Once in a while customers ask me about what probes do they need to go with their high bandwidth oscilloscopes.

Agilent already has application notes about how to probe properly at high frequencies to ensure what you see on the scope represents the reality faithfully, but they are a little dry. Bob Pease Show at National Semiconductor (now acquired by Texas Instruments) talked about it and it’s great infotainment due to Bob Pease’s character:

This show has significant product placement by Tektronix, but the information there applies equally (and fungibly) to all major name brands such as Agilent/HP/Keysight and Lecroy. They all live up to the specs advertised.

What I’ve learned from the video

• No-name brand probes might not live up to the claimed specs. I wouldn’t trust a Chinese probe beyond 100Mhz (or even 60Mhz).
• Shorten the ground leads as much as possible, especially high frequencies. Wires are inductors/antennas.
• Do not use the poor-man’s differential probes (aka subtracting the channels on the scopes): the channels aren’t matched perfectly, the probes aren’t matched perfectly either.
• Design for testing: plan your PCB so you can probe easily.
• For digital designs, high bandwidth scope users care more about (time-domain) step response: rise-time, ringing, settling, than it’s frequency domain (I don’t have a fast pulse generator, this is why I test it with a RF generator to check the specs)
• Active probes have less loading and attenuation. You can use passive probes if you have a large enough signal to burn.
• Probe capacitance (loading) kills a fast circuit (by damping it down)
• Don’t be happy because you see nice waveforms and nothing bad happens with a low bandwidth scope+probe: you are just failing the capture transients.