Using NPN or PNP transistors at micro amp bias and load levels.

[onewilly]

I am trying to build a water detection circuit that will operate for over one year using a 9 volt alkaline battery and therefore need to operate the detection cuircuit at the lowest possible current level. I am using two stainless steal electrodes, one at ground potential and the other electrode biasing a PNP transistor via a 4.7 mΩ resistor with a 10 mΩ hold off resistor from the base to Vcc. Since there is a timing circuit involved else where in the design, the operating voltage is regulated down to 5 VDC for stability. The nominal voltage across the electrodes varies from 0.5 to about 1.5 VDC. I assume that the voltage fluctuation is a result of electrolysis where hydrogen and oxygen tend to build up on the electrodes. A brief jolt of reverse polarity to the electrodes will drop the voltage across the electrodes back to about 0.5 VDC. Operating at µAmp bias levels, the circuit is sensitive to any form of leakage or contamination which is why I have included the 10 mµ hold off resistor. The collector output drives a 10 meg resistor. A CMOS invertor gate input is connected to the collector. The electrodes are nornally submerged in tap water where the PNP bias is at about .63 uAmps. When the water level drops below one of the two electrodes, the base is open (except for the 10 meg hold of resistor). The voltage fluctuation at the collector goes from esentiall Vcc to ground changing the inverter output state. After this wordy description here is my probelm. Data sheets on transistor don't specify gain at µAmp bias levels. What is the best choice for the PNP transistor and are the bias levels reasonable? Will the current levels be below the device cut off currents? I need to assure that going from submereged to out of the water conditions causes the collector voltage to exceed the inverter low and hi switching voltages.
THANKS!!!

 

[crutschow]

Modern transistors typically have low gain at very low currents. If it is not specified in the data sheets then you may need to measure it. Generally the smaller the transistor (the lower its maximum current rating) the higher gain it will have at low currents.

Can you use a small MOSFET transistor instead?

 

[alec_t]

Or can you do away with the NPN/PNP transistor and simply connect the probe, plus any necessary bias resistors, to a CMOS gate input?

 

[onewilly]

Using NPN or PNP transistors at micro amp bias and load levels. REPLY

Thanks for your response!
Measuring the gain at the µA bias level is a good suggestion since it is not published on the data sheet. Using a PNP 2N3907, with my Fluke it measured about 315. At about 0.60µA it had dropped to about 200, well in excess of what is needed to fully load the 10 mΩ collector resistor.

I am not that familiar with the use of MOSFET transistors. However, as I understand it, they are voltage sensitive as opposed to a transistor that is more driven by current. Originally I had the resistor in series with one of the probes feeding the gate of my inverter through a 10 mΩ resistor. The high Z of the inverter in conjunction with a slightly higher humidity level in the water vessel allowed enough leakage to driver the inverter gate into a false alarm condition. In addtion the voltage across the two stainless steel probes could drift high enough that if the water level dropped below the probes causing an alarm condition, when the water level rose and covered the probes, the voltage would not drop low enough to change the inverter state. So, I felt that I had to utilize an interface between the logic gate and the probes that would be less sensitive to voltage fluctuations. I hope that I have achieved that using transistors? After reading the responses I fell that using transistors and current biasing may be the best way to proceed? Or, is there a way to use a MOSFET that will be less sensitive yet more resistant to humidity and or contamination?

Thanks again! This is the first time I have used a Forum, what a great tool for learning!!

 

[alec_t]

Or, is there a way to use a MOSFET that will be less sensitive

The sensitivity/gain can be set by use of suitable resistors in the drain and source current paths. You can make a CMOS gate input less sensitive by using bias resistors (e.g. ~100k) connected between the gate input and V+, V-.

Edit: BTW, immersed probes can become plated and present a changing resistance over time if fed with DC. It is common to feed them AC instead, to overcome this. If you search this forum you will find another thread dealing with AC water level sensing/control.

 

[onewilly]

Sounds like an alternative that I should experiment with. It sounds like I should be able to set the threshold voltage at say 2.5 vdc with a 5 vdc supply. The highest I have seen the electrode voltage climb to is about 1.9 volts above ground. It seems that the physics of the electrolysis process allow very little control over the voltage between the electrodes. With the electrodes out of the water, the voltage should approach the supply voltage. I assume that the MOSFET output should swing pretty much from rail to rail? That may solve the probelm I had with the CMOS Inverter turn off point (worst case tolerance) below the 1.9 volt limit and keep my current consumption as low as possible.

 

[alec_t]

I assume that the MOSFET output should swing pretty much from rail to rail?

That will depend on how you bias the gate, the value of the turn-on threshold of the MOSFET, and the values of any source/drain path resistors.