Need a transistor which as high off-state resistance

[hkBattousai]

There is a certain node in my circuit at which I want to limit level of voltage. The voltage at the node is pulled up by a resistor with an amount of about 1MΩ. I want to use a transistor to ground this node if excess voltage is delivered.

But I don't want to pull-down the voltage at normal operation (i.e.; the node voltage is below upper limit).
For this purpose I need a transistor which has an off-state resistan much higher than 1MΩ.

Which transistor type has the highest possible off-state resistance? BJT, FET or MOSFET?
Can you suggest me a part number please, if you can?

 

[Grossel]

Why doesn't just use a zener?

 

[hkBattousai]

Why doesn't just use a zener?

Zener has too much leakage current for my design.

 

[audioguru]

Ordinary transistors have a low leakage current if the pcb is clean and if the ambient is not too hot.

 

[hkBattousai]

I checked for some MOSFET datasheets, unfortunately they don't mension about cut-off drain-source resistance.
Is there any way to calculate it from other specification data?

 

[smanches]

Try Drain-Source leakage current.

EDIT: I see it as "Zero gate voltage drain current" on one of the datasheets I have. This one is 1uA at the max drain voltage.

 

[crutschow]

Either a FET drain or a bipolar transistor collector look like a reverse biased diode when off. They typically have off-leakage well under a µA at room temperature and at voltages some below their maximum rating.

 

[kchriste]

My question is how you are going to sense that the voltage at the node has risen above spec without loading it down. You could use a pre-biased zener and diode to accomplish this. The zener voltage would be apx 0.7 lower than the desired clamping voltage:

 

[hkBattousai]

My question is how you are going to sense that the voltage at the node has risen above spec without loading it down. You could use a pre-biased zener and diode to accomplish this. The zener voltage would be apx 0.7 lower than the desired clamping voltage:
View attachment 38735

There is a voltage divider in my design. A 1MΩ (at high side) and a 10kΩ (at low side) are connected in series. 10K is connected to ground and 1M is connected to 500V. So the voltage on 10K resistor is only 5V. If it goes above 5V, I will use a comparator to swith on the MOSFET to ground the voltage on the node where 10K and 1M resistors are connected.

In my first post I said that the value of the low side resistor was 1MΩ, but I changed my mind and decided to use a smaller resistor.

I checked some MOSFET datasheets, they show a "drain-source voltage vs drain current (VDS vs ID)" graph but it passes from origin. It doesn't give any information about the values about origin.

 

[kchriste]

If it goes above 5V, I will use a comparator to switch on the MOSFET to ground the voltage on the node where 10K and 1M resistors are connected.

The problem with that is once the MOSFET turns on, the voltage drops below 5V which turns off the comparator. This causes the voltage to rise again until the comparator trips again and the cycle repeats. In essence you'll have an oscillator with a waveform that will spike above 5V for a duration depending on the speed of the comparator, FET, gate/node capacitance, etc.

I checked some MOSFET datasheets, they show a "drain-source voltage vs drain current (VDS vs ID)" graph but it passes from origin. It doesn't give any information about the values about origin.

For a FET look for the Idss specification.

Since the current will be low, you could simply rely on the input clamping diodes of the ADC or whatever is connected to the node. Or you could get fancy and use an OpAmp as a voltage clamp. The OpAmp has the same speed issue due to it's slew rate so some overshoot will occur: