How was testing done for Ghz to THz range circuits and devices before fast enough scopes and frequency counters existed?

  • 2
    \$\begingroup\$ "No measurement equipment is capable of signal measurement at these ranges." Where did you get that idea? Here's a high frequency scope: teledynelecroy.com/100ghz A few seconds with google reveal innumerable pages on mmwave measurements. \$\endgroup\$
    – user133493
    Dec 24, 2017 at 6:55
  • 3
    \$\begingroup\$ Ok, but lets go to a higher speed then, how do you verify the operation and diagnose it if it fails if measurement equipment doesn't exist? Surely there were Ghz circuits before Ghz scopes and even Ghz frequency counters. \$\endgroup\$ Dec 24, 2017 at 7:00
  • 5
    \$\begingroup\$ That's not a helpful comment, \$\endgroup\$ Dec 24, 2017 at 7:15
  • \$\begingroup\$ This is an interesting question, since it touches one of those issues that seems like a classic chicken-and-egg problem, often recurring in bleeding edge-engineering. I hope to see some interesting answers. \$\endgroup\$ Dec 24, 2017 at 8:52
  • 1
    \$\begingroup\$ @LorenzoDonati Eggs predate the chicken by thousands of years. Reptiles and fish laid eggs before birds flew and chickens in particular even existed. \$\endgroup\$
    – winny
    Dec 24, 2017 at 11:17

4 Answers 4


For some perspective, consider that optical signals are still too high frequency for the instantaneous electric field to be sampled and measured, but there are still lots of different kinds of measurements we can do on an optical signal.

  • With a power sensor (a photodiode or even an LDR) we can measure the power of the signal.

  • With a prism or diffraction grating we can build a spectrometer and get a rough idea of the signal's spectrum and/or pulsewidth.

  • With an interferometer we can mix the optical signal with a delayed version of itself and measure the coherence time (bandwidth) of the signal with perhaps gigahertz resolution.

  • With a tunable local oscillator (laser), we can even down-mix the signal and measure its spectrum with an RF spectrum analyzer, getting 100's of kHz resolution.

All of these measurements have analogs in the microwave regime and were or could be used by microwave engineers prior to the advent of multi-gigahertz oscilloscopes.


Long ago they relied of the speed of Gunn diodes for sampling the input signal waveform with a control pulse duration so that the difference frequency could be displayed on a slow timebase oscilloscope. If the sample duration was short enough to capture only the point on a recurring waveform the the waveform was preserved.

Gunn diodes were useful since they had low negative resistance so once triggered, that would accelerate then hold the result once the bias charge was depleted.

The key to reception of a frequency higher than can be observed or detected is to use imaging down-conversion to a useful IF frequency or direct to base band depends on the conversion efficiency, power level and SNR.

Methods such as interferometry, Diode detectors, pulsed samplers, where the harmonic of the sampling rate has sufficient harmonic energy in the band of interest.

Nonlinear mixers such as; "high temp" step-edge Josephson junction, varicaps, GaAs diodes and heterobarrier varactors (HBV) or optical pump with extreme fast rise times from small inert gas arc gaps.

These aliasing down-conversion type scopes were called Sampling oscilloscopes. ( but only useful for repetitive waves) enter image description here

further reading

  • \$\begingroup\$ This is interesting and explains how it works, I suppose you can see if it's working right, but can it be used to reconstruct broken waveforms? Seems challenging. \$\endgroup\$ Dec 24, 2017 at 23:31
  • \$\begingroup\$ @FourierFlux, it certain can be used to reconstruct a waveform. The Keysight 86100 series sample at 40 kSa/s, but can reconstruct signals with bandwidth to 80 GHz. \$\endgroup\$
    – The Photon
    Dec 25, 2017 at 4:58
  • \$\begingroup\$ How though? You only sample a finite number of points and without some constraints on the input waveform you can't really say anything. \$\endgroup\$ Dec 25, 2017 at 5:05

'Fast enough' oscilloscopes are a trick for displaying signals that vary in time, but they aren't the only trick. A 1 GHz oscillator, for example, will heat a resistor. It will also resonate with a cavity length of about 120mm (which can be determined by sensing the heating of resistors). The combination is called a 'wavemeter'.

A crude wavemeter is a length of wire placed on a paper plate, in a microwave oven. The (about two inch) right length of wire gets much hotter, and scorches the plate to a darker color, than other wire lengths.

You can tell the frequency, without a 'frequency counter', of light by using a diffraction grating (a blank CDROM has 1 hour playing time, at 1 revolution per second, so you can measure the band with a ruler and use it to diffract a laser beam...) and measure the wavelength, thus (knowing the speed of light) the frequency.

If you have a non-sine-wave, the various harmonics will ALL show up, and with a little care in measurement one can identify square and triangle waves.

Most people wouldn't call that CD blank a 'measuring instrument', but it does the job. It just isn't convenient and precalibrated. Neither is the paper plate in the microwave oven (and if you value the flavor of your food, you need to clean out smoky byproducts).


There are many ways to analyzed terrahertz device so long one is not too interested in the precise time domain information. You can always use a mixer/downconver, and perform digitization, and analysis on frequency domain.

A company call Virginia Diode produce such mixer.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.