Advise re options for logic-level SMD MOSFET

Thread Starter

Doceave

Joined Nov 16, 2019
49
Hi there

In an effort to slim down my design I hope for some advice re selecting smaller MOSFETs

Requirements:
> ESP32 will be tasked with driving 8 MOSFETs directly
> MOSFET should thus be happy with logic-level inputs
> MOSFETs should be happy with PWM frquencies of 1000Hz
> SMD package
> Loads expected ----- 12V 3A or 5V 1A

I am currently using IRL540's in TO-220 package which is a bit bulks and overspec'd

Thanks.
 

jpanhalt

Joined Jan 18, 2008
8,717
"Logic level" today is not what it was 10 to 15 years ago. Are you talking 5 V. 3.3/3.1 V or lower?

There are over 20,000 selections for single mosfets here: https://www.digikey.com/products/en/discrete-semiconductor-products/transistors-fets-mosfets-single/278?k=mosfet&k=&pkeyword=mosfet&sv=0&pv7=2&pv7=250&pv7=17&pv7=6&pv570=377049&sf=0&FV=-8|278&quantity=&ColumnSort=0&page=1&pageSize=25

Try defining your needs a little better. Start with N-channel or P-channel. What will your gate voltage be? Then decide on a package size you can deal with. Don't be misled by photos. Read the dimensions. Something the size of a fly speck may look like an SOT-23.
 

dl324

Joined Mar 30, 2015
9,986
I'd use AO3400 for N channel and AO3401 for P channel. Threshold voltages are 1.45V and -1.3V, respectively. They both handle several amps. Package is SOT-23.
 

Wolframore

Joined Jan 21, 2019
1,609
Those are nice, btw threshold is only where it conducts 250 uA so take that with a grain of salt, I wouldn't use it for much less than 2V Vgs. Even with a few go to's I end up spending a lot of time shopping for MOSFET's, there's always something else I need. I believe the ESP32 is 3.3V device. - Confirmed it operates from 2.2-3.6V.
 

dl324

Joined Mar 30, 2015
9,986
Those are nice
Those transistors are only 2 cents each on AliExpress. I wouldn't normally buy semiconductors, or any other components for that matter, from them, but inexpensive MOSFETs are otherwise nonexistent, so I took a chance.

The ones I tried work, but I haven't checked with a significant load. Breadboards are only good for an amp and I've been too lazy to wire up a test jig.
 

Wolframore

Joined Jan 21, 2019
1,609
I would love to see a test: 5A @ 5V and 12V. Steady and pulse... I would measure for voltage drop so we can calculate power dissipation and Rds.

My question is 5V @ 5A it's dropping 0.15V (assuming worst case 30 mΩ) That's 3/4 watts... can that SOT-23 dissipate that much heat?

RθJL (junction to lead) is approx 70C/Watt assuming 27C ambient in best case... I don't think it's going to like it.

Regardless of the hopeful datasheet, it's a good jellybean MOSFET for switching with some decent characteristics.
 

Wolframore

Joined Jan 21, 2019
1,609
An ESP32 drives the MOSFET so Vgs of 3.3V but I'd be happier with something that is closer to fully saturated at lower voltage.
Define fully saturated. Your saturation depends on what you're driving. 24-52 mΩ (at 3.3V) is a nice switch it has the resistance of about 12-24" of 24ga copper wire. I believe what you're referring to is gm or transconductance which has a relationship to voltage. It's a tiny MOSFET that has pretty incredible specs that uses basically no current to turn on and can pass as much as a couple Amps without issue that's very useful even at 3.3v.
 

jpanhalt

Joined Jan 18, 2008
8,717
According to Renesas thermal performance of the SC-59 package is (https://en-support.renesas.com/knowledgeBase/16981420 ):

1582053340608.png

Since the chart starts at 200 mW, that may the the highest practical dissipation. Type of board is not stated. 200 mW w/ Rgs(on) = 50 mΩ = 2A

You may want a bigger package or a really good heat sink or a better analysis of thermal characteristics. I did look for junction to case characteristic and junction ambient, but didn't find it right away.
 

Thread Starter

Doceave

Joined Nov 16, 2019
49
Wow this is a much more involved decision than I had expected ---- I will seek something a little more chunky and revert for expert opinions on my choice.
 

jpanhalt

Joined Jan 18, 2008
8,717
TI has been around for a long time. I do like its technical literature. Awhile back, I was prompted to read more about thermal performance of various packages. If you are trying to keep small, adding a 4 cm^2 heatsink seems counter productive. Source= http://www.ti.com/lit/an/spra953c/spra953c.pdf
In light of the fact that RθJA is not a characteristic of the package by itself but of the package, PCB, and
other environmental factors, it is best used as a comparison of package thermal performance between
different companies. For example, if TI reports an RθJA of 40°C/W for a package compared to a
competitor's value of 45°C/W, the TI part will likely run 10% cooler in an application than the competitor's
part.
And here is a link to some package-specific data: http://www.ti.com/lit/an/snva753/snva753.pdf Data for the SC-59 id not there, but it is for the SOT-23 (Section 2.2.1 and Table 2). Junction to ambient is about 200 °C/W. The number of legs can change that a lot. A 6-legt SOT-23 is much better than a 3-leg version.
 
Top