What we do
Amateur Radio

Some modifications to the R3500D receiver for use on top band.


The R3500D (also known as the PJ-80) is a simple direct conversion radio receiver intended for use in amateur radio direction finding events on the 80 metre band (frequency range approximately 3.5-3.6 MHz). These notes cover some modifications to the receiver to enable use on top band, in our case 1.843 MHz. They also identify some avenues for further experimentation.

How does it work?

Before we deal with any modifications, let’s take a whistle stop tour of the circuit to see the key features. Figure 1 shows the schematic.

Figure 1: R3500 schematic

The receiver uses a ferrite rod antenna L1 which is tuned to frequency by means of the capacitor CT . The signal is then coupled by means of L2 and C1 into the base of transistor V1 which amplifies the radio signal. There’s also a sense antenna which can be switched in and out of circuit by means of S1. The ferrite rod antenna is directional but for a full rotation of the antenna, there will be two maxima and two nulls making it impossible to determine the direction of the transmitter without taking a further bearing. The sense antenna resolves this 180 degree ambiguity.

The base-emitter bias (Vbe) of transistor V1 is set by a voltage divider formed by R2 and R3 but we note in passing that the bias voltage is affected also by the adjustable potentiometer RP1-1. This adjustment runs in parallel with the adjustment (RP1-2) on the audio amplifier chip because a dual gang potentiometer is used here. The adjustable bias on the base of transistor V1 allows the gain of the amplifier to be varied. Figure 2 shows an example of the relationship between the base-emitter bias voltage Vbe and the collector current Ic. We can see from the figure that nothing much happens until Vbe rises to about 0.6 volts at which point Ic increases in a rising curve with increasing Vbe until an approximately straight line relationship is reached. Varying the bias point on the curved portion of this line allows us to adjust the degree of amplification.

Figure 2: Example relationship between base-emitter voltage and collector current in a transistor

On the collector of transistor V1 there is not only resistor R6 but also an inductor formed by the primary of T1 paralleled by a capacitor C3. And on the emitter there is capacitor C2 in parallel with resistor R4. Both the collector and emitter circuits are frequency sensitive, particularly the tank circuit on the collector.

The amplified signal is coupled into the next stage of the circuit via the secondary of transformer T1. It is here that the inaudible pulses of radio frequency (rf) are made audible. The rf signal appears at the anode of the germanium diode VD1 along with the signal from the oscillator which is shown in the circuitry in the lower half of the schematic. We’ll have to look at this in more detail when it comes to modifying the radio for top band but for now we’ll simply say that the oscillator and rf signals are ‘mixed’ by the diode and one of the outputs is an audio signal. The output of the mixer in a direct conversion receiver typically consists of a whole host of signals that differ widely in their strength and frequency. To obtain top performance, the mixer should be properly terminated at all of the output frequencies [see refs 1, 2 and 3], a process which may involve the use of a carefully designed diplexer to ‘split’ unwanted low frequencies from unwanted high frequencies (rf) and the (wanted) audio and to present each group of frequencies with the correct terminating impedance. In the present case a simple approach has been adopted. The combination of C5, C6 and R7 provides low pass filtering by grounding high frequency signals via capacitors C5 and C6. The following combination of C7 and R8 provides high pass filtering. So, taken together, these parts of the circuit form a band pass filter for the audio signals which are fed to the audio preamplifier consisting of transistor V2. We can gain some idea of the effects of the ‘audio’ section of the circuit by creating a model using LTSpice. Figures 3 and 4 show the circuit and predicted response respectively. A 2N2222 transistor was used in the model rather than a 9014 because a model of the former was readily available in the LTSpice library. But a range of other transistors was also used and the choice did not affect the response in any significant way so the results are probably broadly indicative of the operation of the R3500D circuit. Those who wish to check this can prepare a model for the 9014 and run the simulation for themselves!

Figure 3: LTSpice model of audio circuit (note – using 2N2222 transistor)

Figure 4: Predicted frequency response of audio circuit

From the 10k potentiometer RP1-2, the signal is fed into the audio amplifier chip, the output of which drives the headphones.

Well, that wasn’t too bad, was it? But we’ll have to use this information to adapt this receiver to work on top band.

Adapting the R3500D for top band use

To adapt the rig, we need to look at those parts of the circuit which are critically important in determining the radio frequencies that are received. They are summarised in Figure 5. We’ll start at the antenna and work our way ‘downstream’.

Figure 5: Key parts of circuit determining the radio frequencies received


The ferrite rod antenna forms part of a tank circuit – an inductor and capacitor in parallel. The information provided with the receiver indicates that it is designed to cover around 3.5 – 3.6 MHz. We note that the capacitor CT is a trimmer capacitor, adjustable by means of a screwdriver to tune the receiver before the build is completed but not designed to be adjustable during regular use. The ‘front end’ of the receiver is therefore not very discriminating but even so, the frequency we need (1.843 MHz) is so far outside the design specification that we’ll need to modify the circuit.

For a tank circuit, the relevant equation is:
Frequency = 1/(2π{LC})
where L is the inductance (in Henries) and C is the capacitance (in Farads).

We might expect the 3dB bandwidth to be at least 100kHz which requires a Q (quality factor = centre frequency/bandwidth) of no more than about 3.55/0.1 = 35.5.

Our desired receive frequency is about half that for which the receiver is designed. From the equation above we can see that we’ll need to increase the capacitance or the inductance fourfold to do this (or increase both the capacitance and the inductance by some lesser amount – e.g. doubling them both). For purely practical purposes, we’ll increase the capacitance because this is a simpler procedure. Changing the inductance would involve increasing the windings on the ferrite rod. But, in passing, we’ll note that keen experimenters might like to compare the results achieved by these two approaches and some comments are added in Box 2. Measurement of L1 indicates an inductance of about 160 µH (depending on exactly where the sliding coil L1/L2 was placed on the ferrite rod). So, from the equation in Box 1, we’d need a total capacitance of about 47 pF to tune the antenna to 1.843 MHz. We already have an adjustable amount of capacitance of about 5 – 20 pF so padding a 33 pF capacitor across the adjustable one should do the job and allow us a bit of scope for tuning around within top band.

After this modification, what does the frequency response of the antenna actually look like? To investigate this, a signal from a signal generator was coupled into the antenna by means of two turns of wire around the ferrite rod a few millimetres from one of its ends. The sliding coil L1/L2 was placed in the centre of the rod and the tank circuit consisting of the coil and capacitor CT (plus the additional 33 pF capacitor) was disconnected from the following circuitry for the purposes of this experiment. The output of the signal generator was set arbitrarily at 1 V and the frequency was adjusted. Using an oscilloscope, the peak-to-peak voltage seen across the tank circuit was measured. The frequency response is shown in Figure 6. The antenna response shows a nice peak at the frequency we want, with a -3dB bandwidth of about 110 kHz. These measurements suggest a circuit Q of about 17 when unloaded by the following stages.

Figure 6: Frequency (x-axis, in MHz) vs voltage across antenna tank circuit (y-axis, in millivolts).

Tuned collector

The next stage we need to look at is the tuned collector. Here we have another tank circuit that we need to tune to a new, lower frequency. As with the antenna, the easiest way of achieving this is by increasing the capacitance. Measurements across the primary of transformer T1 indicate an adjustable inductance of between 23 and 48 µH. Using the equation shown in Box 1, we can see that if we increase the capacitance from 47 or 68 pF (as shown on the original circuit diagram) to 220 pF, we should be able to achieve resonance at 1.843 MHz with an inductance of about 33.9 µH, nicely within the range of adjustment offered by the transformer. This was found to be the case when peaking the receiver during final testing.


A standard direct conversion receiver has an oscillator that operates on a frequency that differs from the desired receive frequency by only a very small amount (typically around 700 Hz). When mixed with the received signal, one of the outputs from the mixer is the difference between the two frequencies (ie the offset) and it is this frequency (a nice audio tone of say 700Hz) that is then amplified by the subsequent stages of the receiver and heard by the user. But the R3500D receiver is unusual in this regard because it is designed to have an oscillator that runs at half the receive frequency (ignoring the small offset mentioned above). So does this mean that we have to change component values in the oscillator to reduce its operating frequency in a similar way to what we did in the previous circuits? We note that our desired receive frequency of 1.843 MHz happens to be half the frequency for which the R3500D is designed to receive so as it happens, the unmodified R3500D oscillator will already be operating on a frequency that would be chosen for a ‘normal’ direct conversion receiver on the frequency we want. We need to consider this in more detail before deciding what to do, having particular regard to how the oscillator and mixer work together (is the mixer a special one that will only work with a subharmonic oscillator?) and what the pros and cons are of the ‘normal’ vs subharmonic designs of direct conversion receivers. Many treatments of subharmonic mixing deal with something called the ‘Polyakov detector’ [e.g. Refs 5 & 6], an arrangement of two antiparallel diodes. But the mixing in the R3500D is performed by a single diode VD1, a very simple arrangement that is unusual nowadays except in microwave applications [Refs 7 and 8]. How does this actually work? Is there something special about this particular arrangement that means it only works as a subharmonic mixer? Is the oscillator producing a ‘dirty’ signal with lots of harmonics and the diode is actually mixing a harmonic signal instead? Or is something else going on?

A standard model of diode operation is what’s known as the Shockley diode equation, written in its simplest form as:
I = IseV/(kT/q)
where I = diode current, Is = reverse bias saturation current, V = diode voltage, k = Boltzmann constant, T = temperature and q = magnitude of charge of an electron.

At a typical room temperature, this approximates to:
(= IseV/0.026)

Using this equation, a trigonometric identity, a series expansion and a bit of algebra, we can see that a single diode operating in accordance with the Shockley diode equation will produce a whole host of mixing products including the difference between the received signal and both 1) the oscillator frequency and 2) twice the oscillator frequency (ie a doubling process - see Appendix for further details). So, based on this analysis, it seems that the diode mixer in the R3500D should work both as a subharmonic and ‘normal’ mixer.

Looking at the matter from another angle, we can try creating an LTSpice model of a simple diode mixer and see what happens when it is fed by a subharmonic oscillator signal and also an oscillator at receive frequency (with a small audio offset) in the conventional way. Figures 7 and 8 show the results of a diode (in this case a 1N5817) being fed by two voltage sources and driving a resistive load of 50 Ohms. One voltage source is set to 1 V and represents the oscillator output. The other is set to 10 mV and represents the rf signal at 1.843 MHz. Both voltage sources are set to produce a perfect sinusoidal output. Figure 7 shows the output when the oscillator is running at 1.8423 MHz. Figure 8 shows the output when the oscillator is set to half the frequency of that in Figure 7 (ie 921.150 KHz). In both cases there is a 700 Hz audio signal of similar magnitude although the output in the subharmonic case (Figure 8) departs markedly from a sinusoidal form (unlike that in Figure 7).

Figure 7: LTSpice model showing audio (700 Hz) output at a receive frequency of 1.843 MHz when oscillator is set to 1.8423 MHz

So much for the theory, but what happens in practice? First, let’s have a look at the output from the oscillator. Figure 9 shows what is seen on the oscilloscope. The output approximates a sine wave.

Figure 9: Output from oscillator, measured between ground and the anode of diode VD1. Each major vertical division represents 0.1 V. Each major horizontal division represents 0.1 µS.

To investigate things further, an attempt was made to compare subharmonic vs conventional mixing in terms of minimum discernible signal. A signal was coupled into the antenna using a signal generator as previously described. In one case, the oscillator was running at the receive frequency (minus audio offset). In the other, the frequency of the oscillator was halved by padding extra capacitance (680 pF) across C10. The sensitivity of the receiver appeared to be marginally increased (by about 3 dB) when the oscillator was running at receive frequency compared with when it was running at half the receive frequency. The experiment was repeated using a different germanium diode (1N60) which had a slightly higher forward voltage drop than the 1N60 (303 mV vs 289 mV) and which was found to be slightly less sensitive as a mixer/detector. In this case the receiver appeared to be about 6 dB more sensitive when the oscillator was running at receive frequency compared with the subharmonic modification. There is clearly an opportunity for further experimentation here, one improvement being to meas-ure the voltage across a load at the audio output under the different sets of circumstances rather than using the ears as a measuring instrument. Another avenue of investigation would be to use an antiparallel diode pair as the mixer rather than a single diode whilst running the oscillator at half the receive frequency (e.g. along the lines of the designs in Refs 2 & 9). Does this yield a worthwhile advantage? Possibly, but this form of mixer is said to be very sensitive to the LO drive level (Wes Hayward W7ZOI, pers. comm.) and the magnitude of the output from the oscillator may have an effect on the relative efficiency of these different methods. Practically speaking, opportunities for increasing oscillator output are probably quite limited within the constraints imposed by the size of the plastic casing of the receiver.

Based on the results reported above, what approach should be adopted? It appears that the re-ceiver may be marginally more sensitive when the oscillator is run at receive frequency rather than as a subharmonic oscillator. But if that really is the case, why would a designer ever make a receiver with a subharmonic mixer? Rob Kalmeijer PA3CJD [Ref 2] argues that the main advantage of the subharmonic mixer (at least in antiparallel diode form as designed by Vladimir Polyakov RA3AAE) is that it minimises leakage from the oscillator. In a ‘normal’ direct conversion receiver some of the signal from the oscillator ‘leaks out’ and is transmitted via the antenna which is tuned to the same frequency as the oscillator. But there is better isolation if the oscillator is running at half the receive frequency because the antenna radiates the signal much less efficiently. Other authors also claim additional benefits from a subharmonic mixer including improved oscillator stability. In the present case, re-radiation of the signal from the oscillator is not a big problem on club DF hunts. It is certainly detectable over a close range, but does not present real difficulties for other participants and falls below the ambient rf noise levels once one is more than a few metres away from the receiver. Also, the oscillator of the R3500D seems sufficiently stable as it is. So, from the point of view of simplicity, it seems appropriate to leave the oscillator frequency as it is and to run the modified R3500D as a conventional rather than subharmonic receiver. But for those who want to squeeze the best performance out of this simple receiver, there are various avenues for further research.....


  1. Rick Campbell KK7B. High-Performance Direct-Conversion Receivers. QST August 1992, pp19-28. https://www.arrl.org/files/file/Technology/tis/info/pdf/9208019.pdf
  2. Rob Kalmeijer PPA3CJD website: http://www.robkalmeijer.nl/techniek/electronica/radiotechniek/hambladen/radcom/1991/04/page39/
  3. VK6FH website: http://www.vk6fh.com./vk6fh/DIPLEXER 1.htm
  4. Chavdar Levkov LZ1AQ website: http://www.lz1aq.signacor.com/docs/fa-eng/Weak_signals-mag_loop_engl.htm
  5. Rick Andersen KE3IJ website: http://www.ke3ij.com/DC-80.htm
  6. Roger Lapthorn website: http://g3xbm-qrp.blogspot.co.uk/2012/08/28mhz-polyakov-mixer-rx.html
  7. Wes Hayward W7ZOI. Introduction to Radio Frequency Design. American Radio Relay League.
  8. Wes Hayward W7ZOI, Rick Campbell KK7B & Bob Larkin W7PUA. Experimental Methods in RF Design. American Radio Relay League.
  9. QRP technical notes. http://noding.com/la8ak/c21.htm

Appendix 1

  1. I = IseV/(kT/q) (= IseV/0.026)
  2. e(x/a) = 1 + (x/a) + ½!(x/a)2 + ⅓!(x/a)3 + ¼!(x/a)4 + .....
    For equation [1], at any instant the voltage presented to the diode is the sum of the signal voltage Vs and the oscillator voltage Vo
  3. V = Vs + Vo
    Assuming sinusoidally varying signal and oscillator voltages and writing the amplitude of the signal as As, the frequency of the signal as Fs and time as t:
  4. Vs = Ascos(2πFst)
    Similarly for the oscillator and writing the amplitude as Ao and frequency as Fo:
  5. Vo = Aocos(2πFot)
    So, from [3], [4] and [5]
  6. V = Ascos(2πFst) + Aocos(2πFot)
    It follows from [1] and [6] that
  7. I = Ise[Ascos(2πFst) + Aocos(2πFot)]
    Equation [7] contains e raised to a power and we can expand this in the form of a power series [2]. Focusing on the cubic term ⅓!(x/a)3, gives
  8. (⅓!)(1/0.026)3[(Ascos(2πFst))3 + 3(Ascos(2πFst))2(Aocos(2πFot)) + 3(Ascos(2πFst))(Aocos(2πFot))2 + (Aocos(2πFot))3]
    Equation [8] contains four groups of terms. Concentrating on the third group,
  9. 3(Ascos(2πFst))(Aocos(2πFot))2 can be written as:
    3(Ascos(2πFst))(Aocos(2πFot)) (Aocos(2πFot))
    = 3/2AsAo2[cos(2π Fst - 2πFot) + cos(2π Fst + 2πFot)] cos(2πFot)
    = 3/4AsAo2[cos(2π Fst - 2πFot- 2πFot) + cos(2π Fst- 2πFot + 2πFot) + cos(2π Fst + 2πFot - 2πFot) + cos(2π Fst + 2πFot + 2πFot)]
    = 3/4AsAo2[cos(2π Fst - 2π(2Fo)t) + cos(2π Fst + 2π(2Fot) + 2cos(2πFst)]
    From [9], we can see that amongst this plethora of mixer products are terms that involve the signal frequency minus twice the oscillator frequency and the signal frequency plus twice the oscillator frequency.

Appendix 2 – some ideas for the experimenter

Changing the windings on the ferrite rod would enable both L1 and the coupling inductor L2 to be varied. The strength of a signal received by means of a ferrite rod antenna is tiny (possibly several thousand times smaller) in comparison with what might be expected from an equivalent half wave dipole (Ref [4]) and this can result in weak signals being lost in the general background noise. Work by Chavdar Levkov LZ1AQ (Ref [4]) indicates that the signal to noise ratio in a ferrite rod increases linearly in proportion to the number of turns on the ferrite rod (L1 in our case) but is inversely proportional to the square root of the bandwidth and the square root of the resistive losses in the winding. This suggests that an increased number of turns might improve the signal to noise ratio since the beneficial effect of increasing the number of turns should outweigh the adverse effect of increased resistive losses. But increasing the resistive losses would, in turn, also affect the bandwidth. Furthermore, increasing the number of turns on L2 would increase the signal (and its associated noise) to be amplified by transistor V1 which may be a good thing but, on the other hand, it would increase the loading on the tank circuit which would act to further increase its bandwidth. In addition, any such changes seen from the base of the transistor may affect the noise figure of the transistor itself since this component will add noise of its own to the signal which it amplifies. Further investigation could be directed towards changing the ferrite rod since the permeability of the rod and its size will also affect the size of the received signal (although for our purposes the size is more or less fixed and determined by the size of the plastic case of the radio). There is obviously more scope for investigation and experimentation here.....


Richard (M0HNK)

About Us
Contact Us