Protection for analog signal with <5pF

[luky315]

I have an analog signal (0..5V, 100MHz) with a long cable (unfortunately no coax, only a twisted pair) and a harsh environment. I don't need / want to pass any standards, but nevertheless I need "the best possible" protection (ESD, short to GND, VCC...) which has a neglible impact (5pF to GND seems to be ok) for it, the cost and space is not an issue. What would you use in this case? GDT + TBU?

[moffy]

I would probably start with TVS diodes as used for USB protection, a GDT is normally for higher voltage/energy strikes but they can be quite low capacitance. Are you expecting high energy strikes?

[luky315]

I have to expect misuse, maybe even a short to 230VAC

[Wolfgang]

Look at oscilloscope front end circuits and their protection.

[PGPG]

I have never considered protection against short to 230VAC so only some thoughts.
The first question is if the circuit have to still operate when short is disconnected or it have only protect itself against bigger damage.
I assume second.
I would start thinking from 5V SMB (600W) transil (or more powerful one). If it will be pre-polarized its capacitance is out of consideration. Then enough strong (I think 1A DC, 50A pulse) diodes. If remember well 1N4007 have about 40pF. Never searched for 1A low capacitance - don't know if you find something. But I remember seeing an Ethernet cable protection made of lot of 1N4007 diodes (don't know if it was 10M or 100M Ethernet).
I assume fast, low current fuses at the input. Don't know how fast are fast fuses and what are the lowest current available. May be smaller than 1A diodes (so less pF) can be used.
Important is what is the minimum distance from device where short can occur. It will determine the twisted pair resistance limiting the current pulse when short to 230VAC. May be a common mode choke can be used to have no influence at differential signal but have limiting pulse current effect when 230VAC short allowing smaller diodes to survive until fuse break.
High voltage diodes are not needed here. In switching circuits Schottky diodes are used. May be they have smaller capacitance...

[ftg]

Maybe smallest 230VAC rated smd fuses from Littelfuse on both cables and then low capacitance SMD ESD protectors to ground from those.
And possibly a GDT to clamp the 230VAC between the ESD protector vaporizing and the fuse acting.

[Andree Henkel]

take closer look into datasheet of receiving Amplifier first stage.
in case you find: absolute maximum rating: max input current - you have to use current limiting resistor additional to your protection diodes

this gives the required protection, so IC will work again once fault is gone

but it will kill your bandwidth - good luck 100MHz protected against 230V

take a deeper dive what kind of input protection the IC uses, basically there are three types:
*two back to back diodes between inputs, so differential input voltage is limited to one diode drop - as long the max input current is not exceeded
*two diodes to supply at each input -one to +, one to -, son input voltage is limited to supply + one diode drop, as long max input current is not exceeded
*no protection diodes - special amplifiers where extremely low input leakage current / capacity is necessary - like electrometer grade amplifiers - there you have no chance - only one: shorten input as long as fault is present

[Shay]

A decent value series protection resistor with diodes to clamp the voltage to VCC/VEE
the diodes should be lowest capacitance possible, the series resistor limits the current
However, you have an RC filter with the diode capacitance and the resistor, so take that into your design considerations. Maybe buffer it with a JFET if you need lower impedance.

[youngda9]

Fuse + TVS

[T3sl4co1l]

Impossible. At least for the effort worth posting on a forum -- perhaps a consultant could craft a solution with modest VSWR.

Closest, easiest solution is probably transformer coupling + fusing.  You'll need to encode the signal to remove DC of course, if it doesn't already.

Tim