Is my MOSFET switch design sane?

[Oaklander]

I'm designing a controller for LED module operating at 24 volts. The module will use about an amp of current and is controlled by Atmega32.
After research I found that N-channel MOSFET used as low side switch is the best approach. I also learned that two resistors are recommended on the gate. One is pull down resistor and the other is in series with the gate to reduce some oscillation.
This was clear but the selection of the FET was not. My requirements were at least one amp of drain-source current, easy to solder SMD-package and logic level. There were hundreds of parts matching this criteria on Digikey and I settled on DMN6140LQ for now.

The circuit I designed is in the attached picture. The label Y is connected to AVR output and the connector on the top is connected to the cathode the LED.

My questions are the following?
1. Is my design sane or is there some error?
2. Are there some rules to select components like the FET in this example? There are so many nearly identical options. Is the best way to just select the cheapest?

[JustMeHere]

You need to figure out the power draw of your LEDs.  Also, I would lean toward a FET that has a heatsink.

[digsys]

FET selection is usually by critical use requirements.
IF you are not switching faster than say 10Hz, then it is primary just a power switch. The main spec I'd be looking at is Rds(on).
140mR is quite high, at 1A that is ~140mW, a lot for that package. It means you now have to worry about heatsinking.
Look for the lowest Rds(on) and if needed a slightly larger package. They may even be cheaper.

[Lt_Flash]

You need to put pull-down resistor before gate resistor, otherwise you are creating a voltage divider. The idea of that pull-down is to keep transistor closed when no input is connected to its gate, so it sinks current when you disconnect input and it doesn't matter if its doing it after gate resistor (current flows from MOSFET's gate to ground via these two resistors when you disconnect input). I know that because of your choice of resistors it doesn't really matter cause final voltage would still be almost the same, but if you put 1k-2k resistor in series with gate - the voltage would start to drop and that's not what you really want as many transistors require at least 4.5V to open and you don't want to lose any voltage. So just put that pull-down before series gate resistor and it should be fine

[IanMacdonald]

If you're driving from 5v logic you need to ensure that the threshold voltage is low enough to ensure full conduction. The device you've chosen is OK in this respect.
The gate resistors don't really do anything useful. If you want gate overvoltage protection, you need a series resistor and a ~12v zener to ground. Shouldn't be necessary if the logic supply is a guaranteed 5v though.
Always use a device that's a bit beefier than the current suggests. I'd suggest a TO220 device for this kind of application. Remember it may have to stand abuse such as shorts, as well as normal working. On which subject, don't overlook the need for fusing or current limiting.

[ogden]

I'd suggest a TO220 device for this kind of application. Remember it may have to stand abuse such as shorts, as well as normal working. On which subject, don't overlook the need for fusing or current limiting.

No need for TO220 in case of application with no switching losses. Just ensure low RDSon, thus low dissipation.

Mentioned transistor is fine because at Vgs=4.5V & Id= 2A 1.3A it's RDS_on is 0.17 Ohms. So at 1A dissipation will be less than 0.17 W, way below 0.4W max. Fuse is good idea indeed.

[Zero999]

I'd suggest a TO220 device for this kind of application. Remember it may have to stand abuse such as shorts, as well as normal working. On which subject, don't overlook the need for fusing or current limiting.

No need for TO220 in case of application with no switching losses. Just ensure low RDSon, thus low dissipation.

Mentioned transistor is fine because at Vgs=4.5V & Id=2A it's RDS_on is 0.17 Ohms. So at 1A dissipation will be less than 0.17 W, way below 0.4W max. Fuse is good idea indeed.

No, the maximum rating is much less than 2A! If you use this transistor for 2A continuously it will melt! A fuse will also do little to protect the transistor. In reality, the transistor is more likely to protect the fuse!

Don't pay any attention to the figures listed in the product summary! Look at the maximum ratings. It's clearly specified for continuous operation of 1.6A at 25^oC and 1.2A at 75^oC.

It will be fine as long as the continuous RMS current is below 1.2A. Note I said RMS current, which might not think is applicable for DC application, but if you're doing PWM, then it is important.

[Oaklander]

The FET must be surface mount and switching frequency won't be a problem. I will be well under one hertz.
How about this one? It should withstand much more current than needed, the Rds(on) is very small and the Vth seems to be low enough.

[ogden]

No, the maximum rating is much less than 2A! If you use this transistor for 2A continuously it will melt! A fuse will also do little to protect the transistor. In reality, the transistor is more likely to protect the fuse!

Don't pay any attention to the figures listed in the product summary!

Thank you for pointing out. Did not know that product summary can be blatantly misleading. I will fix original post

[edit]  Just checked that NXP/Nexperia does not cheat, quick reference data is identical to actual specs:  https://assets.nexperia.com/documents/data-sheet/PMN30UNE.pdf

Luckily I don't use Diodes components 

[Zero999]

No, the maximum rating is much less than 2A! If you use this transistor for 2A continuously it will melt! A fuse will also do little to protect the transistor. In reality, the transistor is more likely to protect the fuse!

Don't pay any attention to the figures listed in the product summary!

Thank you for pointing out. Did not know that product summary can be blatantly misleading. I will fix original post

I agree, it's misleading. It had me fooled too, until I applied Ohm's law and found the temperature rise would have been excessive.

I was about to post that the maximum ratings on the data sheet were BS, but then I realised the real maximum ratings are on page 2!

This does annoy me. It's like the manufacturers want you to use their device, so they make it sound better, than it is, but that's no good if you blow it up!

[digsys]

The FET must be surface mount and switching frequency won't be a problem. I will be well under one hertz.
How about this one? It should withstand much more current than needed, the Rds(on) is very small and the Vth seems to be low enough.

Much lower Rds(on) = 30mW = bugger all. (compared to 140mW). For extra margin, thicken any D/S tracks to take up any free space.
Leave the center part of the tracks exposed, for greater heat dissipation to free air. ie mask them off.

[Lt_Flash]

Well, usually that's why you're checking not only max current in datasheet, but also thermal characteristics. This particular MOSFET can cope with 3A:

VGS = 4.5 V; Tamb = 100 °C [1] - 3 A

With Note 1 saying: [1] Device mounted on an FR4 PCB, single-sided copper, tin-plated, mounting pad for drain 6 cm2

So if you have a look at thermal characteristics, calculate required heatsink to keep junction temperature below maximum - it should do just fine.

[Lt_Flash]

Sorry, my mistake in regards to heat sink - I thought that's already a TO-220 package. But anyway, if junction temperature would be less than 150C - it should be just fine.