ESD protection is insufficient

I was looking at the analog part circuit and found this vital problem.
The component (GDT_SH90)you are using for ESD protection is way less sufficient.
the analog input of the following circuit is very sensitive to ESD, a spark gap cannot provide enough protection.
because the trigger voltage is around 100V, but in realistic, the overshoot can be several hundreds of volts.
And due to the property of the gas, there is a time lag between the primary discharge and the secondary discharge, it could be tens of nano seconds, with that long time, the following circuit could be already dead.
Highly recommend that you can change the spark gap to a TVS diode. I'm an ESD expert. I can give you suggestions, many components from ST and Nexperia are very good choices.

ESD is the most dangerous transient event that damage the scope, and it is very easy to happen to every  one and every day. Even in a humidity environment, there still could be several hundreds of volts on your body.

Pengyu

Hi Pengyu and thank you for your comments, but there is no problem with the ESD protection in the analog inputs.

In the analog front end we have implemented two levels of ESD protection. The first level is the gas discharge tube (GDT) which absorbs the main part of the ESD energy and lowers the peak voltage of the overvoltage transient. We have used the GDT component because it provides a very low input capacitance and more importantly, this capacitance is non-variable. On the other hand, the input capacitance of the TVS diodes change with amplitude and frequency and this is not desired when measuring high speed signals. Any capacitance non-linearity at the input will cause a non-linear frequency response. So, using a TVS diode directly at the input is not an option.
The second stage of ESD protection is located at input amplifier. Here we have used two diodes (BAV199) to protect the input amplifier from the ESD energy that is left after the GDT component. Diode capacitance non-linarity is not a problem here, because this capacitance is relatively small and is isolated from the input with the input RC divider.

Hi Dejan,

I understand your point, but I have to say both stages cannot provide enough protection.

It is pretty easy to find a low cap TVS from ST, Nexperia, Infineon,etc. The bandwidth of this design is 100MHz, so any diode with capacitance less than 100pF should work(100ohm+100pF->BW=100MHz).

You didn't see any ESD failure probably because of the series resistors, they can limit the ESD current effectively. But the resistor could be degraded by ESD as well, and it cannot recover, you gonna see the 100ohm becoming first to several tens of ohms, then after more strikes it becomes several thousand of ohms.

More of that, the position of BAV199 cannot help the ESD, that is because the series resistors will have the most voltage drop and residual voltage cannot trigger this diode, and the diode inside the opamp has a much lower trigger voltage, so BAV199 is doing nothing there.

I don't know the ESD performance of AD4817, but usually the opamps are sensitive to ESD events.

I highly recommend that you would change the component, if you plan to have 50ohm input impedance or higher bandwidth, in that case, you can't have so much series resistance to limit the ESD current.

Hi

This is an interesting topic and I appreciate your concerns, but I have to disagree with some of your statements.

If you can find a suitable TVS diode, then please suggest one... The TVS diode breakdown voltage must be above 25 V and very important factor is also diode capacitance as a function of frequency and voltage.

I have to argue that BAV199 is doing nothing. This diode it is clamping the input voltage to the ADA4817 below the maximum allowable limit (note the absolute maximum ratings in the ADA4817 datasheet). The input series resistors are relatively low value compared to the input impedance of ADA4817 (500G Ohm), so they can not provide enough protection without the BAV199.