原创 Guide to probing (Part 1)

Back when the Earth was beginning to cool and the first creatures were crawling from the slime, we were building a very fast system with plenty of discrete logic because the processors of those Paleozoic years couldn't keep up with our data rates. I was probing a pair of signals on a fabulously-expensive oscilloscope, but their relationship made no sense. Eventually I realized that one of the scope probe leads was a full meter longer than the other. This was the first, but certainly not the last, time I cursed light's snail-like pace. And it was the first time I was forced to think about scope probes, other than the routine of compensating them.

Eons later I still see engineers rooting around in a bin and untangling a random, often abused, probe, carelessly connecting it to the scope that they spent a week evaluating and comparing to a host of other products. But that $10k or $20k instrument (or more; I read a press release for a nifty $400k model this week) can be completely hobbled by a lousy or poorly-selected probe.

A scope probe is not a wire. Sure, it's an electrical connection between a node on a board and the oscilloscope. But its resistance, capacitance, and inductance have serious consequences as speeds increase.

Plenty of white papers talk about proper probe selection, but to my knowledge none show how the wrong probe can either cause your circuit to fail or even physically destroy components. Let's look at some of the issues.

For readers without an EE, here's a bit of background. There will be a little math, but fear not! Mostly this is as simple as Ohm's Law, which is:

E is voltage, I current, and R resistance. But there's really no such thing as a perfect resistor. Every component, be it a wire, resistor, or anything else has some amount of capacitance and inductance associated with it. So, while Ohm's Law holds true, as frequencies increase we need to include these effects in our thinking about resistance.

A capacitor passes AC but not DC, and the higher the frequency, or the larger the capacitance, the more easily it passes current. This is known as capacitive reactance, which is a form of resistance to alternating current. It's expressed as:

C is the capacitance in farads and f is the frequency in Hz. If f is zero—DC—the reactance is infinity and a perfect capacitor completely blocks current flow. As the signal gets faster, its reactance drops.

Inductors are similar:

In this case L is the inductance. So an inductor, too, has a resistive-like effect, although it is inverse of that of the capacitor.

If a circuit has inductance and capacitance, the total reactance is, well, it's thankfully not important to this discussion, but suffice it to say that one can compute a net reactance. Toss in pure resistance and the total impedance—which is merely resistance to waveforms—in a parallel circuit is:

Ohm's Law still holds, except we now replace R with Z. So it's possible to compute current flow in a complicated circuit at any given frequency if one knows the voltage and impedance.

I said there's no such thing as a perfect resistor. A typical ¼ watt resistor has 0.5 pF of shunt capacitance. A wire-wound resistor is a coil and has inductance (although some have reverse windings to limit this). A normal film or carbon resistor has very little inductance.

A scope probe has capacitance, resistance and some inductance (mostly in that black ground wire), and looks something like Figure 1.

A scope probe, because of its impedance, is rather like a resistor that gets placed in parallel with the node you're testing. The net impedance of two in parallel is:

Thus, placing a probe on a node reduces the impedance of that point. Whatever drives the node must, in effect, push harder or the signal will degrade.

But is this really a big deal? Scope probes are typically specified at 10 million ohms, which is an enormous value. Put that in parallel with a digital device, whose resistance is orders of magnitudes lower, and there will be no measurable effect.

But what's important is impedance, not resistance, and as noted earlier, impedance depends on frequency. If you're probing a digital circuit odds are that things are switching pretty quickly. The frequency may be quite high.

That's where the second part of a probe's spec, the tip capacitance, comes in. This ranges from under a pF to 100 pF or more. And that has a huge effect on the probe's impedance. Tektronix is one of the few vendors that gives graphs of probe impedance. Their TPP1000, a 1-GHz passive probe, has the impedance vs. frequency curve in Figure 2.

At 100MHz the impedance appears to be around 500 ohms. The probe's capacitance is 4 pF; running the reactance numbers we get 400 ohms at that frequency, pretty darn close. At low frequencies the reactance gets very large so the 10-MΩ resistor dominates.

This is a $900 probe. Cheap probes may be considerably worse. The capacitance of Pomona's $117 5812A, rated for 300MHz, is listed as "<17 pF." Let's assume 17. That's 93 ohms at 100MHz. In a 5-volt circuit, the probe will add a 54-ma load.

Surfing the net one finds lots of cheapies at 100 pF or worse, which nets a terrifying 20 ohms at 100MHz. To put that into perspective, in a 2-volt circuit it would take 100 ma to drive the probe alone, and few gates can do that. Even at 10MHz we're talking 10 ma. 74HCT gates, for instance, are generally rated at 4 ma. Other parts, notably processors, may be capable of driving less current. TI's OMAP35xx parts are spec'd at 2 ma on most pins.

In other words, connect one of these puppies to your board and the circuit may very well stop working. I think some of those homeless people I see in Baltimore are engineers who cracked when they found their troubleshooting tools made troubleshooting impossible.

The important take-away is this: when buying a probe don't be seduced by the 10-MΩ rating. Think impedance, which is related to tip capacitance.

(A side note for those working with analog. Generally analog signals aren't as fast as digital so the probe-loading issues may be less severe. On the other hand, many analog nodes have a much higher impedance than digital. Even at DC putting a 10-MΩ probe across a 10-MΩ node will zap half the signal).

Better probes exist. Tek's 4 GHz P7240 has a tip capacitance of only 0.85ÿpF. Running the numbers we get 2ÿKΩ at 100MHz, which exactly matches the chart in the device's manual. But this is an active probe, one loaded with electronics, and it'll set you back $5,000. Which explains the sign held aloft by one gaunt street person last week: "Please help. Will work for a pair of P7240s."

Logic analyzer probes also exhibit capacitive characteristics. Agilent, for instance, sells sets that range from 0.7ÿpF to 3 pF per tip. The probe sets for their MSOs run 8 to 12 pF. Cheapies advertised on the Internet have higher capacitances, and an astonishing number don't have a rating at all. Never connect a logic analyzer to a circuit unless you've thought through the probe impedance issues.