DIY 6 GHz comb generator

Kenneth Wyatt- November 27, 2012

Most simple DIY comb generators seem to run out of steam about 1 GHz. I recently ran into David Bowman's 2.4 GHz circuit and measured an upper usable range of about 6 GHz. While greatly attenuated above 3 GHz, this circuit should still be valuable for measuring semi-anechoic chambers in the GHz ranges.

Referring to Figure 1, the design starts with a 96 MHz crystal oscillator (third-overtone crystal), capacitively-coupled to a standard MAR3 broadband amplifier, and capacitively-coupled to a back-to-back set of low-capacitance PIN diodes. If you wanted the harmonics to be on 100 MHz centers, you should easily be able to substitute a 100 MHz third overtone crystal. This circuit was originally designed to produce a reference signal in the 2.4 GHz amateur radio band; hence the 96.013 MHz crystal. The circuit is designed to run from 15 volts, which is regulated down to 12V using a 78L12 linear regulator.

Specific part details and schematic may be downloaded from David's web site: http://www.g0mrf.com/source2.htm.

For those wishing to build one of these, the whole kit, including circuit board, is available from the AMSAT-UK Online Shop (http://shop.amsat.org.uk/shop/catalog/browse?shop_param=), then click on Hardware and Signal Sources). The price is 15 GBP, plus 3 GBP shipping (to the UK). Shipping for the U.S. is an additional 3.50 GBP, but there seems to be no way to add U.S. shipping within their order process, so I simply donated an extra 5 GBP (the smallest amount available) on their regular web site (http://www.uk.amsat.org) and that made them happy.

The BFS17 transistor, MAR3 and HSMP-3822 PIN diode pack are all surface-mount devices, so extra care is required to solder them to the PC board. I used tweezers to hold the package while I tack-soldered one lead. After soldering the remaining leads, I reflowed solder on the first lead. Note that the transistor is mounted to the bottom-side of the board. While there is no silk-screening, it should be obvious where it mounts. The board is designed so that a standard PC board edge-mount SMA connector may be soldered directly on the output trace, as shown in Figure 2. The lead with the diagonal cut is pin 1 (input) of the MAR3. Once all the parts were soldered in, I powered it up and everything worked as advertised. I did have to peak up variable inductor, L1, slightly to maximize the output. Be sure to use a plastic hex tool for this, as a metal tool can fracture the ferrite core.


Figure 1 - A block diagram of the 6 GHz comb generator.

A very similar 10 MHz-based circuit may be found on the discussion board, edaboard.com (http://www.edaboard.com/thread200668.html), although, because the oscillator frequency is so much lower, I doubt this circuit will get you past 1 GHz.


Figure 2 - The completed 5 GHz comb generator board, ready for test.


Figure 3 - A typical harmonic output from the comb generator.

One experiment to try might be to install a simple high-pass filter (with cutoff frequency about 500-1000 MHz) to equalize the high-amplitude low-frequency harmonics, making the comb response a little flatter.

With the on-board regulator, this self-contained harmonic comb generator should be very stable and could be used to at least partially verify the day-to-day performance of your semi-anechoic chamber. For more on harmonic comb generators, see my article in RF Technology International (http://rfti.com/wp-content/uploads/2012/05/RFTI0412_Wyatt.pdf).