Long, floppy leads on high speed active probes?

[jjoonathan]

My understanding was that the principle of high speed active probes was basically "get the amplifier as close to the circuit as possible to minimize capacitive / inductive / transmission line / stub loading, resonance, etc." In the low GHz range active probes follow exactly this pattern, with the amplifier right in the probe tip, as close to the circuit as possible. However, after 5GHz or so, the designs seem to start bringing the amplifier further and further from the tip, in some cases so far that the distance from tip to amplifier is roughly as far as the distance from amplifier to scope (see the keysight link below)! How does this work?

My first thought was that the long, floppy leads in front of the amplifiers are not as simple as they seem, that they're actually clever high-characteristic-impedance coaxial transmission lines which can simultaneously present high impedance to DUT, carry the original signal voltage, and shield from the surroundings, but if that's the case why bother with an in-line amplifier at all? Why not bring the high-impedance coax all the way back to the scope's front end amplifier? Some of the "tips" are already long enough. Ditto if the trick is "high-resistance tip in series with transmission line to act as attenuator plus amplifier to bring the signal level back up." What gives?

http://www.keysight.com/main/editorial.jspx?cc=US&lc=eng&ckey=1615761&id=1615761&cmpid=zzfindextreme

http://www.tek.com/datasheet/z-active%C2%AE-p7300-differential-probe-family-datasheet

[alm]

The Keysight N5450A InfiniiMax Extreme Temperature Extension Cable is specified for temperatures up to 150°C. I imagine it would be hard to design an amplifier that will work up to 150°C. The long cables are necessary to reach inside the environmental chamber. The long cables do degrade the bandwidth (extension cable with N5441A 12 GHz, as opposed to 20 Ghz for the N5441A without the extension cable), possibly due to skin effect loss in the transmission lines.

The other probes do have leads that are much shorter than the cable from amplifier to scope (~10 cm), so putting a Z₀ cable between amplifier and scope still saves a bunch of capacitance. Most of lower bandwidth probes, like this old 1.5 Ghz Tek probe, or this 750 Mz Agilent probe, came with simple copper wires that connected to the probe for connection to SMD parts by either soldering down or clips. You can't very well clip a probe directly around a TSSOP lead or BGA ball and hang the whole mass of it, like you could with old through-hole parts and probes with retractable hooks. Especially if it has to fit in a narrow space. Not much choice of connecting a probe to a tiny lead or test pad apart from soldering down or handholding. Making the leads thinner and 'floppier' helps reduce mechanical stress on any test points.

The leads shipped with the high-bandwidth probes are generally more advanced with low-value isolation resistors (~50 Ohm) as close as possible to the DUT and sometimes some other passives, and likely a lossy (resistive) wire or terminated transmission line (as is used by the InfiniiMax stuff) to dampen any ringing. Pretty sure at least Keysight/Agilent and probably also Tektronix have published information about this. This Agilent appnote, for example.

Apart from the environmental probes, I don't really see the leads from DUT to amplifier getting any longer. Just more advanced and more aimed at soldering down as opposed to various types of clips.