[Radio Wrangler]

The achilles' heel of the scope probe is the need to make a cable as low capacitance and as high impedance as possible. it is then used in highly mismatched and uncertain circumstances. We then want it to give a flat response and clean, fast risetime.

Aye, right! as they say north of the border.

Once I get above a few tens of MHz, I prefer to use probes with a real, broadband, well screened 50 Ohm coax, a scope with a switched 50Ohm termination, and a simple series resistor at the pointy end. These do not give megohm impedances at DC, but they can be a lot better than what the routine divide-by-ten 10Megohm probe degrades to at those frequencies. I have some Sealectro probe tip ends with SMC connector back ends which go on a length of thin 50 Ohm coax. Tektronix had a family of complete probes like this.

So, once you start climbing up the MHz, you leave attempts at high-Z behind. Often, they were futile anyway.

Once I go higher still in frequency I leave scopes behind and switch to the frequency domain and use the resistor-on-coax probe on a spectrum analyser. I'm probably thinking in frequency domain terms anyway at this frequency.

I've spent a lot of time in the last several years probing GHz transmitter circuitry and the resistive probe on a 50Ohm spectrum analyser was the only tool left in the box. Pulsed transmitters reduced the net long-term power dissipation, but the instantaneous peak voltages would have instantly destroyed high-Z active probes.

The one thing out of this is learning how critically the way you ground the outer of your coax at the pointy end to the circuit under test affects results. Very low inductance is needed.

Plan-B is to put resistor networks on the board being developed to act as interfaces with a 50 Ohm connector. No probe as such, just a wandering 50 Ohm cable is needed.

With the resistive probe on 50 Ohm cable, slipping a few ferrite beads on the business end of the cable can help control/avoid common mode effects.

With circuitry in llittle screening boxes accessed through removed lids, you can rest the grounded shank of the resistive probe against the box wall very close to where the pointy bit is probing.

Once you switch from a scope to a spectrum analyser you suddenly get an awful lot more sensitivity available, and you can exploit this to use more lossy probing/sampling arrangements which are the price of getting less loading on the circuit under test. This sensitivity also leaves far behind what diode probes could do.

If handling moderate RF powers in the milliwatt and upwards region, thermal power meters enter the scene. These can be very flat, very accurate. They tell you the power level present. They don't do pulses. The tell you not one jot about waveforms and spectra, just how hot the signal makes a 50 Ohm resistor. Frequency wise, this technique works to and beyond optical frequencies, if you can make a terminating absorber at your frequency, you're in!

So scopes were common to 100-200MHz with a few at 400MHz, and one at 1GHz. Apart from sampling scopes they were never pushed higher, because people switched to spectrum analysers. Folk working on ultra-fast logic wanted time domain, so multi-GHz sampling scopes were developed. At first these were subsamplers, needing repetitive signals, but nowadays there are digital samplers running at several GS/s that can give you useful one-shot bandwidth and sometimes a few samples up risetimes..

with the resistive probe, I have a useful bit of kit, the probe into transmission line adaptors made for the HP vector voltmeter. A stainless N-type length of line with a side port to allow a probe to hard ground to the outer metal and to just touch the inner line conductor. True 50 Ohm Zo throughout apart from the side port. Stick it on a sig gen, with a good 50 Ohm termination after it, and you now have a fairly accurate voltage on the line to measure your resistive probe calibration. You can look at flatness this way, you can also use an attenuator and analyser after this probing adaptor as the 50 Ohm load, and you can see the effecto of applying and removing your probe as a sanity check.

The vector voltmeter, now I've mentioned it, is a handy bit of kit that will act as an RF voltmeter down below mV and up to 2GHz. 1GHz for the classic one with built in probes on cables and moving coil meters. Two probes, and you can get the relative phase measured too.. The probes contain diode sampling bridges driven by an LO in the main instrument. The LO is phase locked to the incoming signal.

It has its limitations, but the limitations are different to other instruments, so they let you see things you wouldn't see elsewhere.

No one size fits all. Diode probes are cheap to make. Other stuff can be very expensive unless you hit a lucky find.

So there you have it. Many methods, plenty of overlap, but each with strengths. The comment on probe ground referencing, and the ferrite beads apply to all.

I'll have one of each, please.

David