Battery Discharge Current Rate

[geobabu]

Hi All,
I was just trying to implement a controlled battery discharge rate calculator, as shown in the below image. A 1KHz pulse with 5% Duty Cycle is applying to a MOSFET gate. Here I got some confusion regarding the power ratings of the resistors.


1) If its a normal continuous power system, if we neglect the R1 (.01Ω,3W shunt resistor) & MOSFET's RDSon (~100mΩ), The current I = 12.8/1 = 12.8A .
The power dissipation = 12.8x12.8x1 = 144W (Bigger power rating !)

Since here a 5% duty cycle is applying, therefore for 50usec a 12.8A is passing through the resistor and rest of the time its OFF. So while considering the power ratings , whether we need to consider 12.8A x .05 = 640mA therefore P = .64 x .64 x 1 = .4096W. So in safe side just a 1W resistor is required here ? or we need to consider bigger power ratings.

2) Is it also possible to replace the 1Ω , and use the 0.01Ω(Shunt) + 0.1Ω(RDSon) for the current path. In this case, while simulating , ~150A is showing as the peak current and from the datasheet the MOSFET's Pulsed drain current is 60A. (Suppose if we are using some other MOSFET version) .
So the current will be 150A x 5% = ~ 7.5A. And power dissipation in
Shunt Resistor = 7.5 x 7.5 x .01 = 0.56W
Across MOSFET = 7.5 x 7.5 x .1 = 5.63 W

If anything wrong is found, Please correct me & give some guidance in this issue.
Thank you ...
Regards,
Geo

 

[crutschow]

For a given duty cycle you multiply the power, not the current, by the duty cycle. Thus if the resistor is dissipating 144W when ON, as in your example, you multiply that by the 5% duty-cycle to give the average power or 144W x .05 = 7.2W. So you need at least a 10W resistor, not a 1W. The reason you can't multiply the current by the duty-cycle is that power is proportional to the square of the current in a resistor.

 

[MikeMl]

Let LTSpice do the power measurement for you!

Look at my version of your circuit, below:

I first ran the sim for exactly an integral number of cycles (.tran 10m), 10 cycles in this example.

I asked LTSpice to display the power being dissipated in (my) R1 by placing the cursor over R1 while simultaneously holding down the Alt key. Note that the cursor changes to a thermometer when the Alt key is pressed. When you left-mouse-click with the cursor on the resistor, LTSpice automatically generates the expression for power in R1= V(N001,D)*I(R1) and plots it as a trace with the units of Watts. Note that the peak power is indeed ~141W, but with a 5% duty cycle.

Next, move the cursor to the plot pane and place it over the ( V(N001,D)*I(R1) heading. Hold down the Ctrl key while left-mouse-clicking on the heading. That pops up the small box, which automatically computes the average power in R1 for you (7.036W).

When ever you think you know how to use LTSpice, there are always a hundred more tricks like this to do useful stuff...

 

[geobabu]

Thank you for helping me.
@ Mike : Really it was a new LTspice lesson for me !

Regards,
Geo

 

[dr pepper]

What they said.
Also in point 2 you mentioned high discharge at 60a, you also mentioned the rdson resistance of the fet, at this kind of current you also need to take into account switching speed, the fet doesnt instantly turn on it changes from off to on over time, during this time is acts like a resistor and disspates energy, so the dissipation of the fet will be higher than you think and at high ampages it can be a lot higher, also the inductance of the resistor if wire wound also plays a part here.
You can use a mosfet gate driver chip to control and speed up switching times, this design process can get tricky, and the board layout.
I'd reccomend you stick to dumping the energy into a load resistor, at least this way you can stick a fat resistor on to sort it out if it gets to hot, messing up calculations with fets usually ends up in bits of fet on your bench, anyone who's designed and built a switcher will have blown a few up (ahem).
P.S. with lead acids there can be issues with plate resonance with squarewave current discharge waveforms, you might want to look into this.