Why do VNAs typically lack ESD protection?

[Torrentula]

At an internship I've been working with VNAs a bit and I've come to wonder why VNAs (and other RF test  gear) lack ESD protection diodes like you would find on a silicon chip for example?

Is it because diodes would add too much distortion even at low signal levels because they are non-linear devices?

Thanks!
Elia

[heavenfish]

The capacitance would lower down the bandwidth I guess.

[Marco]

They can make low enough capacity ESD diodes that you could put them into a 50 Ohm transmission line in theory, but at the extremely low voltages they'd have to clamp to protect the input transistors the capacitance is significantly nonlinear. Can't bootstrap at high frequencies either.

[tefe]

A parallel diode can do this, but it would distort S11 of input line and is hard to calibrate.
So I thik it's a compromise?

[Torrentula]

Thanks for your replies!

[David Hess]

Protection can be done at high frequencies without shunt diodes.  Instead use a diode bridge driven by balanced currents.  Then if the current through the 50 ohm transmission line exceeds the drive current to the bridge, the bridge disconnects the input from the output so protection is instantaneous.

This works great up to 1 GHz and beyond in time domain instruments like oscilloscopes but I assume still introduces too much distortion for frequency domain instruments because they are so much more sensitive.

For an example of this, check out the design of the Tektronix 7A29 1 GHz vertical amplifier.

[Howardlong]

High speed sampling oscilloscopes frequently lack any protection, with warnings everywhere in the documentation together with appropriate panel markings as a result.

I guess any attempt to clamp the front ends is going to introduce some unwanted capacitance, degrading rise time and bandwidth.

[David Hess]

I do not know of any sampling oscilloscopes which include protection but very few are as slow as 1 GHz and even the 1 GHz ones that I know of do not include any special protection either.

There are other ways however to build a fault protected sampling head; start by using an impedance other than 50 ohms.  The 1 Mohm 350 MHz Tektronix S-5 is almost at tough as an oscilloscope input.  The 100 kohm 1 GHz Tektronix S-3 is not quite that tough but is still much better than an unprotected 50 ohm sampling head.

Active probes suffer from the same problem; the highest performance ones have no input protection because it would compromise performance.

[Marco]

Protection can be done at high frequencies without shunt diodes.  Instead use a diode bridge driven by balanced currents.  Then if the current through the 50 ohm transmission line exceeds the drive current to the bridge, the bridge disconnects the input from the output so protection is instantaneous.

This works great up to 1 GHz and beyond in time domain instruments like oscilloscopes but I assume still introduces too much distortion for frequency domain instruments because they are so much more sensitive.

Yeah, instead of the non-linear capacitance you now suffer the non-linear dynamic resistance.

[David Hess]

Protection can be done at high frequencies without shunt diodes.  Instead use a diode bridge driven by balanced currents.  Then if the current through the 50 ohm transmission line exceeds the drive current to the bridge, the bridge disconnects the input from the output so protection is instantaneous.

This works great up to 1 GHz and beyond in time domain instruments like oscilloscopes but I assume still introduces too much distortion for frequency domain instruments because they are so much more sensitive.

Yeah, instead of the non-linear capacitance you now suffer the non-linear dynamic resistance.

Since the output follows the input and the current driving the diodes is constant, the current through the diodes is constant within the limitations of the impedance of the current source and sink.  Whatever distortion it produces is not visible in a time domain instrument like an oscilloscope but I am sure a spectrum analyser would see something.

They make diode shunts for protecting RF input stages from overload but they are intended for low signal levels.  Avago has a bunch of application notes on the subject.  Commonly a PIN diode is used anti-parallel to a schottky diode; the schottky shifts the DC level on the input coupling capacitor and once the PIN diode turns on, it shorts out the RF.  Oscilloscope inputs usually use shunt diodes which operate as PIN diodes because they are slow but with a large DC bias.

One reason they might not bother protecting the input is simply because an overload would destroy and short out the protection diodes anyway in which case the instrument becomes useless.  The constant current diode bridge at least would tolerate a higher level of overload.

[Marco]

Since the output follows the input and the current driving the diodes is constant, the current through the diodes is constant within the limitations of the impedance of the current source and sink.

The input however is not a current source, it's a 50 Ohm impedance voltage source. Lets say Rs of a diode is around 5 Ohm, that would also be the approximate dynamic resistance it would be biased at ... but you can't bias it too far into it's linear region without losing the ability to limit voltages too. So the entire bridge would then also have around 5 Ohm equivalent resistance and you'd put a 45 Ohm impedance buffer amplifier on the other side.

AFAICS the changes in dynamic resistance of the diodes in the bridge will not cancel out exactly as the current through the legs changes (it goes up both when input goes positive and negative) so the equivalent resistance of the bridge changes, so you get non-linear distortion.

[dacman]

I know a laboratory that uses their VNA on an ESD mat.  (They pointed it out.)

[G0HZU]

FWIW some VNAs do use overload/ESD protection diodes. I've seen these fitted to HP VNAs that can operate up to 3GHz.

[Howardlong]

I know a laboratory that uses their VNA on an ESD mat.  (They pointed it out.)

How much of this is to mitigate real risk, and how much to instill a culture of care is up for debate.

At the end of the day when you're testing filters, amplifiers, or other muti port devices, almost certainly the coaxial outer connection will maintain a common ground so the ESD mat offers limited benefit.

However what I am an advocate of is the psychology aspect: if you engender a culture of care where people think twice before each step rather than go gung ho into everything, that can have significant benefits, particularly in a lab where equipment is shared, and absolutely in a student lab, where stuff inevitably happens.

In later years I've been chastised in some clean rooms, for bringing in paperwork (yes, the paper fibres are a no no in some clean rooms) or a laptop with a fan in it for example. You just don't thnk about that stuff until it's drawn to your attention.