How come my MOSFET was a "smoking", I thought it was rated high enough.

The thermal resistance is the temprature rise per watt, so at 19 c per watt, you have two wats of power and the temprature rises by 38 degrees.

apples:
Okay I have made up a diagram, and have noticed a wrong hook up. You will see a red line where it is wrong.

Yes it is connected to a PWM output on the arduino.

I have a smaller mosfet to turn the bigger one one, like a darlington pair I guess.

See the schmatic.

Yes, that is a mess and the red wire is confusing, it's wrong because it's there or it's wrong because it wasn't there and you added it? Either way that is not a good gate driving circuit because it's relying on a passive pull down resistor to turn the large mosfet off, which means it has to allow for a time constant of 10k ohms X the mosfet gate capacitance to fully turn off, which means the mosfet is dissipating significant heat while in that slow transition time from on to off.

A far better solution would be to just use a 'logic level' power mosfet to begin with and eliminate the need for a 'driver' mosfet. Here is an example that would work well. N-Channel MOSFET 60V 30A - COM-10213 - SparkFun Electronics

Lefty

ahh, so instead of being the 120deg as mentioned before with no heatsink, with the heatsink it would only be 38deg then?

Have a read of this
http://www.thebox.myzen.co.uk/Tutorial/Power.html

No one seems to have mentioned it, but your driver ckt is completely wrong.

You do not bias MOSFETs the same way you bias BJTs. A Darlington ckt is
wired as it is to increase the base-current drive into the 2nd BJT, ie IC1
becomes IB2.

What you need to do with n-MOSFETs is make the gate-source voltage large
[ie, typically 10V plus]. With your ckt, you are actually losing gate-source
voltage for your 2nd MOSFET.

  1. as others mentioned, try a logic-level MOSFET, eg IRL540, where you'll
    get a good turnon with Vgs=5V.

  2. go look up how to design a proper MOSFET driver.

OK that driving circuit is plain wrong. The first device (for which you've used an NPN symbol, not an n-channel FET) is in source-follower mode - this is no good at all, it won't pull up the second devices gate more than a volt or two.

The first device must be a common-source stage, with a drain load resistor to the +12V rail (something like 1k perhaps). This will level convert from 5V to 12V (and invert the logic). This can then drive the second devices gate at suitable levels to actually turn it on.

yeah righto , totally confused now, lol.

This is where I get my parts from as the hve a shop in town. http://www.jaycar.com.au/index.asp

Can some one find me a suitable mosfet, the one you mentioned the IRL540 is not listed (I do not think) .

MarkT:
The first device must be a common-source stage, with a drain load resistor to the +12V rail (something like 1k perhaps). This will level convert from 5V to 12V (and invert the logic). This can then drive the second devices gate at suitable levels to actually turn it on.

This looks like the right circuit to drive a 12V 3A load for my project (LED ribbon). Can I drive the load at 2Khz (switch on/off 2000 times a second) with this circuit, without wrecking the FET?

Also, are there alternatives to this circuit, e.g. a single component that integrates an equivalent of this whole circuit into it -- logic-controlled high-amperage 12V amplifiers in a single package? (even if a few dollars extra)

EDIT -- Found a ready made multiple-channel Arduino amplifier board suitable for 12 volt control.
http://www.elechouse.com/elechouse/index.php?main_page=product_info&cPath=&products_id=1812

You want almost any n-channel MOSFET in TO220 package with
an "L" in the part-number.

MarkT showed a reasonable MOSFET driver ckt. It has an inverter 1st stage,
so the I/O pin works backwards, ie "1" output turns off the load.

how about a BUZ71
www.jaycar.com.au/images.../BUZ71.PDF

has a vgs of max 4v

should only need one as it can switch 50v 14A

ohh, and a heatsink

How about posting a link that works?

http://www.jaycar.com.au/productView.asp?ID=ZT2225

there is a pdf data sheet there

No that FET needs 10V to turn it on. The gate threshold is where it just starts to turn on not when it is on fully. Look for the voltage the quote when they quote the on resistance.

By the way if you switch a 50W load the switch does not dissipate 50W, only the on current flowing through the on resistance.

Forget Jaycar, I've had a quick look at their range and none of their power mosfets appears to be logic level. Try one of these http://au.element14.com/jsp/search/browse.jsp?N=204245+110158845+110169484+110158844+110137572+751&Ns=P_PRICE_FARNELL_AU|0&locale=en_AU&appliedparametrics=true&getResults=true&suppressRedirect=true&isRedirect=&originalQueryURL=/jsp/search/browse.jsp%3FN%3D204245%26No%3D0%26getResults%3Dtrue%26appliedparametrics%3Dtrue%26locale%3Den_AU%26divisionLocale%3Den_AU%26catalogId%3D%26skipManufacturer%3Dfalse%26skipParametricAttributeId%3D%26prevNValues%3D204245.

yeah I agree I have been chasing my tail looking up numbers and I too think they are limited in what they sell.

dc42, wow great link, and I can now browse them all via the rds(on) voltage. cool.

One thing to watch for is that Rds(on) is often quoted at less than the maximum rated drain current. Always look at the datasheet and check that Rds(on) is quoted at or above the maximum drain current that you want to switch, for Vgs = 5V or lower.

btw I should have checked the 4V and 4.5V boxed for Rds(on) test voltage as well as the 5V box.

Many mosfets have Rds(on) quoted at both 10V and at 5V or 4.5V, and in these cases the Element14 site often indexes them under 10V. So there are probably many other suitable mosfets available at Element14.

If you definitely won't need a heatsink, then you can consider power mosfets in packages other than TO220, for example IPAK. One of my favourites is http://au.element14.com/international-rectifier/irlu8726pbf/mosfet-n-ch-30v-86a-ipak/dp/1698317 (Rds(on) = 8 milliohms max @ 4.5V 20A).

mdrejhon:
This looks like the right circuit to drive a 12V 3A load for my project (LED ribbon). Can I drive the load at 2Khz (switch on/off 2000 times a second) with this circuit, without wrecking the FET?

Also, are there alternatives to this circuit, e.g. a single component that integrates an equivalent of this whole circuit into it -- logic-controlled high-amperage 12V amplifiers in a single package? (even if a few dollars extra)

The only issue with my circuit is that 1k pull-up resistor - it limits how fast the IRF3205 can turn on. The total gate charge is about 150nC for the 3205 so it'll take on the order of 10us to pull high - at 2kHz that means about 2% of the time will be in this slow switching zone. For 12V and 3A that'll be OK (the rough estimate for dissipation during switching is VI/4 - so at 2% of the time that averages 0.18W)

The 1k resistor can be made a lot lower in value, speeding up the switching, but it'll then be dissipating more power (if 100 ohms it'll dissipate 1.44W).

The simplest alternative to the circuit is using a logic level FET in place of the IRF3205 - then the driver FET can be eliminated. You still have limited current to drive the gate (Arduino outputs absolute max is 40mA so the lowest value gate resistor you should use is 150 ohms)

For more demanding loads you can use a MOSFET driver chip like the MIC4422 between the Arduino and the power FET - you'll get much faster switching but you'll need good decoupling on the driver chip. A gate resistor of 10 ohms or less would then be sensible giving 100ns switching or so.

dc42:
One thing to watch for is that Rds(on) is often quoted at less than the maximum rated drain current. Always look at the datasheet and check that Rds(on) is quoted at or above the maximum drain current that you want to switch, for Vgs = 5V or lower.

Rds(on) doesn't change much with current if the device if Vgs >> Vds and you can model it as a simple resistance. Furthermore you won't normally go anywhere near the max current rating in a sensible circuit :wink:

For instance a 0.01 ohm device at 50A will have Vds = 0.5V - if Vgs = 10V then Vgd = 9.5V and the channel is basically the same geometry as it is at zero current (where Vgd = 10V). It will be dissipating 25W too which means v. good heat sinking needed. For
50A I'd be tempted to go for lower Rds(on) or multiple FETs - thus reducing Vds.

So I'd reword this as "calculate Vds(on) and check its a lot less than Vgs(on)"

MarkT:
Rds(on) doesn't change much with current if the device if Vgs >> Vds and you can model it as a simple resistance.

You can't calculate Vds(on) and know that Vgs >> Vds until you know that you can model the mosfet as a simple resistance. So your argument is circular.

Data sheets always quote Rds(on) at a particular drain current. Where Rds(on) is quoted for two different values of Vgs, Rds(on) at the lower Vgs is typically quoted at a lower drain current - often much lower than the current rating of the device. The reason is that at lower Vgs, the drain current at which the mosfet transitions from resistance mode to transconductance mode is lower.

By ensuring that the current you are switching is not greater than the value at which Rds(on) is quoted for the Vgs you are using, you ensure that the mosfet is fully turned on for that drain current and the Rds(on) figure is valid to use in your calculations.