# Limiting current WITHOUT dropping voltage

I'm trying to limit a current of a 5V power supply to 100uA. I first thought of putting in a 50k resistor. That will never allow above the threshold.

Unfortunately, the load requires at least 4.5V and varies in the amount of current in needs. So this is obviously a too simple approach for the problem.

How can I extend this circuit to limit the current and keep the load voltage current independent?

• Sounds like a fuse would be just what you need, if 100μA fuses existed. I think the challenging part of this problem is reliably detecting 100μA. That's a pretty small current and you'll need to give some consideration to noise for a robust solution. Oct 7, 2016 at 2:44
• From 5V? That allows you to only drop 0.5V in the current limiter. See if you can get lucky and find a LDO regulator with adjustable voltage AND current limit. If you can relax the input constraints - say, starting with 9V in - that'll greatly widen your options. Otherwise - Spehro's approach looks good.
– user16324
Oct 7, 2016 at 10:13

Presumably you are okay with a small drop, provided it is less than 0.5V.

Here is a circuit that will limit at 100uA and will drop less than 100mV before it limits.

simulate this circuit – Schematic created using CircuitLab

The MCP6001 is an inexpensive rail-to-rail input/output op amp that will operate from a 5V supply. The op-amp will saturate at ground until the load current reaches about 98uA nominally (with the values shown). The supply thus 'looks like' 5V with ~1K in series (the MOSFET contributes less than 10 ohms with Vgs =-5V), so it will drop between 0 and 100mV for load resistances of infinity down to 50K.

For lower load resistances the circuit regulates the output current to ~98uA.

The circuit draws about 200uA from the 5V supply, in addition to the load current of 0~100uA.

• Are you sure about your calculations? Voltage divider (R2, R3) gives 238mV at non-inverting input of opamp thus it allows 238uA through 1k resistor (R1). Apr 29, 2019 at 11:58
• @Chupacabras Yes, I’m sure. Apr 29, 2019 at 12:37
• You are right, now I see where I made mistakes. One of them is I saw 49.9 ohms there, not kiloohms. My bad. Apr 29, 2019 at 13:41
• Rail-to-rail is relevant because the inputs have to be within the op-amp's common mode range- a part such as LM324/358 will not work well in this application without much more voltage across R1. Input bias current is a direct error on the current. At 1mA any op-amp is okay, at 100nA many are not. Offset voltage and tempco of the Vos (TCVos) are important if you want to minimize the voltage across R1. At 1 volt, any op-amp pretty much is okay- but at 10mV many will contribute a lot of error. May 28, 2020 at 13:55
• @SpehroPefhany oh, sure - thought this was a new question as it was on top, but I visited the wrong SE landing page so it was only on top because of some activity, sorry about that. Oct 23 at 11:39

The Ohm's Law police will come after you if you attempt to regulate both voltage and current into a fixed load.

Unless you have a very unusual load, the load will draw whatever current it needs if you supply the correct voltage. Any attempt to reduce the current will reduce the applied voltage.

• It's for simulating a power supply that can give 4.5V/100uA. The load sometimes has spikes with more than 100uA. There is a decoupling capacitor to handle that but I skipped all that to try to focus on the main goal (getting from 5V/100mA to 4.5V/100uA power supply). You think it's important for the question? Oct 6, 2016 at 23:30
• What kind of supply are you trying to model? A battery limited by internal resistance? A benchtop supply with a foldback circuit? A wall wart with a fuse? For each one you would want to model it a different way. Oct 7, 2016 at 0:28
• @ThePhoton I want to simulate a point in the I-V curve of a solar cell. Oct 7, 2016 at 14:12
• So just a quick enquiry. Would you be able to regulate the current if you figured out the impedance of the circuit and finding the correct voltage? Feb 22, 2021 at 19:57

Generally your current limiting circuit will require a bit of voltage "headroom" to operate in. That means you'll need an unregulated PSU of > 5 V and regulate it down to 5 V while monitoring the current.

It's a while since I've read up on the old LM723 voltage regulator but they offer voltage and current limiting. These were very popular once upon a time so you should find plenty of sample configurations on a web search.

Figure 1. Basic Low Voltage Regulator (VOUT = 2 to 7 Volts). (Figure 4 of datasheet).

It's a bit late but I would use a current regulator diode S-101 from Semitec. S-101T is the SMD version and they are available in a range of currents.

The simplest solution is NCh JFET with gate shorted to source.

The spec is called IDSS with Vgs=0

Here are some choices, which require ESD precautions and reverse voltage failure.

http://www.digikey.com/product-search/en/discrete-semiconductor-products/transistors-jfets/1377093?k=&pkeyword=&pv609=99&pv609=59&FV=fff40015%2Cfff80345&mnonly=0&newproducts=0&ColumnSort=609&page=1&quantity=0&ptm=0&fid=0&pageSize=500

I see these are either not in stock, obsolete or last time buy

Plan B

quick and dirty solution

LM317 cct. for 0.1mA R1 = 1.25V/0.1mA = 12.5k

• Can you share a circuit reference for NCh JFET with gate shorted to source? Oct 6, 2016 at 23:25
• I don't have any handy. It's common EE knowledge widely documented on web. Pick any datasheet I linked to read IDSS but also becoming OBSOLETE. Generally we must use a stable Vreg regulator CC design to be a constant current depending on V and I tolerances unknown. Look at LM317 and CC design Oct 6, 2016 at 23:43
• Current regulation isn't the same as current limiting Oct 7, 2016 at 0:42
• ... and neither of these solutions meets the requirement of zero overhead voltage. Oct 7, 2016 at 0:58
• The LM317 current regulator definitely won't work at 100uA - it needs some mA, depending, in part, on the input voltage, and it will need to drop several volts (1.25V plus the regulator drop-out voltage) in order to work. Oct 7, 2016 at 6:36