# Constant current load under varying input voltage

I am building a battery discharge system for LiPo batteries which accepts a charged battery and discharges it at a CONSTANT CURRENT regardless of the battery voltage, until a threshold is reached. The threshold is the LiPo storage voltage.

The purpose of the system is to measure actual delivered power under variable battery environmental conditions.

The question I have is regarding the load itself and the feedback system required to vary the load to meet the constant current requirement.

So what is the recommended load? e.g. motor or heater

How could the load resistance be varied in direct proportion to the battery voltage?

Any circuit references would be valuable.

(sorry I'm an ME not an EE)

• It's surprisingly easy if you only need the load to work between around 3v and 50v. Depending on the accuracy you need you could just do it with one zener diode, one npn transistor and two resistors. Search for "constant current sink". A motor or heater will not do it! Jun 25, 2016 at 22:09
• Can you explain the LiPo storage voltage - What voltage is it? What range is acceptable? What currents are we talking about? The load itself would be mostly the transistor, be it NPN, MOSFET or similar. If you are talking about dissipating over about 50 Watts then it'll get more complicated because you will need circuts to balance load across several. If you want accuracy of 1% or better then it also gets hard. But an op-amp feedback would achieve that much. Jun 25, 2016 at 22:50
• Some more details of the system under test: The LiPo battery systems are all assembled from one or more battery cells connected in series with rated voltages ranging from nominally 3.7v (1 cell) to typically 22.6v (6 cells). Between each cell is a connection used to balance charge the battery and to monitor the individual cell resistances. A 6s battery initially charges to 25.2v and fully discharges to 19.2v. To improve battery life, LiPo batteries are stored partially charged at 3.8v per cell or in this case 22.8v - this is the threshhold voltage to end the test. Jun 26, 2016 at 18:08
• LiPo capacity is the parameter under test - i.e. how long does the battery take to fully discharge from 100% charge to the storage voltage. The load will dissipate approximately 5 amps - 110W so heatsinks will be needed. Jun 26, 2016 at 18:12

If you want a constant current load, design a constant current load. As you say, this is a circuit that draws the same current over some range of applied voltage. That's your spec right there.

There are various ways to achieve this. The conceptually simple is to get a measure of the current, then use feedback that adjusts a pass element to keep the current constant. Some current sinks do work this way, especially if a separate supply is available to power the control electronics. For example:

All current entering at SINK goes thru R1. The voltage across R1 is therefore proportional to the current being sunk. Due to how the feedback is arranged, the opamp adjusts the gate of Q1 to maintain the same voltage across R1 as Vref. Vref is therefore the control input that sets how much current to sink.

There are details of stability, bandwidth, and accuracy beyond the scope of a simple answer here. However, such circuits do work and are used in reality. In this case, you'd want a separate power supply for the opamp in most cases.

Choosing R1 is a tradeoff. You want it large for good signal to noise ratio, considering the opamp offset voltage as one source of noise. On the other hand, you want R1 low for lower minimum voltage the current sink can operate at.

Another less accurate but much simpler current sink is:

This exploits the property of a bipolar junction transistor where the collector current is largely independent of the collector voltage over a decent range of collector voltages. As before, all of the sink current flows thru R1. In this case, the base current also flows thru R1, but that's small compared to the collector current due to the gain of the transistor. For example, if the transistor has a gain of 50, then about 98% of the current thru R1 is the sink current and the other 2% is the base current.

The voltage on R1 will be Vref minus the B-E drop of the transistor. At first approximation, that can be considered constant, around 700 mV. This kind of current sink is also used in real life when a few percent accuracy and some load on Vref are OK. This is the case often enough to make this a useful circuit.