forward leakage of a diode ...

[Njk]

Just nonsense for an engineer. But if there are a design rules in the organization, e.g. for every component, the electrical load shall not exceed 30% of its rating for each the parameter, and you have to get the design passed the formal review, that nonsense can create a problem

[arijav]

Hello, I am having a similar problem. In my case I have to perform leakage measurement tests of semiconductors up to 1000V down to 1nA. I have a setup with a logarithmic amplifier but I need to protect its input, which is at stable 0.5V. I have tried putting an InterFET SMPPAD1 Diode between the input and ground in forward direction. The issue is that it seems that at 0.5V I have a leakage of around 8nA (very different as the stated 1pA in reverse). I have tried as well to put 2 Diodes, one from the input to the guard, which has the same voltage as the input (+/-20mV) and the guard to ground, it improves but I still have 6nA leakage.

Is there any better suggestion to protect my input (while keeping the 0.5V constraint) and still being able to measure 1nA?

Thanks

[David Hess]

I have tried as well to put 2 Diodes, one from the input to the guard, which has the same voltage as the input (+/-20mV) and the guard to ground, it improves but I still have 6nA leakage.

If I understand you correctly, that is the circuit change I would have recommended.  If it did not improve the leakage significantly, then could there be another leakage path?

Could your InterFET SMPPAD1 diodes be fakes?

[Phil1977]

Hello, I am having a similar problem. In my case I have to perform leakage measurement tests of semiconductors up to 1000V down to 1nA. I have a setup with a logarithmic amplifier but I need to protect its input, which is at stable 0.5V. I have tried putting an InterFET SMPPAD1 Diode between the input and ground in forward direction. The issue is that it seems that at 0.5V I have a leakage of around 8nA (very different as the stated 1pA in reverse). I have tried as well to put 2 Diodes, one from the input to the guard, which has the same voltage as the input (+/-20mV) and the guard to ground, it improves but I still have 6nA leakage.

Is there any better suggestion to protect my input (while keeping the 0.5V constraint) and still being able to measure 1nA?

Thanks

If you don't need very fast transient response and can live with the input impedance, then I would propose such a setup:



The resistors in general prevent super high currents that magnetically couple to somewhere else. The gas discharge tube (GDT) does most of the work by dissipating the energy of ESD or high voltage arcing. The cap smooths the transients to a rise time that can be handled by the following diodes.

With this setup I could successfully protect FET-inputs of opamps from 8kV-HV-pulses that regularly arced over in a ionization sensor. Most of the time we just used a neon-indicator bulb as a GDT and it had no measurable leakage current. For the resistor between the GDT and the cap we needed to take a high-power high voltage type because usual small signal resistors also flash over at 3kV.

[arijav]

Thanks for the answers. The SMPPAD1 have been bought some days ago over Mouser, so they should be the real deal. I am wondering if the leakage is provoked by the rework soldering to fit the diodes, because I was specially disappointed with the results of the test with the 2 diodes. Since the guards have roughly the same voltage as the input I was expecting an If in the range of picoamps. I will try to rework another channel, this time with blackened white or blue LEDs with high Vf and check if the results improve. I am using as LogAmp the ADL5303 by the way.

Regarding the GDTs, it seems a nice solution but my boards need to fit 3 channels in one Eurocard and with them it will be tricky, as I have as well some overcurrent protection circuit which opens Reed-Relays (in the future maybe compact SSR) when the current approaches 10mA.

I am wondering if I just add a 80K resistor in series and check if the protection is sufficient without diodes. Any thoughts on this?

Any other black magic suggestion to solve this challenge is very welcomed!

[Phil1977]

A GDT for this purpose can be just a small neon indicator bulb like in a power switch, not something big as for lighting...

A TVS-diode may work too, at voltages <1V they also have very low leakage current (<1nA is typical)

[Kleinstein]

Some resistance that can withstand the high voltage is definitely needed. Bootstrapping the diodes, so that there is not much voltage at the low leakage diode is definitely a good idea. My choice would be something like BAV199 for the low leakage part and a 2nd pair for the bootstrapped voltage.
A dirty PCB can also show quite some leakage.

The ADL5303 itseld may have some bias current. It may not be the best choice for currents below 1 nA. For the stability the RC combination at the input is likely also important.

[voltsandjolts]

... I need to protect its input, which is at stable 0.5V.

The stable low input voltage is an advantage. Maybe something like this.

[arijav]

Great inputs from everybody. I will definitely have a look to the small GDTs and low leakage TVS

Some resistance that can withstand the high voltage is definitely needed. Bootstrapping the diodes, so that there is not much voltage at the low leakage diode is definitely a good idea. My choice would be something like BAV199 for the low leakage part and a 2nd pair for the bootstrapped voltage.
A dirty PCB can also show quite some leakage.

The ADL5303 itseld may have some bias current. It may not be the best choice for currents below 1 nA. For the stability the RC combination at the input is likely also important.

I will try other diodes like the BAV199 from the Input to the Guards and later to ground maybe TVS. Regarding the ADL5303, with the PCB as it is I have measured down to 700pA with a one teraohm resistor and a 700V HV source. The board itself seems to work quite good and the ADL should be able to read down to 100pA if well designed. The issue are the dirty modifications for the protection diodes. Nevertheless, would you suggest something different to the ADL5303? I am open to explore other designs. I need to be able to measure 1na till 10mA