Walter Zwickel, OE2TZL

Pre-mixer for 23 and 13cm

VHF Communications 1/1997

1. INTRODUCTION

The measuring equipment available to private operators is often limited to 500 MHz or below, and is thus not fully satisfactory for the 70cm band. If you nevertheless wish to include the next highest bands, 23cm and 13cm, in your measurement system, either new measuring equipment has to be obtained, or the existing equipment has to be extended by appropriate additions.

1. INTRODUCTION

Supplementary equipment is described below which increases the measurement capacity of spectrum analysers and test transmitters in the 1,000 - 1,500 MHz and 2,000 - 2,500 MHz ranges.

As radio shacks often contain elderly test transmitters giving results which are fine in themselves, but unfortunately cover only a restricted frequency range of just 500 MHz, this addition, which covers the 23cm and 13cm bands, together with their processing frequencies, brings about a welcome expansion in the possibilities of measurement. Existing wobble transmitters also expand their useful frequency range to cover these two areas. The same applies to simple spectrum analysers such as Hameg and the like. ATV amateurs can thus evaluate their output signal directly for band width, modulation symmetry, tone lowering, etc.

If you assemble two units of this kind, and if you have the appropriate tracking generator, you can also carry out wobble measurements at 23 and 13cm. The logarithmic representation and the a dynamic range of app. 60dB come fully into their own here.

With directional couplers which can be set to these frequencies, even matching measurements are no longer a problem.

2. SUMMARY

A VCO for 1 GHz is stabilised by a PLL. After a buffer stage, the signal divides into a 1 GHz fraction and a 2 GHz fraction. Both paths are selectively boosted to app. 10dBm and fed alternatively to a ring mixer through a PIN diode switch. This keeps the high frequency input signal and transmits the spectrum ranging from 0 to 500 MHz at its output. Moreover, the input and output are exchangeable, so that you can obtain upward or downward mixing, depending on the application.

3. CIRCUIT DESCRIPTION

3.1. VCO

The VCO oscillates directly at 1 GHz with a PNP transistor, so that the collector circuit can be directly earthed. Loose decoupling is provided at the emitter through 220 Ohms (Fig.2). To ensure a stable conversion value for 1 or 2 GHz, a simple PLL circuit is used, which is completely adequate for the purpose in hand. In addition, some HF is decoupled for the divider: 64 at a collector circuit tap. The type used, U664B, is highly sensitive at 1 GHz and requires only a few mV of input voltage; hence the small coupling capacitor and the tap. The weaker the connection, the less feedback through the divider!

The output signal from the pre-divider controls an integrated SO42 mixer, which has a crystal oscillator circuit at the second mixing input. The difference frequency and the sum frequency of the two signals are available at the output (pin-2). The sum signal is attenuated through the subsequent loop filter. In the case of parity of frequency between the divided VCO frequency and the crystal frequency, a DC voltage comes about as the control voltage for the capacitance diode on the VCO (mixing at frequency 0 Hz). Should discrepancies occur - due, for example, to changes in temperature or voltage - this control voltage always pulls the VCO back to the rated frequency.

The VCO is followed by a booster stage, equipped with a BFR91, since this type created the least feedback of any of the UHF transistors tested. There is already about 2mW power available at the output. This power is now divided into two paths:

a. To the final booster with T5, which is already generating just 10mW, enough to control the C-3 integrated mixer.

b. To the doubler with T3. The BFG types are indispensable here, since the low-inductive earth connection can be created only through the two emitter lugs. This is vital for high efficiency duplication.

3.2. Booster for 1 GHz

To obtain better harmonic suppression, the 1 GHz branch is again selectively laid out. This is also the reason why no MMIC's are used here. However, it is generally not a simple matter to obtain 1 GHz or 2 GHz alternatively with the same processing, since the PIN diodes used for switching no longer display ideal characteristics. For this reason, there is a switching diode in the collector circuit of this stage, which tunes out the output circuit in 2 GHz mode. If this stage is not needed, the operating voltage is also switched off as well.

Naturally, the second harmonic is relatively strong, due to the moderate quality of the strip-line resonators. However, a PI filter, fitted as an experiment to reduce the harmonic, had only a very modest success and was therefore dispensed with. The spectrum purity obtained for the 1 GHz mixing signal overall is just 40dB. If more is required, we can not do without separate assemblies, since even within the housing selection is restricted by parasitic coupling.

3.3. Doubler and Booster for 2 GHz

A standard single-phase doubler is used in the 2 GHz branch. Even with the small amount of control available, a low bias makes it possible for the doubler to operate reasonably well. At the output, a bandpass filter, together with a suction circuit, provides for the necessary selection. But the suction circuit (idler circuit) also improves the efficiency of the doubler.

If you also incorporate a second suction circuit, as described in the "Setting-up" section, the spectrum purity of the mixing signal looks considerably more favourable. With optimum setting-up, 60dB can be attained. A booster stage, which also has a BFG91 transistor, provides for the level of almost 10 mW required. This output is needed so that, even after the lossy PIN diode switchover, there is still a high enough level available to control the mixer.

3.4. Change-over Switch and Mixer

The BA479 PIN diode can still just be used in this range, but the track resistance is already clearly traceable, particularly at 2 GHz. The relevant diode is therefore located directly at the high point of the circuit, without burdening the latter excessively. A connection leg from the other diode is bent round to the 1 turn choke, and thus reduces the drain of 2 GHz signal capacitive coupling.

The C-3 SMD mixer needs an oscillator power of 7dBm (5mW). This type is given preference here because, firstly, it can be obtained at a reasonable price, and secondly, because it can safely be used at 3 GHz. Let's not conceal that it has a disadvantage as well. Compared with the well-known ring mixers in the metal housing, its matching behaviour is markedly worse. Since a simple expansion of the measurement range of this kind should not be expected to give precision of measurement to a tenth of a dB anyway, this is no disadvantage here. Moreover, with series-connected attenuators, automatic matching can be obtained at 50 Ohms. An attenuator of this kind, with SMD resistances, calculated for 3dB, can still easily be fitted in on the strip line of the high-frequency connection. We thus obtain a total mixing attenuation of 10dB pretty accurately.

At the mixer output, a twin-circuit low-pass filter ensures that higher frequency content does not have direct access to the subsequent analyser, where it could mix with harmonics to give fake signals which are completely non-existent.

The high-pass filter at the mixer input might not be absolutely necessary, but it simultaneously provides a secure DC barrier - an advantage which should not be underestimated for a good many OM's, furiously tinkering and measuring away.

Apart from this simple high-pass unit, no further image frequency filters have been provided for, since they would not have given the desired selection in the planned board format. So if image frequency suppression is required, it should be brought about by means of external series-connected high-pass units in a coaxial structure.

4. ASSEMBLY

The equipment should be assembled step by step, in functional blocks, so that each circuit component can be functionally tested as soon as it is completed. This procedure is described in Chapter 5 - Setting-up.

A printed circuit board (Fig.3) with dimensions of 148 x 54 mm., which can be incorporated into a standard tinplate housing, has been developed for the circuit.

The double-sided board has a continuous earth surface as its top. The holes on the top are slightly countersunk in the usual fashion, using a 3mm drill bit, in order to avoid short-circuits.

No holes need be created for the component connections with links to earth. Instead, these are soldered directly onto the earth surface. These soldering points are indicated by small solid circles on the component diagram.

The integrated circuit connections are an exception. The earth connections also have 0.8mm holes, but they are not countersunk. The IC earth connections are to be soldered onto the component side (Fig.4). No bases should be used for the IC's.

The transistors T1 to T5 are placed in 5mm holes, so that the shortened terminal lugs lie flat on the track in question and can be soldered (see Fig.1b).

Be careful how you mount the transistors! For T1, the lettering should face toward the soldering side, for T2 to T5 the lettering should face toward the earth surface in each case.

For T2 to T5, the small emitter legs should be bent sharply upwards and soldered on the earth side. The trimmers, Tr, are of the green SKY type. The thin, flexible little terminal lug is bent sharply through 90 degrees and soldered to earth. In each case, the round pin goes through the hole to the oscillator circuit connection.

As test assemblies of the suction circuit with L3 as a strip line were of insufficiently high quality, this circuit was executed as a wire strap on the soldering side.

After a great deal of consideration, I opted for resistors in the 1/8 or 1/10 W format. 1/3 Watt resistors can hardly be obtained in capless format now, and are thus unusable for assemblies in the GHz range. SMD resistors were not used, to make the structure easier to reproduce.

The SMD mixer is inserted into a rectangular recess, flush with the earth surface. The soldering connections face the components side. Two narrow strips and one wide strip, made of copper foil, act as the earth connection from the mixer to the continuous earth surface (identified with xxx). One leg of each of the coupling capacitors is inserted into the corresponding board hole, and the other leg is soldered directly to the square soldering surface of the mixer. These three coupling capacitors have not been shown in the components drawing, as it is difficult to depict them. The band pass filter coils are soldered to the earth surface, with a clearance of 1mm.

The tinplate housing - dimensions 55mm x 111mm x 30mm - is prepared with two high-frequency jacks, preferably SMA or SMC, and three feedthrough capacitors.

The fully-assembled and pre-balanced board can now be firmly soldered all around into the housing. With regard to the fitting height, make sure that the jack pins lie directly on the tracks in question. The earth connection for the high-pass strip line, near the high-frequency input jack, is soldered directly to the tinplate housing.

5. SETTING-UP

When the voltage controller, the two IC's, the T1 and the passive components have been assembled, a thin measuring cable is soldered onto the position of the T-2 base - still vacant - and connected to a frequency meter which measures values up to 1 GHz.

The current consumption should measure app. 60mA at +UB 12V, stabilised, with the lion's share of this, 45mA, going to the pre-divider.

If the VCO is oscillating, it should be possible to bring about a latched condition with the trimmer on T1, which is signalled on the meter by a stable reading of 1 GHz.

A DC voltage of 2 to 8 Volts is applied at the test point, TP. If the VCO has not been locked, an AC voltage can be measured here, using the oscilloscope, the frequency being the difference between the divided VCO frequency and the crystal frequency. Should there be locking problems, first check that the crystal oscillator is oscillating satisfactorily, using an anti-capacitance high-frequency probe. The : 64 divided frequency of the VCO can then be checked at pin-6 or 7 of the U664B.

Buffer stages are now assembled around T2, and the 1 GHz section around T5. A sensitive milliwatt meter is connected at the PIN diode switch through a 3.9pF capacitor. Control voltage is now fed to + UB and subsequently to the 1 GHz switching input. A maximum of over 5 milliwatts can be achieved with two new trimmers added. Check again to find out whether the VCO is actually locked. If necessary, restore this condition through a small correction to the first trimmer.

The components around T3 and T4 supplement the frequency synthesis for 2 GHz. For setting-up, couple the milliwatt meter to the diode switch again, and apply control voltage at + UB and also at the 2 GHz switching input. The remaining trimmers are adjusted to give a maximum 2 GHz signal. This is the case for trimmers which are almost switched off.

The suction circuit for 1 GHz can not be correctly balanced except with a spectrum analyser, as even without it the reduction in the unwanted signal is already better than 30dB. It is of assistance if this trimmer is also balanced to a maximum 2 GHz signal, since this circuit markedly improves the doubler efficiency.

Anyone who has the appropriate measurement equipment for setting-up can also incorporate a second suction circuit - dimensioning as L3 - on the collector circuit of T4. This circuit operates considerably more efficiently than the first one, as the filtering is carried out directly at the output to the switching diode, and thus parasitic couplings of the 1 GHz signal no longer create any interference. In this way, if setting-up is carefully done, following incorporation into the housing, a spectrum purity of almost 60dB can be attained for the 2 GHz signal. This circuit can not be adjusted except with a spectrum analyser which can, as a minimum, represent the 1 GHz signal directly. For this reason, it was not incorporated into the layout either. Finally the SMD mixer is incorporated, as described.

After incorporation into the housing, carry out a final re-setting-up of the VCO setting at a mean control voltage of approximately 5 Volts.

6. COMPONENTS LIST

not included here

7. TYPICAL APPLICATIONS

The pre-mixer described above was hooked up to a spectrum analyser as per DB1NV.

Fig.5 shows an FM-modulated ATV signal at 2,410 MHz, together with a sound sub-carrier, at 6.5 MHz. It can also be clearly recognised that the modulation symmetry is imperfect, conditioned by the characteristic of the capacitance diode in the ATV control transmitter.

In this case, the spectrum analyser output signal was fed to two A/D converters in the PC, to make it easy to save and print out test certificates.

HOME PAGE