Power Dissipation Sanity Check

Other Hardware General Electronics
1

I'm trying to work out how to rig up a basic circuit to effectively protect my projects from my own stupidity without losing much in the way of power. One microcontroller has already been sacrificed upon the alter of 'which way round did this power connector go?'

The idea is to have the unregulated battery pack on one end and the load on the other and have the protection circuit only allow power to flow if the battery pack in in the right way round. I know there are lots of ways to do this including rectification which would allow the battery to go in either way but a P channel power mosfet seemed to offer much greater efficiency and flipping the battery pack round isn't exactly a huge hustle.

I apologize for the following

The mosfet is IRF4905, the resistors are pair of 1W 50kOhm (not 47kOhm) resistors while the diode is a 1W 15v zener. The voltage input ideally could be anything from 4V up to 15V. I assumed 20V for my calculations for a nice oups margin.

The issue I have run into is that I am not entirely sure how to calculate the power dissipation in the different parts. If the battery is in backwards then the mosfet opens and power instead goes through the resistors and zener which are there to protect the mosfet gate from over voltage. Between those I am not sure how to work out the power dissipation so the following represents my best guess at how to work any of this out and seems to good to be true and therefore likely to be wrong.

Resistors: As far as the resistors go if I considered that if they were the only thing in the circuit and the input voltage is 20v (much higher than the 15v max I intend) then the worst case power dissipated would be (V/R)*V = (20/100k)*20 = 0.004W right? So I could easily get away with a 0.25W resistor.
This means I could do away with the two big resistors in parallel and use the 1/4W metal film resistors I have. Anything in series with them can only reduce the amps and/or voltage through so these are the most pessimistic numbers.

Diode: For the zener the voltage drop would be the zener voltage right? So with a 15v zener it would be 15v volts out of the 20v input. Current would be determined by the resistors in so far as I am aware so 0.0002 amps? That leaves 0.003W for the zener to dissipate. So ΒΌ watt is again fine.

Mosfet: For the mosfet. Fully on (at -4v on the gate) the resistance is 0.02 ohms. So, with a short at 20v that would likely desolder itself right off the board... but the load will be what determines the amps through the mosfet and since that load draw will be substantially less than 5A then 5A20v = 100W. The mosfet is rated for 200W so it is fine.
For the heat sink... Well, at 20v and 5A input and the full on resistance of 0.02 ohms, the voltage drop should be 0.1V through the mosfet. So 0.5W converted to heat?
Junction to ambient ta resistance is 62 so (taw)+ambient would be 62*0.5+30 or 61C working without a heat sink. That seems toasty, but not a problem, so a small heat sink would be enough?

The End? So as far as I can tell I could swap out my two resistors for a single 1/4w metal film resistor without issue, saving ample space. I could swap the 1W zener I have for an 0.5W zener since those were cheaper (I don't have any 0.25w zeners), and a small heat sink on the mosfet would be enough to keep things warm rather than scolding.

Someone please tell me where I went horrible wrong.
And ideally explain it in a way someone dreadfully sleep deprived could understand.

2

The fet listed is not a logic level type.
The 0.02ohm specification is at -10volt Vgs and pulsed (mosfet not getting hot).
At 4volt, the fet could be anything between 0.1ohm and not conducting at all.
Leo..

3

Oh? Was I wrong in reading the gate threshold voltage as min-2v and max -4v?
I thought all things being proper that would put 4v at the source, 0v at the gate, so -4v difference between them leaving the mosfet fully open.
While flipping it round the other way, it would be 0v at source, 4v at gate and so +4v difference between them so the mosfet must be closed.

4

The gate threshold voltage is the voltage where the fet just STARTS conducting.
You could say, the point where the fet turns OFF.
That threshold voltage varies per individual fet, and could lie somewhere between 2volt and 4volt.
To fully turn the fet ON, you need at least 10volt between gate and source (for this fet).

Leo..

5

The odds are that I am wrong but that video seems to suggest I am right.
video

So at -4v the mosfet is open and the resistance drops to almost 0.
He is using a FQP47P06 which has a -2v to -4v gate threshold voltage just like the one I am using.
What state is it I should be looking at to see the ideal on voltage?

So I am not entirely sure where I made a mistake there.

Edit:
Found this and starting to see a detail or two I missed.
Oups

From what I can tell the circuit should still work. It just may not be as effective over as wide a range of voltages as I had hoped.

Edit again:
Well, simplest way to find out if it works or not is to wire it and see and I can safely say I have either wired it wrong or it doesn't work at 8v

6

Ahh ha! I wired it wrong.

It works at 8v, 5v & 3.3v with a very small load. I will test higher loads and other voltages.

7

If you are going to limit yourself to 15V input, then you don't need the diode or the resistors. The gate is insulated so zero current flows in the steady-state condition. The only time that current flows in or out of the gate is when it's switching on or off.

The only way I know to estimate the mosfet's on-resistance for an actual drive voltage is to look at the top-left chart on the first page of charts in the datasheet. Pick the line that's closest to the drive voltage you have (eg, pick 5V) and see what the voltage is for the current you want. Other charts will give the resitance in ohms but I think that chart is top-left for every manufacturer's datasheet because it's the most important.

8

Actually I made the circuit.
I have been measuring the voltage drop across the source and drain and amps going through and it looks like the mosfet resistance is holding at about 0.6ohms when closed. A lot more than the ideal 0.02ohms but I can't say 0.6ohms is really all that bad.
It certainly seems better than the single diode alternative or putting resistance on the ground line using a nchan mosfet.

I shall see how it fares at higher loads and reconsider my alternatives from there.
And I am aware that the zener is unneeded if the input voltage is limited to 15v but this is idiot proofing. I'd throw the kitchen sink at it if it could potentially save my burning anything out through careless miss-wiring.

I'll have to take another look at schottky diodes. See if any offer a better option. They would certainly be simpler. Assuming I can find one with a high enough voltage and amp limits.