ultra low current devices

What sort of devices are used in applications such as creating a 1nA current source?

Most transistors (both bipolar and JFET) that I find enter cutoff at this level. collector cutoff current of the analog devices MAT02 matched NPN BJT pair is 3nA (these things cost 20 bucks a pop)
IDoff of the J113 JFET is 1nA

a circuit with extremely high gain in a feedback configuration might be able to hold these devices on the edge of cutoff to make them source such low currents, but are there [discretely packaged] devices available for extremely low current applications?..I'm having a bear of a time finding any


EDIT: its also rare that datasheets spec devices down to such low levels, or even provide any information at such low levels. I'm not sure if I fully trust the SPICE2 models for these devices either, at low levels.

opamps and FETs...

the cutoff spec is just how they measure it... what you are missing is that you are trying to generate is in the PCB leakage range... can you do better, of course, but you need to keep it clean - no rosin traces and perhaps ground isolation tracks around the critical opamp inputs

thanks for the reply

PCB design will incorporate driven guard traces to minimize leakage, among other low leakage techniques (matching metallic junctions to minimize seebeck effect, teflon standoffs for sensitive devices, etc).

what exactly do you mean by "the cutoff spec is just how they measure it" ... i assume this is when they set the Vbe to 0V or lower, and measure how much current flows in the collector - this would be the absolute smallest amount of current that the device can pass. i.e., the OFF current.

my current solution is using high performance BJT pairs and a really nice opamp (LT1885) since it has over 1 million open loop gain

solis365 said:

thanks for the reply

PCB design will incorporate driven guard traces to minimize leakage, among other low leakage techniques (matching metallic junctions to minimize seebeck effect, teflon standoffs for sensitive devices, etc).

what exactly do you mean by "the cutoff spec is just how they measure it" ... i assume this is when they set the Vbe to 0V or lower, and measure how much current flows in the collector - this would be the absolute smallest amount of current that the device can pass. i.e., the OFF current.

my current solution is using high performance BJT pairs and a really nice opamp (LT1885) since it has over 1 million open loop gain

Click to expand...

wow .. you are well beyond most of the people that frequent these boards. matched pairs should not be needed since you are looking at a good opamp though BJT singles will not work since you are looking at low currents and will always be fighting the beta tempco.

in the case of BJTs you have the base collector leakage multiplied by the very temperature sensitive beta. there are circuits that can compensate for that and you are obviously using one. I am assuming you are using a sense resistor, though you do not mention how much sense voltage you are using.

how they measure it ....

the https://www.electro-tech-online.com/custompdfs/2010/07/BC846ALT1-DPDF.pdf specs it at 15nA@30V Vbc but expands it to 5uA@150C ... before you take that at face value you need to look at the actual symbol: Icbo

that means the base is open and the leakage spec is assuming no base current or in other words the collector leakage current is being gained up by that very temperature sensitive transistor beta. Very few if any spec sheets give the spec you are actually interested in, which is when DRIVEN off what does it still conduct.

Dan

interesting... it appears you are nearby as well. I too am in CT

Ubergeek63 said:

wow .. you are well beyond most of the people that frequent these boards. matched pairs should not be needed since you are looking at a good opamp though BJT singles will not work since you are looking at low currents and will always be fighting the beta tempco.

in the case of BJTs you have the base collector leakage multiplied by the very temperature sensitive beta. there are circuits that can compensate for that and you are obviously using one. I am assuming you are using a sense resistor, though you do not mention how much sense voltage you are using.

how they measure it ....

the https://www.electro-tech-online.com/custompdfs/2010/07/BC846ALT1-DPDF-1.pdf specs it at 15nA@30V Vbc but expands it to 5uA@150C ... before you take that at face value you need to look at the actual symbol: Icbo

that means the base is open and the leakage spec is assuming no base current or in other words the collector leakage current is being gained up by that very temperature sensitive transistor beta. Very few if any spec sheets give the spec you are actually interested in, which is when DRIVEN off what does it still conduct.

Dan

Click to expand...

Thanks, that helped quite a bit. I guess I will just order some devices and see how much current I can pass with a carefully driven base.

I am a grad student in analog EE, and I am working on some instrumentation circuitry. I may be "advanced" but there is still a lot I don't have much experience with. It's hard to get this information without asking a few people and hopefully having one prod you in the right direction.

The idea is to maintain a precise, low noise 1 nA bias at the output of the circuit, and to measure and maintain it without injecting significant noise. I have not worked much with bipolar devices so this is new territory for me, especially over temperature. I am not expecting these devices to be hot (only passing 1nA), nor will they be near any other hot components. Since the system only has to work in a room with temperature and humidity controlled 24/7, might it be safe to ignore the beta tempco? probably not...

I am worried because I don't think this C-B leakage current can be "corrected" by even a high gain feedback controller... the current is generated internal to the device

My home is CT but I am currently at school in Rochester