# Power Dissipation VS Switching Power

It amazed me recently to have found 3 tiny Sot-232 transistors inside a 12V led controller. It can't be serious, but after digging and testing it with my 10 watts power LED mounted on a CPU heatsink for 30 minutes, the huge heatsink was hot to touch, yet when stupidly touching the tiny SOT-232 transistor, it was not even warm. It's a AO3400 N-Channel MOSFET; According to the datasheet, it's Max Vds is 30V, and Id is 5.8A.

Question 1, so technically is this tiny transistor capable of switching up to 30V * 5.8A = 174 watts?

Question 2, but according to the datasheet, it can only dissipate 1.4 watt. What is the difference between Power Dissipation VS Switching Power?

Question 3, Again, according to the datasheet, we only get 5.8A when VGS=10V; but since I am planning to use this with a MCU which has a HIGH voltage level of either 3.3v or 5v, what is the approximately expected switching current? and how it's calculated?

• But what is the voltage drop? Jul 16 '17 at 3:06

1. AO3400 30V N-Channel MOSFET ...Can it switch 30V * 5.8A = 174 watts?
• Yes that's the guideline assuming you consider heat and thermal resistance.
• based on Rds with drops with Vgs above threshold and often need 4x threshold.

2, but according to the datasheet, it can only dissipate 1.4 watt.

• yes but only if ambient =25'C then reduced to 0.9W @ 70'C

3.... @ 3.3v or 5v, what is the approximately expected switching current? and how it's calculated? ...

• Pd=Vds²/Rdson using Vgs to get Rdson from Fig 5 at max case temp Tc, where Tc_max=Tamb +Pd*Rθja where I choose max Tc= 85'C at max ambient at T=40'C

• Figure 5 gives nominal RdsOn then compute ; Vds, Pd and T rise from Thermal specs 100~125'C/W unless you choose better board conductor material or thermal design

Its a low Vds max switch with very low Vgs threshold, so it has more switching capacitance than others of higher V but the ratio of max power switching/ max dissipation is the almost same as ratio of max R load at max power to RdsOn minimum

e.g. at 30V 5.8A, load = 30/5.8=5.2R while RdsOn from Fig 5 at 85'C @5V=Vgs ~30mΩ

• thus switch power ratio = 5.2Ω/30mΩ = 173x

Relays often have a power ratio of 1k~2k while BJT's may be as little as 10x

• If Mods marked up good answers here like Jonk's then the point system would be more meaningful. Jul 16 '17 at 15:59

Must be folks are out at the bar, tonight. I'd have figured lots of answers showing up for this one in less than an hour. Nothing. So I'll provide my hobbyist view.

The datasheet can be found here: AO3400 N-Channel MOSFET. It specifies $R_{DS(ON)}\lt 28 \:\textrm{m}\Omega$ at $V_{GS}=10\:\textrm{V}$. Since you are using $12\:\textrm{V}$, it's probably fair to assume that they have access to a $V_{GS}$ of that level. You also mention a mounted $10\:\textrm{W}$ LED. So from this, I can presume that the current is $\le 1\:\textrm{A}$. It follows that the power dissipated in the NFET, based upon $I^2R$, is about $30\:\textrm{mW}$. The maximum thermal resistance to ambient is $125\:\frac{^\circ \textrm{C}}{\textrm{W}}$, so I guess we are talking $\approx 4\:^\circ$C over ambient here?

(It's possible that it is well-connected thermally to the board, as well. And perhaps closer to the typical thermal resistance than the maximum. So it's probably a fair bit less than that.)

1. If you flip a switch to turn on a $300\:\textrm{W}$ xenon lamp, do you expect the switch to get hot? No, you don't. Likewise, if you can use a transistor that will guarantee a low voltage drop, even a small one can switch a large current. Power dissipation is about the device itself, as a switch. And the figure you give, $1.4\:\textrm{W}$, is a maximum and only in the case when ambient is $25\:^\circ$C and where you only pulse it so that $t\le 10\:\textrm{s}$. You need to be careful here, as they are using a thermal resistance of $90\:\frac{^\circ \textrm{C}}{\textrm{W}}$, which I believe was based on the maximum shown in the first line under "Thermal Characteristics."

Now for question 1:

1. See #2 above. That discusses power dissipation of the switch, itself. So you already have half your answer. The rest, switching power, is a term I can't find on the datasheet. I suspect you may have read it in a context where it's discussing the load being switched (as in that $300\:\textrm{W}$ xenon lamp I mentioned earlier.) Or else you saw it in the context of a CPU where there is "static power" and "switching power," with the switching power being a (nearly) linear function of operating frequency. If it is of this last kind, then it's going to be about operating this device very fast. And since you haven't talked about that part of things, I won't either.

Now:

1. I don't think it's worth nuancing over, given how little you've discussed about your future intentions using it. So, looking at the datasheet as well as what you've mentioned about possible $V_{GS}$ value ranges, I'd just say that you should plan on no better than $50\:\textrm{m}\Omega$ for it. You don't say anything more about your loads, so all I can offer from here is to suggest that you carefully watch those thermal characteristics for the device and check Figure 9 regarding the safe operating areas.

Just to expand a little on #3, where you want to know the maximum current you can use. This depends on a lot of things. If you live in Adelaide, Australia, where the ambient can be insanely high and if your device might be used inside of a car which might be sitting on the dashboard on a hot day there in broad daylight for hours at a time, the answer could be $2.8\:\textrm{A}$. Why? Because the inside of the car can actually reach almost $100\:^\circ$C. And the datasheet appears to say that under no circumstances should you allow the device's internal temperature go over $150\:^\circ$C:

$$\sqrt{\frac{150\:^\circ\textrm{C}-100\:^\circ\textrm{C}}{125\:\frac{^\circ \textrm{C}}{\textrm{W}}\cdot 50\:\textrm{m}\Omega}}\approx 2.8\:\textrm{A}$$

And that is pushing the maximums. So you actually should not want to do that.

On the other hand, if the device will be used pressed up against a block of dry ice in a situation where the ambient will likely be $-50\:^\circ$C, then you could probably push more current through it. You might be allowed to reach that maximum current you mentioned. At least you'd be in the neighborhood.

Again, it's pushing the thermal maximums. So again you actually should not want to do that, either.

You also may want a "margin of 2" on the current. Pushing the maximums while the inside of a hot car is kind of nuts. So is doing the same with dry ice. To figure something more normal as an operating temperature -- say $85\:^\circ$C -- then I'd say you should not plan on more than somewhere around $1.5\:\textrm{A}$ to perhaps $2\:\textrm{A}$ for the device.

Oh. And I should note here that I did not take into account anything that shows in Figure 9, above. This is where you need to know something about your $V_{DS}$ (which you didn't discuss) and the duration of your on-time (which you also did not discuss.) But you will need to consider the safe operating area, too.

We are moving into an area where opinions may vary somewhat. That's mine, sight unseen.

Looks like a nice device and it looks like Arrow is selling them dirt cheap, too.

• 0, it's a 10W power led running at about 1.34A initially (more than 10Ws here) and slightly increases as it heats up; 1, so the answer is YES? 2, Still not exactly sure; 3, I am not sure that ambient temperature is more important than Vgs here. For instance, I am planning to make a simple 12V LED dimmer using a MCU running at 5V, so I suppose Vgs max is 5V. My concern is that if the transistor is not being switched on completely, then it could burn out. Another concern is that if not being turned on completely, LED not bright enough. Jul 16 '17 at 5:18
• @Atmega328 I'm just trying to get you to expand your mind a bit. Everything matters. Some more than others, depending on what's important to you at the moment. I gave you ways to answer your own questions. Or, I think I did. If you can deliver $V_{GS}\ge 3.3\:\textrm{V}$ then the datasheet suggests to my reading of it that you can expect $R_{DS}\le 50\:\textrm{m}\Omega$. What more do you need, other than making sure about the SOA and thermal?
– jonk
Jul 16 '17 at 5:26
• It's sort of confusing, especially with your big temperature equation; Regardless, if I am powering a 12V led with Vgs ≥ 3.3v, then I can get at least 12V/50mOhm= 240A, but of course, it's not true. Since power= I^2 * R, let's assume that temperature is at constant 25 c, then since the mosfet can dissipate 1.4 watt @ 25c, so then the max switching current is about 5A, >> 1.4 watt = I^2 * 0.05 ohm. so after all, what limits the max current is the temperature or Power Dissipation capacity for the mosfet? and similar to any resistors as well? Jul 17 '17 at 22:58
• @Atmega328 No, you can't get $\frac{12\:\textrm{V}}{50\:\textrm{m}\Omega}$!! That's only if you put the entire power supply across the switch! And it would NOT tolerate that for a moment before self-destructing! Darned right that's not true! You complained about the "big ... equation" but I also side-stepped all that equation stuff, so nothing to complain about really. I just wrote what I think you can manage -- maybe 1.5 to 2 amps. That is it for my opinion on it. However, if you want to come up with your own figures, then you need to read stuff and do equations.
– jonk
Jul 18 '17 at 0:00
• Why not. say I want to switch an inductor instead of a LED, then it's full 12v rail power. I think I got it now. My Q2, Power Dissipation VS Switching Power, they are basically the same thing. The max switching power, ie, current flow is proportional to the power dissipation. While a mosfet is a switch, it also can be considered as a 1.4 watt 50mOhm resistor in its simplest form. Jul 18 '17 at 16:48

It amazed me ...

nothing amazing here. the mosfet operates like a switch so what matters is its Rds when on - assuming that switching losses aren't a big factor here.

So if it has very low Rds(on), it can switch lots of current. its power dissipation has no relation to the amount of power being switched, other than the current being switched and the resulting power dissipation on the mosfet.