Triangle waves drive simple frequency doubler
Extend a function generator's range, and get a sine wave signal in the bargain.
Jim McLucas, Longmont, CO; Edited by Brad Thompson and Fran Granville -- EDN, November 23, 2006
If you use a function generator, you may occasionally require a sine-wave output at a higher frequency than the generator can provide. If your function generator also produces a triangle-wave output, you can use a frequency doubler to extend the generator's available frequency by as much as a factor of two. A previously published Design Idea describes a triangle-wave-driven frequency-doubler circuit employing op amps that produce output frequencies limited to about 20 kHz (Reference 1).
This Design Idea describes a frequency doubler that provides a sine- wave output with a frequency of 4 to 6.7 MHz, with an output level that can range from 110 mV p-p to 1.30 V p-p into a 50Ω load. As Reference 1 describes, applying a symmetrical triangle wave to a full-wave rectifier produces a triangle wave of twice the input frequency and offset by a dc level. Any asymmetry in the input waveform allows some of the input signal's fundamental frequency to pass through to the output. Also, the circuit's input transformer, T1, may cause amplitude or phase imbalance, allowing some of the input signal to pass through to the output.
To construct a wideband transformer with good amplitude and phase balance, twist three AWG #30 enameled wires together at about 10 twists/in. Wind seven turns of the bundled wires onto a Fair-Rite 2643002402 toroidal core. (Each pass through the core's central opening counts as one turn.) Connect the wires as shown in Figure 1. (Refer to Reference 2 and Figure 2 for additional information on this type of transformer.) This technique results in a wideband transformer with good amplitude and phase-balance characteristics.
To achieve maximum input-frequency attenuation, use a matched pair of Schottky diodes for D1 and D2. However, the prototype produced high-quality signals with unmatched Schottky diodes. In Figure 1, diode D3 applies a small negative bias to D1 and D2 that allows operation at low signal levels. Capacitor C1 passes the rectified and frequency-doubled triangle wave to the bases of a complementary emitter follower comprising Q3, Q4, and associated components. A simple, two-element lowpass filter at the follower's output removes higher frequency harmonics. Use any 1.6-µH inductor with a Q of 20 or greater for L1. Although an inductor with a Q as low as 10 will not noticeably change the filter's frequency response, a value lower than 20 increases the inductor's insertion loss and decreases the maximum available output-signal amplitude.
A simple, two-element, lowpass output filter provides adequate performance for a symmetrical-triangle-wave input because the output's frequency components consist of the doubled input frequency signal and only the desired output signal's odd harmonics. For a 5-MHz output, the third harmonic occurs at 15 MHz with an amplitude of –19 dB relative to the 5-MHz signal. The lowpass filter imposes 15 dB more attenuation at 15 MHz, diminishing the 15-MHz signal to –34 dB relative to the 5-MHz output signal and attenuating higher order harmonics to even lower levels.
The complementary emitter follower's unfiltered output signal consists of a triangle wave of twice the input signal's frequency, plus odd harmonics of the doubled input frequency. For example, applying a 2.5-MHz triangle wave to the circuit's input produces a 5-MHz triangle-wave signal at the lowpass filter's input. For a nearly perfect triangle wave, the filter's input consists of a 5-MHz fundamental and only its odd harmonics. At –19 dB below the 5-MHz signal, the 15-MHz third harmonic represents the closest spurious signal and one that you can easily filter.
To use the circuit at higher frequencies, divide the values of output-filter components L1 and C8 by a factor of FNEW/5, where FNEW represents the desired output frequency in megahertz. For example, a nominal output frequency of 20 MHz requires division of the values of L1 and C8 by a factor of four, producing new values of 0.4 µH and 140 pF, respectively. Simulating the circuit with the revised filter in Spice shows adequate harmonic rejection over an output range of 16 to 26.8 MHz. Although designed for 5-MHz operation, the remainder of the circuit works well at 20 MHz without additional modifications. This frequency doubler also accepts a sine-wave input signal. However, the circuit's unfiltered output contains higher levels of the desired signal's even- and odd-order harmonics and requires additional filtering to produce a high-quality sine-wave output.
| References |
|
-
There are a couple of errors on the schematic of my Design Idea.
(1) R6 should be 1500 ohms, not 500 ohms.
(2) In the description of inductor L1, the core is manufactured by Amidon, not Amicon.
Regards,
Jim McLucas
Jim McLucas - 2006-27-11 17:32:00 PST -
Hi Alex, Thanks for your feedback. The idea of using a triangle wave signal to drive a frequency doubler is a clever one, and it allows the implementation of the doubler to be quite simple.
I was most interested in higher frequency applications; hence my focus on a practical high-frequency application.
Best regards,
Jim
Jim McLucas - 2006-27-11 17:20:00 PST -
Hello Jim, Thanks for your kind attention to my old EDN design idea, mentioned in the Reference 1: "Frequency doubler operates on triangle waves," EDN, March 14, 1996 (well, it’s been a 10-years anniversary). I am glad that you found the new use for my old design concept (“The general idea is to apply the triangle waveform to any full-wave rectifier….”– as it was stated in Ref 1).
As a little amendment to your interesting article I just want to point out, that in my original design this concept was combined with another one, namely: building a full-wave DIODE-LESS rectifier based on a Rail-to-Rail Op Amp in a single-supply mode. Component selection and some topology specifics lead to the limited frequency range. This limitation is quite easy to overcome by choosing any fast full-wave rectifier of a different topology (e.g., with diodes); I have seen several ones, published in the EDN within past 10 years. Their frequency range could be pretty wide, and the implementation, in general, is rather simple: couple Op Amps, two diodes and a set of matched resistors.
My best,
Alex B.
Alex B. - 2006-24-11 14:58:00 PST
