Remote Active Aerial for Reception

I've constructed a remotely tuned loop aerial and it worked OK but was cumbersome to erect and adjust and it failed due to corrosion, so I decided to try instead a ferrite rod aerial in its place. Why should I bother when a random wire is simpler to use? The answer is local interference levels within one's house and immediate surroundings can be significantly reduced by moving the antenna to one's property boundary. Accurate tuning of a desired broadcast will also reduce general interference levels. In practice tuning will be relatively broad and increasingly so the higher the frequency. Optimistically I'd like to cover the 40m and 80m amateur bands down to VLF and I'll be using the usual Armstrong rotation method.

I intend to use similar hardware to that of my tuneable loop aerial viz. a single transistor amplifier to match the high impedance aerial circuit to a low impedance coax feed. A varicap diode driven from a control box on my computer desk will enable the coils wound on the ferrite rods to be tuned and a few miniature relays will enable bandswitching between a few coils (maybe on different ferrite rods).

I learned from my loop aerial that the circuitry needs to be carefully protected from weather so I've ordered a plastic case fitted with a weatherproof seal. I've also procured a selection of ferrite rods and a few coils. The aim is to make five tuning ranges as below, but there will be some problems, similar to those met in designing my loop aerial coils...

Range	Min Freq KHz	Max Freq KHz	Inductance	Min Tune pF	Max Tune pF	Coil pF	Notes
1	2000	8000	11.3uH	35	530	?	-
2	500	2000	191uH	35	530	?	-
3	128	500	2.9mH	35	530	?	-
4	38	150	32mH	35	530	?	high coil pF
5	10	40	452mH	35	530	?	high coil pF

The table above lists the various wavebands and, using basic design equations, the numbers seem OK (unless I've made an arithmetic error..), however all is not what it seems. Firstly the selected ferrite rod(s) may affect the range of frequencies in different ways and secondly, each coil will have some inherent effective (parasitic) capacitance which will significantly affect tuning range for the lowest ranges. To reduce this capacitance I'll wind (or select) the coils to minimise it and the table above includes a column to add a figure for each coil. There's a really clever way to determine the parasitic capacitance by feeding a pulse into the coil and examining its ringing frequency on an oscilloscope. The alternative is testing the resonant frequency using fixed capacitors.

As an example I've taken Range 4 to investigate the effect of parasitic capacitance on tuning range. No problem with the low end, but the high end is significantly lowered. In fact, in the best case I'd have to reduce the coil inductance from 32mH to 15mH to tune 150KHz, this lower inductance value would then increase the low end from 38KHz to 54KHz. Recalculating Range 5, I'd need 116mH tuning 54KHz down to 20KHz. Adding a Range 6 requires 844mH tuning 20KHz down to 7KHz but of course that coil would have loads more parasitic capacity so negating these calculations! I suspect the results in practice would preclude tuning anything lower than say 25KHz, but I need to investigate typical coil parasitic capacity values... Another option is to use two varicaps in parallel which would improve the tuning range as these would give me a capacitance range of 100pF to 1090pF (bearing in mind 10pF stray capacitance wouldn't materially change). 15mH would tune 128KHz to 39KHz. To go lower perhaps three varicaps in parallel... 126mH giving a range of 40KHz down to 11KHz without recalculating parasitic capacitance.

Looking at the ratio between the max and min frequencies below you can see the effect of undesirable capacitance. In theory, twin varicaps and triple varicaps result in ratios of 3.28 and 3.6 respectively.

Range 4 (effect of coil parasitic capacity)

Frequency	Min Tune pF	Max Tune pF	Parasitic pF	Frequency	Parasitic pF	Frequency	Parasitic pF	Frequency	Parasitic pF	Frequency
38KHz	-	530	40	37KHz	75	36KHz	150	34KHz	300	31KHz
150KHz	35	-	40	103KHz	75	84KHz	150	65KHz	300	48KHz
Ratio 3.95	-	-	-	Ratio 2.78	-	Ratio 2.33	-	Ratio 1.91	-	Ratio 1.55

Because of practical considerations I guess I'll be using at least two ferrite rods. The aerial will be fed via a long coax cable and controlled using CAT5 wire which carries 4 pairs of wires through which power will be supplied to the amplifier, varicap control voltage from a potentiometer and relay selection (yet to be worked out) via a set of wires to miniature relays used to select the desired coil.

A simpler alternative to contructing a low capacity VLF coil might be to look through examples of chokes etc or even an ex-R1155 LF coil having the right internal diameter to fit a ferrite rod and selecting one with the best characteristics. Preset tuning to match the required tuning frequencies can be made by positioning the coil on the ferrite. A quick check on two sample coils revealed their inductances of 0.11mH and 0.01mH increased to 0.53mH and 0.08mH respectively when fitted at the end of a ferrite rod and in the former case increased to 1.21mH at the centre of the rod. The permeability factor being from 5 to 11 compared with unity for air. With the former coil a standard dust core changed the inductance from 0.11mH to 0.18mH so for the two VLF coils (15mH and 126mH) I'm looking for inductances measured in air of about 1.5mH and 12.5mH.

I found a suitable coil in my collection which was on a half-inch (12.5mm diameter) former with a dust core. I was able to carefully push the coil off the former and fit it onto a ferrite rod (10mm diameter) with just enough space to slide a paper shim under it to make it fit more tightly. This measured as 16mH when fitted at the centre of the ferrite rod.

Initially I tried a small coupling coil to inject enough signal to test the resonance of the coil, but changed to a standard medium wave coil originally fitted to the ferrite rod which gave me better results. The circuit above is the one used in my loop aerial but with slightly different components fitted. I used the 1S149 in place of the BB551 and a pair of 6.8 volt protection zener diodes in series. As you can see the circuit is constructed on a small piece of tin plated steel. The yellow capacitors are 100nF. The input was the tracking generator and the output connected to my spectrum analyser. The first thing of note is the response of the circuit is much sharper at the low frequency end and drops off as the inductance to capacitance ratio increases.

The table below shows three tests. The first is with the coupling coil close to the tuned coil, the second with loose coupling and the third using two parallel varicaps (they worked much the same wired directly in parallel or controlled via their own 100Kohm feed resistors)

Control voltage	0V	1V	2V	3V	4V	5V	6V	7V	8V	9V	10V
close coupled test signal	42	49	55	62	71	85	100	111	116	119	120	KHz
loose coupled test signal	39	47	53	59	67	81	97	107	114	115	115	KHz
twin varicaps	29	35	40	45	53	67	83	94	103	104	104	KHz

It will be relatively easy to now plug the output from the circuit into an SDR and use this to monitor the VLF band on an adjacent computer, as it's received on my workshop bench. Looking at the circuit one can see that the loading on the tuned circuit is relatively high and one improvement would be to substitute a FET, with appropriate bias resistors, for the 2N5109 transistor. That would also enable me to significantly increase the size of the 50pF coupling capacitor and hence to increase the strength of the received signal.

I tried the bench prototype as a front end for my SDR but it initially it failed to work and, despite changing coax leads, nothing would persuade my probe to work. For convenience I used the probe rather than a direct connection, but then tried a direct connection and it worked. What was wrong with the probe? I knew the batteries were long past their best but I had noticed that I'd needed to jiggle leads in the past couple of days in order to get things going. The explanation was clear once I looked into the BNC socket on the probe.. the socket had been pushed out the back of the connector. I imagine this might be a common fault that's upset experimenters for years and a dab of superglue fixed it.

Once the ferrite aerial was working into the SDR I could see MSF. Not too strong, (around -110dBm compared with -106dBm with my 80m dipole), because the aerial was sitting on the bench, but what I did see were several large spikes which reacted sharply as the ferrite rod was rotated. Of course, thinking about this, the VLF ferrite rod has a very useful secondary use. It can be used to pinpoint interference, and is much better than my long wave receiver because that doesn't tune low enough to see fundamental sources. As a test I noted a strong spike around 62KHz. The ferrite rod indicated that it was coming from my cordless house phone and sure enough when I picked this up from its rest and placed it back down the spike changed into broad pulsing noise, before turning back into a spike after about a minute or so. There was another huge spike at 58KHz (as yet unidentified) which completely disappeared when I used my 80m dipole.

Looking closely at the two pictures below it's highly likely that the Gigaset phone fundamental signal is half that seen at 56KHz because you can see a spike at around 28KHz whose size is proportional to that at 56KHz. Currently 28KHz is lower than the range of the coil under test, so that remains to be proved. Perhaps a third varicap would allow the coil to tune below this frequency?

The first picture has the ferrite rod pointed at the phone and the second at right angles. Note that some spikes will be teletype signals which are plentiful in the VLF band, with the spike in the centre of the display an artifact from processing.

This is a typical scan of the ferrite rod tuned circuit after changing the 2N5109 for a 2N3819. The latter has a very much higher input impedance allowing 2.2M and 10M to be used for bias in place of the 10K and 22K resistors and hence less input signal damping. I tried various gate coupling capcitors but decided 50pF was about optimum for response sharpness versus output voltage. I also changed the 220 ohm for an 820 ohm in the FET drain to give me a higher voltage gain.

LEFT: The tuning sharpness at 62KHz (about 4.5 volts) improved as the varicaps were tuned lower and lessened as they're tuned higher. Best amplitude was achieved at a drain supply of 10 volts.

The best signal for MSF was something like -88dBm, which compares with the best figure of -110dBm from the 2N5109. I did find however, that I was getting breakthrough of what is probably Radio 4 (198KHz) audio across the receive band so more experiments are needed.

I was looking at that picture above and wondered about the long trailing edge which reminds me of a discharging capacitor. Could the shape of the curve be associated with the DC blocking capacitor connected to the varactor diodes? If so, the 100Kohm resistor won't help so maybe reducing it will improve the shape of the curve? 5 volt x 100K x 100nF gives 10mSec but 5 volts x 100K x 10nF gives only 1mSec. As it stands I'm experiencing breakthrough from something which could well be Radio 4 on 198KHz and that curve seems to be heading for -40dB at 198KHz. Better if it was greater than -60dB.. But no.. changing the capacitor to 12nF had only the effect of altering the tuning range slightly.

I did however try a second coil, this time a standard long wave coil (about 2mH) and using the twin varicaps tuned it from 101KHz to 402KHz with a much better defined peak so maybe that falling slope above is caused by the parasitic capacitance of the coil?

To investigate the cause of breakthrough and possible spurii I'll check the FET amplifier using a signal generator. This will let me see if there are any harmonics due to non-linearity or overloading.

The latest circuit using a 15mH coil on the ferrite rod.

Note the JFET is operated under automatic bias conditions where the drain voltage determines the gate bias and hence the source current.

The varicap diodes are rated up to 15 volts and the zener diode is fitted to prevent damage if the tuning voltage were to be greater than a nominal 3 volts which can easily happen during testing.

Test signals are injected via a second coil slid onto the end of the ferrite rod.

Below: The latest tuning details. As the coupling coil is moved towards the tuned coil the frequency was observed to drop to 33KHz at 0 volts, at the expense of the highest upper frequency which I suspect is due to a mutual inductance effect. Also, swapping to the FET seems to have improved the tuning range, adding over 15KHz to the high end without materially raising the low end. .

Control voltage	0V	1V	2V	3V	4V	5V	6V	7V	8V	9V	10V
KHz	34.7	41	46	52	61	71	97	112	125	128.9	129.9

I haven't checked the current consumption of the circuit yet and I'll probably leave that exercise until I've tested for linearity (easier said than done because my choice of signal generators is not ideal as far as purity of output is concerned). Another thing I can try is taking the output from the FET (undecoupled) source.. a lower voltage swing, but this may improve the operation of the SDR and hopefully get rid of overloading on strong signals. Of course the aerial will work perfectly well with a standard communications receiver. Before further testing I'll include a couple of pictures showing conditions in the VLF and longwave bands using a random length long wire (not yet the new ferrite rod antenna).

I carried out several tests aimed at picking out coils for the other wavebands and found a long wave coil, a medium wave coil from a spare ferrite rod. The shortwave band was trickier and I wound different numbers of turns of 16SWG btc wire on the ferrite rod until I found that six turns gave me coverage of Top Band, 80m and 40m. Once this job was completed I checked the amplifier for linearity and found overall the loss between input and output was something like 20 to 30dB. The problem I believed was the 50pF coupling capacitor driven by the 50 ohm source impedance of the signal generator. After studying the circuit I realised that the small capacitor cound be removed and DC coupling used instead. This method is fairly common in IF amplifiers and I redesigned the amplifier as below.

As you can see, the coil now connects directly to the FET gate and the FET is operating under automatic bias with the bias suppy decoupled to ground via the 100nF capacitor.

Tuning is carried out exactly as before. I've dispensed with the source resistor which is unecessary and the output is taken from the drain as before via a 100nF capacitor.

During tests I found that the presence of more than one coil on the same ferrite rod resulted in odd resonance effects so that I'm probably going to use twin ferrite rods each carrying separated pairs of coils.

The next step is to rewire the amplifier to the new circuit and try that, in particular I need to recheck the linearity, because this was difficult to measure with the large losses previously.

The table below shows the latest revised figures for coils. Some tweaking will be done once the circuit board has been fitted with the two ferrite rods in the enclosure.

Range	Min Freq KHz	Max Freq KHz	Inductance	Min Tune pF	Max Tune pF	Coil pF	Notes
1	1670	9100	826uH	27 + 10	1065 + 10	10	2 varicaps
2	344	1604	300uH	35	535	?	1 varicap
3	89	466	3.67mH	52	1040 + 25	25	2 varicaps
4	24.4	122	28.4mH	85	1585	?	3 varicaps
(5)	8	32	412mH	85	1585	?	no coil available

After further testing I realised the amplifier could be simplified. The zero-bias FET drain current is governed by the drain voltage and at 6 to 10 volts is relatively low and with a supply voltage of 12 volts with a small negative bias from a 47 ohm source resistor, the drain sits at 6 volts making the drain current about 5mA which is fine for its job.

Left is the final circuit and below the results including a typical scan using the selected long wave coil and then a picture of signals received by the ferrite rod on the bench.

Redundant components have yet to be removed from the tin sheet and once the circuit is finalised with range switching relays etc it will be tidied up before fitting in the aerial case.

A scan of the long wave coil using a coupling coil of a few turns in the centre of the ferrite rod, as shown above.

The 8 turn coil is the short wave coil which tunes across top band, 80m and 40m. The tracking generator output is -20dBm so only a small loss of signal is apparent.

I'm now connecting the spectrum analyser directly to the amplifier output capacitor instead of using the probe. This is so I can determine true voltage readings.

Below, a picture of the amplifier output showing long wave signals with tuning adjusted for Radio 4 on 198KHz. The signal level of -56dBm represents 0.384mV RMS at 50 ohms. The centre frequency is 200KHz with 40KHz horizontal divisions so the two spikes right of centre are RTL Radio 1 on 234KHz and RTE Radio 1 on 252KHz.

The tuning range is 89 to 466KHz but this can be modified by sliding the coil along the ferrite rod.

Even with the ferrite rod on the workbench, Radio 4 at -56dBm is not bad considering it's -50dBm with a long wire and ground connection (see the SDR picture above)

The next part of the excercise is to assemble the amplifier and ferrite rods into their case and wire in the control cable and coax. This means salvaging the CAT5 cable from the old loop aerial and when I did this soon discovered the reason for it failing to work. The cable had been cut close to the mast (my garden strimmer!!), but fortunately leaving plenty of cable to do the job of wiring the new ferrite rod-based aerial. The new wiring will be similar to the old loop aerial.. as follows (sticking to universal colour coding where possible).

CIRCUIT DESIGNATION	PSU NEG	PSU NEG	PSU POS	PSU POS	RANGE 1	TUNE	RANGE 2	SPARE
CAT5 WIRE COLOUR	BLACK	BROWN	RED	ORANGE	YELLOW	GREEN	BLUE	WHITE

As previously I'm using two small relays to select the required range, with the added complication of adding an extra varicap diode to a specific coil if needed to provide a wider tuning range. To avoid resistive loss down the control cable I'll be using relays with suitable coils. I measured the resistance of the cable by joining two wires at the far end of the cable and measuring their loop resistance which turned out to be 5.4 ohms (this represents 28 metres of standard CAT5 cable). The varicap control voltage will be via a small 10K potentiometer as used before. A rotary switch in Setting 1 will allow Relay 1 and Relay 2 to be both inactive selecting the Short Wave range, In Setting 2, Relay 1 will be activated, Setting 3 will activate Relay 2 selecting the Long Waveband and in Setting 4 both Relay 1 and Relay 2 are acivated , selecting the VLF waveband. The spare wire in the cable is unassigned but might be used for a subsidiary function. One possible use is resetting the SDR which can sometimes lock up, requiring temporary disconnection of its power supply. Currently the aim is to have three locations viz the control box close to my computer, the aerial on my propery boundary and the SDR (or receiver) in my workshop away from main interference sources. I think the circuits below are correct?

The latest circuit diagrams

Note the addition of the 2N5109 transistor.

The new aerial is now fitted in its plastic box and prepared for final testing and adjustments. Not all the varicaps are fitted and coils not trimmed to match their design bandwidths. The first test was to check everything was wired correctly. The table below was completed by using the response on an SDR rather than the spectrum analyser. The RH metal plate is not grounded and is fitted to hold the ferrite rods in place. The brackets are left open to minimise losses.

Waveband	Min	Max
VLF	24.4KHz	140KHz
LW	86KHz	340KHz
MW	350KHz	1360KHz
SW	2.5MHz	10MHz

Everything worked, but I need a third varactor wired into the VLF circuit. The long wave coil (top right) could do with shifting higher and then the medium waveband should also be pushed up to around 1650KHz, possibly by moving the coil (top left) nearer the end of the ferrite rod. The shortwave band (bottom left) needs shifting lower but hasn't got its second varactor fitted. There's also a little room for it to be slid nearer the centre of the ferrite rod (bottom left).

So, what are the results when the aerial is used?

On the left is the medium wave broadcast station on 828KHz, Smooth Radio. Using the aerial sitting on the workbench and plugged into an adjacent SDR. with tuning and bandswitching located some 80 feet away in front of my Office PC, signal strength is indicated as -80dBm with an S/N of 76dB. The tuning response has a roughly triangular shape over 750 to 920KHz.. in other words the base level of the signal is -104dBm which slowly rises in strength to -80dBm from 750KHz and slowly dropping off until it again reaches its base level at 920KHz. Switching to the shortwave range the 828KHz signal drops to -100dBm. Turning off the power (the power supply is also remote from the aerial close to my PC) results in the signal dropping to -102dBm.

Besides obvious broadcasts indicated by their complex modulation (and in the medium waveband by their channel frequencies) you can also see several other signals. Some are carriers (solid green lines) and some are regularly pulsing (a series of dashes). Some of these do not increase in strength as the tuning control passes through them, others do. The former must be spurious responses developed within the system... maybe inside the SDR, from software or picked up in wiring between the aerial box and the SDR aerial input, or even conducted from the network wiring, the network switch in the workshop (or even the network switch close to the PC).

Testing revealed a couple of ideas to improve reception. One was the output level caused the SDR display baseline to drop below -150dBm showing a straight line at maximum gain (see the picture above.. extreme left) and the peak readings were not as high as I've seen before. The solution turned out to be relatively simple. By adding a simple 2N5109 amplifier the baseline rose and signals were amplified to a level where the SDR input could attenuated. Previously zero attenuation provided the best signal to noise ratio. With the extra transistor roughly wired in place there was a little instability (hopefully this will disappear when the circuit is tidied up) so I guess I'm getting the best results possible.

This is again Smooth Radio on 828KHz. The SDR attenuation is now 24dB instead of 0dB. The peak signal is now -48dBm to -50dBm (was -80dBm) with a similar S/N of 77dB. The tuning hump is the same shape.

As you can see, adjacent signals are also much stronger. Vertical grid spacing is 5dBm.

Switching off power to the circuit resulted in Smooth Radio dropping to exactly the same level, -102dBm, as before I added the additional amplifier. That level represents stray pickup on the coax leads plugged into the SDR. Interestingly, with the aerial in place it can be rotated so that Smooth Radio drops to virtually the baseline noise level of -125dBm so the difference between max and min signals works out at 73dB.

The next job is to tidy up the extra transistor circuit and drill holes for the control cable and coax output plus one for mounting the case on a pole. Once the extra transistor had been fitted the gain was too high causing instability and feedback cured by inserting a fixed resistor in its emitter lead whose value needed to be selected by trial and error to be 130 ohms. I think I just re-discovered a reaction control!!

As the gain was increased to the point where the amplifier was just stable over the four wavebands I noticed the shape of the tuning response sharpened up from a hump to a steep rise and fall. This means that rejection of interference is improved (the main object of the new active aerial).

If the emitter resistor of the 2N5109 was reduced to 47 ohms the shape of the tuning response narrowed considerably and oscillation took place.

Note the two signals at 800KHz and the one at 855KHz. These register -110dBm whilst before (in the picture above) they measured -100dBm, an improvement of 10dB. The noise floor is also lower in the third picture, at -140dBm compared with -135dBm previously.

At this stage the coax feed line into the aerial is rather long at around 50m.

Below are two pictures showing a set of teletype signals at the lowest tuned section of the VLF band (the tuned point is just right of the rightmost signal), the first with the ferrite rods pointing roughly N-S at 1m above ground and the second with the ferrite rods pointing at right angles to that in the first picture, roughly E-W. The top was recorded on 26th November and the second on 27th November 2020.

I've noticed a rather odd effect the reason for which I haven't established. There is a pulsing noise at about 7 beats per 5 seconds affecting some signals. You can just see on the 828KHz signals above as a set of horizontal lines on the waterfall. It's present on 252KHz, where today (27/11/2020, Radio 1 is not broadcasting, and just to the right of 198KHz. I can see a set of signals around 100KHz which seem to be the source of the noise which varies in characteristics over a period of say 30 minutes, sometimes going off, sometimes continuous pulsing and sometimes bursts of pulses. I then figured out the pulsing was present on 100, 150, 200, 250 and 300KHz, and enlarging the spectrum around 200KHz, then turning off AVC, turning the ferrite rod to minimise Radio 4, I found a set of pulses at 200, 201, 203, 205, 206, 208 and 210KHz. To minimise Radio 4 by rotating the aerial, I'd used a pair of cordless phones (with the help of my XYL watching the response) and once I'd retuned the receiver I decided to pinpoint the source of the pulses using the same method. I picked up the phone from its rest, ready to call the other handset... and the pulsing stopped. I placed the phone back on its rest... and the pulsing restarted! So there it is.. the Gigaset phones are causing the interference... Interestingly the same phone causes my wireless keyboard and mouse to also fail.

The RH picture shows the effect of lifting the phone. Oddly, this leaves a second set of signals. Are these from another phone... perhaps belonging to a neighbour, or another handset in my system (which has three handsets in total). At this point my new aerial is midway between my house and that of my neighbour. On the LH picture you can actually see a second set of pulses which remain the the RH picture.

Siemens Gigaset phone type C430A

Here you can see, by speeding the SDR waterfall, that the noise is actually FSK signalling.

Centred around approximately 102KHz I counted 10 discrete channels with each channel having two frequencies spaced by something like 150Hz, for example 102.45KHz and 102.30KHz. I can't speed the vertical speed any more but it's possible that each pulse could be carrying a code. There are about 7 pulses every 5 seconds making the PRF about 700mSec and the FSK pulse about 100mSec long.

Periodically the pattern changes before reverting to this.

In summary then, the new aerial is working very well but because of the very long feed lines there's some RF pickup getting into the receiver. There are a couple of options. Firstly I can swap my CAT5 cable and use a CAT6 cable with an outer screen (something I should use in place of my extensive CAT5 local area network anyway). Secondly I can add RF chokes and decoupling capacitors into the control lines. At present my control box is sitting only a couple of feet away from my landline phone base unit and clearly I must relocate either the base unit or the control box.

pending

Remote Active Aerial for Reception

The latest circuit diagrams

Siemens Gigaset phone type C430A

Return to Reception