small current limiting

[gus789]

Wondering if anyone can offer some advice on a current limiting circuit for small currents (1uA-100uA) with less than 1V supply.

I currently have a 10k-100k resistor in series with a variable load resistance. The current in sinked into the virtual ground of a photodiode (I to V) amplifier.  The quantity of interest is the load conductance, which we are trying to measure in real time by showing the output of the I to V converter on the scope. The problem is that the series resistor makes the output current no longer proportional to the load conductance, and requires additional computer processing of the output signal to extract the desired load conductance. I'd like to possibly substitute the series resistor with a current limiting circuit, simply to avoid damage of the load when its conductance becomes too high, while allowing the conductance value to be read directly from the scope.

I looked around for current limiting circuits and found ones similar to what is described  here http://www.zen22142.zen.co.uk/Circuits/Power/add-on.html.
Would this work down to <100uA level? Are there any better/easier ways to accomplish this?

Thank you!

[moffy]

Yes a single BJT transistor should work down to the 1ua level. Use a small signal device e.g BC547C etc or PNP equivalent BC557. The lower the current the higher the output impedance becomes. Keep them cool (shouldn't be a problem at those currents) otherwise leakage currents can be significant. Use a single LED as the bias diode(in place of the two forward biased diodes in the base) and you get less drift. An infra red LED should give around 1 to 1.5v bias at 1ma.

[T3sl4co1l]

Do you want "constant" as in "independent of output voltage", or "constant" in absolute terms?

The first is easy, a current mirror will do the job.  The output current will be fairly independent of output voltage (give or take Early effect, which still won't be too bad at that voltage, and as long as it stays out of saturation, say Vce > 0.2V), but depends directly on the input current.  If you use a resistor to supply said current, then it's linearly dependent on supply voltage.

Tim

[gus789]

Thanks for the replies! If I understand correctly, the current mirror is a constant current source  - i.e. it adjusts the voltage (up to a limit)  so as to force a set current through the load. I don't think that is what I want. What I'm after is a current compliance circuit that prevents excessive current from damaging the load, but otherwise just behaves like a constant voltage source.     

[dannyf]

What people mean by "current limiting" differs from person to person. Without clearly articulating it, no one can help you.

What I'm after is a current compliance circuit that prevents excessive current from damaging the load, but otherwise just behaves like a constant voltage source.   

Use a comparator + pmos/pnp.

[gus789]

Use a comparator + pmos/pnp.

Would you care to elaborate on that?

[batterypowerguy]

Thanks for the replies! If I understand correctly, the current mirror is a constant current source  - i.e. it adjusts the voltage (up to a limit)  so as to force a set current through the load. I don't think that is what I want. What I'm after is a current compliance circuit that prevents excessive current from damaging the load, but otherwise just behaves like a constant voltage source.   

500:1 current sense xformer, some 1% resistors and a comparator

[dannyf]

The comparator takes 1 input from a current sensing resistor, and another from your target current levels (a pot for example), and controls a pmos / pnp.

The choices of opamp and pmos vs. pnp will depend on your design requirements.

This approach assumes that the current sensing is on the return path.

If you insist on high side current sensing, you can retain the comparator and use an opamp with pnp / pmos to form a current sense amplifier - or use an IC there.

Using pmos is cleaner in that you don't need to worry about gate current / base current.

The basic topology can be found in any number of power supply designs posted here.

[T3sl4co1l]

Use a comparator + pmos/pnp.

Would you care to elaborate on that?

Don't use a comparator, it will oscillate.

You can still do current limiting with a current mirror, you just set the current higher than what the load requires.  It will saturate to, oh, maybe 50mV or less.  If that's not low enough, you can try an inverted transistor (swap collector for emitter), which can saturate down to single digit mV, but has low hFE.  (On the upside, low hFE means you can "suicide bias" it (fixed base current, nothing else) without much trouble.)

These options do tend to draw a lot of bias current (an ideal mirror needs the same bias in..), which shouldn't be a problem for testing purposes but isn't really a good idea for, say, battery operated stuff.

Tim

[gus789]

Thanks for the input. This is not any production or battery operated stuff. It is just an experimental setup. So stability is important, low power operation is not. We are trying  to measure the conductance of a "break junction" that is repeatedly formed/destroyed by crashing a thin wire against a metal coated slide using a piezo actuator. The wire is held at ~300mV, and the substrate is connected to the virtual ground of the photodiode amplifier.  The current limit circuit's function is to prevent excessive current during tip/substrate contact. 

[Marco]

Why don't you just put a constant current through the load and measure voltage to determine it's resistance/conductance?

If you need the 1V how about something like this (use a better opamp though, this one latched up in simulation) :

[gus789]

Why don't you just put a constant current through the load and measure voltage to determine it's resistance/conductance? If the 1V supply is a must then do it like this.

That is basically how I have it hooked up now, that is, the output signal is  proportional to the resistance of the junction. My goal, however, is to get an output signal that is proportional to the conductance (i.e. 1/R).  PS thank you for the circuit diagram - I'll have to try this one on the breadboard. 

[Marco]

Ewww, I just realized why the simulation was screwing up ... the reverse biased JFET on the opamp input start sourcing current when the current limit hits and differential mode voltage gets too large ... ehmm, that circuit needs some refinement (A CMOS opamp seems to work fine in simulation.)

[David Hess]

If you want to drive the output voltage independent of the current limit, then I would use a standard current limit circuit (lots of options here) with high voltage compliance and lower the compliance with a cascode transistor which sets the output voltage.  If the output current rises above the limit, the current drive to the cascode will collapse.

A JFET or MOSFET cascode transistor will allow a very accurate current limit if that is necessary.  A bipolar cascode could work but the base current would have to be limited or it will end up driving the load when the current limit collapses.  Voltage feedback from the cascode output will compensate for the Vgs or Vbe so the output voltage may be arbitrarily accurate.