What difference does a probe make?

When I bought my 1 GHz LeCroy digital oscilloscope in June of 2012 (omg, was it really almost 7 years ago??) it came with a set of four LeCroy PP007-WR probes. These have served me well but they have limitations. The first is that they're only rated to 500 MHz, the second is that their input capacitance is 9.5 pF.

Why on earth would that matter? After all, I'm not working with 1 GHz signals, so why would a 500 MHz limitation be a problem? And 9.5 pF is practically nothing.

Well, when working with digital signals you're dealing with square waves, not sine waves. A square wave consists of the fundamental frequency plus the odd harmonics of the fundamental frequency. Thus a 1 KHz square wave is the sum of a 1 KHz sine wave, plus a 3 KHz sine wave, plus a 5 KHz sine wave, plus a 7 KHz sine wave, and so on. As the frequency of the harmonics increases the amplitude decreases, but they're there (though by about the 9th harmonic they're pretty tiny). There's a nice audio demo buried in the technobabble of the Wikipedia article on square waves.

This is why I was told that if I was going to work with digital circuits that I should invest in a 100 MHz oscilloscope. Of course I was told that in the 1970s, when most digital circuits ran at frequencies below 10 MHz. These days, even system-on-chip microcontroller run that fast. The main clock for the Spartan 6 FPGA on the Digilent Atlys board runs at 100 MHz.

Now let's scale this up a bit. If I probe the output of the Epson SG8002JF crystal oscillator on the Atlys board with my 500 MHz probe I see a nice sine wave. What's wrong with that? Well, the output of that oscillator isn't a sine wave, it's a square wave with rise and fall times of less than 4 nanoseconds. But with a fundamental frequency of 100 MHz, the 3rd harmonic is 300 MHz and the 5th is 500 MHz. If the probe's frequency response was good to even 300 MHz I'd expect to see flattened or wiggly tops, but I don't. Clearly even at 300 MHz this probe's response is significantly diminished.

PP007 Typical Input Impedance

Also consider the loading on the signal being probed. The input impedance of a passive probe diminishes as the frequency of the signal goes up. By the time we get above 100 MHz the PP007 probe's impedance has dropped from 10 MΩ to 100 Ω or less. The input capacitance of the clock pin on the ROM I'm going to use is 8 pF; if I put a 9.5 pF probe across that I'm more than doubling the capacitance. All of this changes the signal's behavior. Given my worries about ringing on the CCLK line of the Spartan 6 FPGA, and the two clock lines on the reconstructed 4004, I want to be able to probe such signals with the minimum effect possible.

For years I've wondered if I should invest in a 1 GHz probe for my 'scope. Earlier this week I decided to buy a used LeCroy AP020 FET-input active probe off eBay. It's old, but so is the 'scope it's going to connect to.

Let's compare the specs:

	PP007	AP020
Bandwidth	DC to 500 MHz	DC to 1000 MHz
Capacitance	9.5 pF	1.8 pF
Impedance	10 MΩ	1 MΩ
Attenuation	10:1	10:1
Max input	400 V	10 V dynamic ±20 V offset

Clearly I'm not going to be sticking this probe on a 120 VAC power line! But the response frequency is double and the capacitance is less than 20% of my passive probes. It also comes with adapters to allow connecting to a nearby ground point with a very short, springy wire. This reduces the inductance of the ground, which also helps reproduce signals accurately.

Unlike a passive probe, the AP020's input impedance is essentially flat across its rated frequency range. In an ideal world I should be able to see a signal containing not only the fundamental 100 MHz oscillator output, but also the 3rd, 5th, 7th, and 9th harmonics. Even losing the 9th the signal should look pretty square. Touching this probe to the same 100 MHz oscillator output I see a distinctly trapezoidal shape with flat tops and bottoms. I can't swear that this is the true shape of this signal but it meets the Epson specifications for this part.

Could I have done better? Sure. At some point this 'scope will break, and I'll probably replace it with something much more modern (and lighter, and quieter, and more capable, and...). But the 3.5 GHz active probe I used recently at work sells for $4,500. My AP020 cost me $200. It even came with calibration paperwork (from 1998, but hey, this is a hobby).