June 20, 2010:
Revised: v2.0

ZL2PD Single Span HF Test Oscillator

This compact RF oscillator covers the entire 0.4 - 30 MHz range in a single sweep of the dial and has a terminated 50 ohm output of more than 300mV across the entire HF band.


Most signal generators use a series of ranges to cover the HF band. This oscillator is a little different. It tunes the entire HF band from 400 kHz to above 30 MHz in a single range. It was designed to test receiver front end designs and HF filters and be compact enough to sit around my workshop bench. The oscillator output level also allows it to be used as a temporary oscillator for testing diode mixers, and it’s sine wave output minimizes harmonics.

Circuit Description

It’s not possible to directly cover the entire HF band in one range with a traditional LC oscillator. However, mixing an oscillator operating on a higher frequency band with a lower frequency fixed oscillator, it is possible to achieve the required range. This is shown in the design’s schematic in Figure 1. A voltage controlled oscillator (VCO) operates from 48 MHz to 85 MHz. The VCO output (100-150mVpp into 50 ohms) is mixed with the output of a 48 MHz crystal oscillator in a diode mixer to give the required output.

Figure 1: Schematic of the test oscillator

The varicap diode is the key to successful wide range VCOs, and the device I used was one of four recovered from an old video recorder tuner. Other wide range varicaps such as Motorola’s MV104 or a Philips BB911 will also work well.

The 48 MHz crystal oscillator is typical of those found in equipment such as printers, video cards and the like. These provide a 5V square-wave TTL- compatible output level. I found two plastic encapsulated 48 MHz oscillators in an old Epson printer. The output of the crystal oscillator I used was unable to drive the diode mixer directly, but the series combination of C5 and R3, a 1000pF ceramic capacitor and a 100 ohm resistor, worked well. The square wave output is also ideal for diode mixers.

The use of a 48MHz oscillator, and the resulting VCO range, was largely based on the availability of suitable parts. If you are looking to substitute parts and to modify the design to suit, the VCO frequency must be high enough to permit the required 30 MHz range to be achieved within a single span. It’s unlikely that any lower VCO frequency range would be successful. Also, the crystal oscillator, which sets the lower VCO frequency bound, must be far enough away from the upper output frequency of 30MHz to allow the simple 3-pole low pass filter stage to filter off any residual 48MHz oscillator signal as well as the sum component of the mixer output. The prototype reached 35 MHz with an output rolloff of about 3dB.

The output of the SRA-1 double balanced mixer (DBM), M1, provides both difference (wanted) and sum (unwanted) products. A variety of diode-type DBMs will work fine here, including one made from 1N4148 diodes and a couple of ferrite beads. The desired (difference) output is selected using a 3-pole elliptical filter. This filter also places a notch at 48MHz to minimize any oscillator feed-through.

The filtered output is then amplified by 20dB with an ERA-5 Mini-Circuits “MMIC” amplifier to give an output of 300 – 400 mVpp into a 50 ohm termination. I used a surface mount version of the ERA-5 amplifier which is around half the size of a grain of rice. Careful soldering is required.

When operating, the unit draws just over 100mA at 12V. Not really suited to battery use, but then this oscillator is really intended for the test bench. A well-regulated supply is required since this voltage directly feeds the varicap.

Photo 2: The view inside the box illustrates the simple construction method used with a section of folded tinplate from a tin can used to form the walls of the box.


Manual tuning of a VCO across a broad range of spectrum like this often requires a multiturn precision variable wirewound resistor. With these now costing as much as $US20, I looked for a cheaper alternative. The solution lay in using a 15 turn preset resistor. These cost more like $US2 each. A further advantage of this preset resistor was the improved MHz per turn tuning rate.

To add a control knob, I used parts from an AM/FM radio volume control potentiometer. Most of these volume potentiometers seem to have a thin edge-adjusted knob which is screwed
with a tiny Philips screw onto a tiny brass rod. This, in turn, is connected to the volume control vane (or wiper) inside the pot.

I broke one apart, salvaging just the brass rod, knob and miniature screw. A couple of minutes with a file added a "flat screwdriver" end to this little brass rod. This was inserted into the adjustment slot of the multiturn preset resistor, which was also made of brass, and carefully soldered in place. Works like a charm!


I built the circuit directly onto a small piece of blank PCB in just a few hours. The diode mixer is mounted through this blank PCB, with its various pins isolated from the ground plane formed by the blank PCB as required. The 48 MHz oscillator (Epson SG-615) was mounted upside down onto the PCB. Ferrite beads (shown as ‘FB’ on the schematic) are used as RF chokes to feed DC to each stage.

The multiturn trimmer is glued onto a scrap of PCB to lift it slightly higher off the PCB, and to allow the shaft of the tuning knob to rotate freely. A slot was milled in the PCB for the knob to go through the PCB.

A box was fabricated from tin plate, cut into an 18mm wide strip and soldered around the edge of the PCB. The front panel layout was designed in CorelDraw, printed out, and covered with contact plastic to make it more durable. This was then glued onto the front of the PCB with contact adhesive.

Coil Details

L1    8 turns 24SWG on 5mm slug-tuned Neosid former, tapped at 3 turns from ground
L2    8 turns 28SWG on Amidon T25-10 toroid
L3    7 turns 28SWG on Amidon T25-10 toroid
T1    10 bifilar turns 28SWG on Amidon T25-10 toroid

General Comments

The oscillator is easy and quick to build, and uses relatively few parts. Many components can be substituted without any problems. To test this, I built another version using an LM375 IC as the VCO (It’s an obsolete National chip similar to Motorola’s legendary MC1648), a homemade DBM made with 1N4148 diodes, and a discrete 20 dB wideband amplifier. It gave similar results.

Stability is not equivalent to a crystal-locked or synthesised oscillator, and tuning across specific bands is very fast, but it’s fine for general test applications where this sort of tuning is often preferable. If you want to tune to spot frequencies regularly, an extra ‘fine tune’ control could be added.

Although I’ve not done so, it’s also easy to add a PLL chip to lock the VCO to selected 5kHz channels for better stability, or it's also possible to add a ‘huff and puff’ stabilizer. The latter is certainly less complex to build, and either approach would improve stability.

Want to go back to the main page? Click here to return directly.