise canceller 'NC2' is automatically selected when the main station antenna switch applies an earth [ground] to the terminal 'Remote NC Select'.

Compensating for loss through the hybrid transformer

Of course, one thing leads to another!  The 8 dB loss between the Main Antenna port and the Receive port of the hybrid transformer (T1/T1B) meant that the usual S9 signals were only just reading S8 on the bargraph-type S meter of the FT707.  To compensate for this loss, a Receive Pre-selector was constructed and connected at the Receive port as shown in Figure 2.  A resistive pad was used at the output of the pre-selector to set the overall loss through the noise canceller to 0 dB.

Construction of T1/T1B

T1 and T1B should be wound with colour-coded wire, such as that obtained from stripped-down internal telephone cable, being sure to separate any 'twisted-pairs'.   Using colour-coded wire makes it easier to identify each conductor when terminating the wires. 

First, the wires should be twisted together at about one twist per 20 mm.  (Four wires should be twisted together for T1, and six wires for T1B.)  Ensure that you allow enough of a 'tail' to terminate the windings, and use a cable tie to anchor the start of the winding.  (The start of each winding is denoted by the black dot in Figure 1 and Figure 2 above.)  Then wind about 13 turns around the ring core, using another cable tie to anchor the end of the winding.  Click on the thumbnail to enlarge the picture of T1B.

Picture of T1B - Click to enlarge!


Possible Improvements

During June 2001, many hours were spent at the bench with the dual noise canceller pulled apart - soldering iron in hand - to help me better understand the present design, and see how it might be improved.  I came to the conclusion that:
1)  the phase shift network is affected by loading effects from the noise antenna;
2)  in practice, it is not possible to realise a full 0-180 degree phase shift - even with very large changes in tuning capacitance.  A maximum range of 130 degrees is more typical; and,
3)  the diode switching in the dual noise-canceller version is responsible for causing break-through from HF frequencies on some evenings.  Even when the diode was biased 'on', shorting the diode would often reduce or eliminate unwanted heterodyne whistles.

The dual noise canceller will now be restored as per the above design because, although it may not be perfect, it continues to provide essential noise cancelling facilities at my QTH.  When time permits, I would like to build a new dual noise canceller with the following improvements:
a)  better switching between the individual noise canceller outputs, possibly using electro-mechanical relays rather than diodes;
b)  provide an impedance step-down transformer at each noise antenna input so that antenna switching is carried out at a low impedance point, thereby reducing the effects of capacitance in the internal wiring;
c)  include an option for using low impedance noise antenna inputs - for use with remote receiving  antennas, fed via coaxial cable;
d)  include additional signal-frequency bandpass circuits at the noise antenna input, with an FET buffer/amplifier, thereby reducing the dependency on the phase-shift network to provide front-end selectivity;
e)  include a two-stage phase shifter comprising a switchable 90 degree shift network, followed by a variable 0-90 degree phase shifter: thereby covering the required 180 degree phase shift in two ranges; and,
f)  provide some front-end selectivity between the noise antenna input terminal and the diode protection circuit (currently across L1).


Further Reading

A paper written by Derek Atter, G3GRO entitled 'A Simple Signal Canceller For 136Khz To Combat Loran or Other Noise Sources' can be found via the Crawley ARC web site, and at:

Another paper written by Derek Atter, G3GRO entitled 'Methods of Interference Suppression Evaluation of Receiving Loops etc. And the MFJ 1025 Signal Canceller' can be found via the Crawley ARC web site, and at:

Use of the JPS Communications/Timewave Technology 'ANC-4' at 136 kHz.  Notes from GW4ALG's initial testing of this noise canceller can be found at:

For details of a 50 MHz noise cancelling receiver see:




Initial Tests of ANC-4 Noise Canceller at 136 kHz ]