Current regulation circuit?

[bountyhunter]

He doesn't need constant current. If these are Li packs, he just needs to limit the current to a safe amount until the final (CV) phase of charge.

 

[Colin]

If the pack is intelligent, it will limit the current, provided you deliver more than 15.2v and 17v would not be enough for the overhead of the intelligence.

I would not charge above 0.5C (2 hours at 2.5amp).

Simply connect the power supply and increase the voltage to get 2.5amps and monitor the current for 2 hours.

 

[spec]

He doesn't need constant current. If these are Li packs, he just needs to limit the current to a safe amount until the final (CV) phase of charge.

The OP says he does require a constant current generator/current control circuit so that his voltage source does not trip out. This seems like a reasonable request.

There are many fast chargers especially for LiPo batteries which are designed to take high charge discharge currents. These are used extensively in models, both cars and aircraft.

In any case, if I just post a circuit the OP can use it or not at his discretion. The current will also be able to be set to any reasonable value.

The primary limiting factor will be the dissipation of the series pass element (PMOSFET in this case) which will need a low Thr junct/case and an appropriate heat sink, as has already been mentioned. If push comes to shove, more pass elements can be paralleled. This is all normal design.

spec

 

[schmitt trigger]

Very sensible answer, Spec

 

[MikeMl]

Here is an "add-on" 4A current limiter that introduces very little loss when the battery current is less than 4A. It comes at the cost of the power dissipation in the regulator transistor M1 when the load tries to draw more than 4A.



Look at the plot. The independent variable is the load resistance, which is swept from 2Ω to 10Ω. If it were not for the current regulator, when the load is 2Ω with a 17.5V supply, the current would be 17.5V/2Ω = 8.75A, while when the load is 10Ω, the current would be 17.5V/10Ω = 1.75A.

With the current regulator, as the load resistance increases, the current is limited to ~4.0A (green trace) until the load resistance reaches 4.3Ω, at which point, almost the full 17.5V supply voltage gets to the load, after which the current is just 17.5V/R(load), so the load current decreases as the load resistance increases. The total voltage lost across M1 and R1 is plotted as the blue trace, V(d), and once the load resistance is above ~5Ω is always less than 150mV.

Note the power dissipation in M1 (red trace) as a function of the load resistance. M1 will have to mounted on a large heatsink. M1 must be a modern power NMOS which has a low Rds(on). A zener is used to limit the voltage on M1's gate (between 0V and 10V) and to derive a reference voltage (~109mV) which, along with R1, determines the current limit = 109mV/0.027Ω ≈ 4A. U1 is any modern CMOS rail-to-rail amp.

I have not simulated the transient behavior. I suspect that some feed back compensation will be needed... I'll work on that when I have some time.

 

[spec]

Here is a constant current generator circuit. It is a bit more fancy than necessary for this application, but it is still reasonably simple to build and the components are low cost.

​

ERRATA
(1) Change R4 from 0R022 to 0R025
(2) Change R1 from 84K09 to 74K

CIRCUIT DESCRIPTION
(1) N1, a precision RRIO opamp, is part of a feedback loop which adjusts the voltage on the gate of the PMOSFET, Q1 to keep the voltage drop across R4 the same as the reference voltage across R1, which is 100mV. RRIO= Rail to Rail Input Output, which means that the opamps inputs work normally from the negative supply line to the positive supply line and the output voltage can swing from the negative supply line to the positive supply line.
(2) By Ohms law, 100mV across 25mOhm gives 4A which will flow out of Q1 drain. As the source and drain currents of a MOSFET are the same, the circuit produces a constant current of 4A from the drain of Q1.
(3) The string of three TL431s probably looks a bit odd but it gives a better accuracy and lower current consumption than using one TL431 and two precision resistors (TL431s have a rather high control current of 4uA worst case). Each TL431 is wired as a 2.5V Zener to give a total reference voltage of 7.5V, which is also the supply line voltage for the opamp. The TL431 are jelly bean price so the seeming extravagance is justified.
(4) Q2 and Q3 also form a constant current generator, in this case of 3mA. This current biases the three TL431 voltage references at around 1.5mA and provides the supply current for the opamp. Having a constant current supply means that the circuit consumes a constant current of 3mA regardless of input supply voltage.
(5) R1 and R2 simply divide the 7.5V reference voltage to feed 100mV to the non inverting input of the opamp, N1.
(6) The circuit may look a bit odd because the reference voltage is referred to the positive supply line rather than the 0V line, but this arrangement is quite normal and is used extensively in electronic circuits especially inside integrated circuits. The opamp does not care where it gets its input voltages from. All it cares about is making sure both of its inputs are exactly the same, in the case of the OPA192 within +-5uV nominal, by adjusting its output voltage, The OPA192 effectively takes no input current so there are no errors caused by input current (unlike the TL431s).

NOTES
(1) At 4A the dropout voltage (overhead) will be around 200mV
(2) The current accuracy is +-0.2% plus the errors due to R1, R2, and R4
(3) The input voltage range is 8.5V to 55V which is limited by the VDS of the PMOSFET, Q1
(4) The output voltage range is 0V to 54.8V
(5) The circuit current consumption is 3mA irrespective of input and output voltage.
(6) The limiting factor is the dissipation in PMOSFET Q1. With a suitably large heatsink, 50W would be possible. This means that at 4A constant current the maximum Q1 VSD is 50/4 =12.5V
(7) R4 is a 'four terminal' resistor to ensure accuracy. As standard four terminal resistors are expensive, you can wire an ordinary two terminal resistor as a four terminal resistor to get a close approximation.
(8) R16 is a gate stopper to prevent the PMOSFET from oscillating at high frequency. R16 should be connected directly to the gate terminal.
(9) The layout and connections should be as shown in the schematic to ensure accuracy and frequency stability.
(10) The capacitors play no part in the fundamental operation of the circuit. Instead they provide decoupling, except for C3 which tailors the open loop frequency response to ensure frequency stability. All capacitors should be physically connected as shown on the schematic.
(11) All capacitors are +-10% ceramic X7R dielectric types. The 22uF types should have a voltage rating at least 10% above the maximum input voltage, remembering to take into account any ripple voltage. But C3 need only be 10V or more. The 100nF capacitors should have a voltage rating of 10V or more.
(12) R1 is the theoretical value to give a 4A constant current and can be made up from 62K and 12K resistors in parallel.
(13) The PMOSFET, Q1 was originally made by International Rectifier but they were formally taken over by Infinion in January 2016, so you may see the manufacturer variously described as International Rectifier or Infinion. Infinion are progressively changing the IR data sheets to Infinion, but the technical data is unchanged.