Difference between revisions of "Surface Mount FETs compared"
Latest revision as of 07:52, 27 December 2019
One particular electronics component that I often found difficult to pick for my projects is the venerable MOSFET. The FET (Field Effect Transistor) is part of the Transistor family of semiconductors. There are many types of transistors, but they basically fall in one of these three broad categories:
- Current-driven: BJT (Bipolar Junction Transistor)
- Voltage-driven: FET (JFET and MOSFET)
- Other (e.g. phototransistors)
In this overview I focus mostly on Enhancement-mode MOSFETs and present a selection guide with examples.
BJTs and FETs can be used for either switching or amplification, but generally one type performs better than the other. Fastest switching speeds can still be obtained by BJTs but due to the low power consumption FETs are often the better choice.
- BJTs require a constant current flow through the base in order for the transistor to open. This base-driven current flow can easily exceed the current rating of microcontroller ports and additional drivers (BJTs or FETs) are often required to enable higher drive currents. BJTs however outperform FETs in terms of amplification factor. Although BJTs can be used for switching purposes, they are not the most obvious choice when we need to save overall power consumption.
- FETs are voltage driven and can usually be connected directly to microcontroller ports. There is the factor of inrush-current if a FET has a high capacitance, but a simple series current-limiting resistor is sufficient to deal with that. FETs may also need to be selected to be compatible with the logic-level (TTL) voltages and may therefore also require additional drivers to enable higher (or lower) drive voltages. They are however a better choice than BJTs when it comes to low-power switching.
Lastly, before we get to our overview, there is one more distinction I want to address: JFET vs MOSFET. The JFET (Junction-FET) has certain benefits over the MOSFET. JFETs have a simpler production process and are therefore usually cheaper. JFETs also have higher gate input capacitance which makes them more robust to ESD (Electro-Static Discharge) compared to MOSFETs. However, since MOSFETs are the FET of choice for big players like Intel their prices have come down and characteristics have improved. It seems that, like BJTs, JFETs are slowly but steadily giving way to MOSFETs for many applications except where their niche is unchallenged.
(This page is a work in progress)
===Assumptions=== (the dangerous part of this page)
Typically, if we're using N-Channel FETs driven directly from a microcontroller, the drive voltage is probably going to be 1.8V, 3.3V or 5.0V. It is therefore best to select FETs that turn (almost) fully-on with the Vcc that we're powering the microcontroller with.
With FETs, I'm primarily interested in the ones that can switch on with mcu Vcc and offer a low RdsOn.
The FET package for my projects ideally should be SOT-23. These are the ones that I can still comfortably solder yet offer a maximum power rating (heat dissipation) of about 0.4W (that's about 80mA @ 5V, or 120mA @ 3.3V). In the datasheets, the 'Typical Output Characteristics' (Id vs Vds) shows the voltage drop (Vds axis) for a given Drain current. Multiply those and you'll get a feel for the heat dissipation in Watts.
The threshold value when the FET begins conducting. A bigger voltage difference between Gate and Source than this value will open the FET more. This value is usually shown in the Id-vs-Vgs 'Transfer Characteristics' graph. At the point where the curve (going up) becomes a near-linear line the FET is fully open. We generally seek the lowest voltage where the FET is about to become linear. The Transfer Characteristics graph however is not the best indicator what that 'fully-open' Vgs voltage is. The Transfer Characteristics graph often shows multiple curves showing how temperature affects the beginning of conduction. Furthermore, Vgsth is often the point at which only a very small amount of current (e.g. 250uA) begins to flow, in many real-world cases this is not the current we're interested in. Lastly, the Vgsth value is often measured at a nominal Vds voltage of about 10V, which in many low-power microcontroller circuits isn't always available. The better curve to show the FET-fully-on Voltage is the 'Output' or 'Typical On-State Resistance' curve. This curve often shows the Vgs voltages that are the tipping point between ever-increasing Rds (not enough Vgs) and Rds remaining nearly constant (sufficient Vgs to turn the FET on).
This is my own parameter that shows when the FET should be fully on. This means that the RdsOn value remains largely constant and the FET is in (or close to) linear (aka 'saturation') mode. The main graph in the datasheet that I use to obtain this value is the 'Drain-Source On-Resistance' curve (showing RdsOn vs Id).
The maximum voltage the FET can safely handle between the Drain and Source. For our purposes we'll mainly look at whether the FET already turns on at mcu Vcc voltages.
The maximum continuous current we can run through the Drain of the FET. This value is dependent on the maximum Drain-Source voltage. FETs running at a higher voltage need to lower their maximum current. The value for this characteristic is usually shown in the Id-vs-Vds(V) graph, aka the 'Safe operating area' graph. I'm usually only interested in the maximum continuous current rating without going over the maximum power rating of the FET (or SOT-23 package, i.e. around 0.4W). In practice, I look for FETs that can handle either up to 50mA or 100mA @ 5V or 3.3V.
The resistance between Drain and Source when the FET is fully open. The value for this characteristic is usually shown in the 'Output' or 'Typical On-State Resistance' curve (Rdson-vs-Id(mA)). As Vgs goes more and more above Vgsth (before the FET is fully-on) the lower the On-Resistance between Drain and Source (Rdson). When the FET is fully-on Rds remains fairly constant. If the FET is not yet fully-on, Rds is lower with low currents.
N-Channel, Enhancement-mode, Logic-Level MOSFETs SOT-23
Fortunately, for all these N-Channel FETs the pinout is the same:
|Si2302DS||0.65V||2.0V||20V||?||0.70R to 1.1R||could work with 1.8V, low cost||File:Si2302DS.pdf|
|NDS331N||0.7V||2.0V||20V||160mA @ 3.3Vds, 140mA @ 5Vds||0.7R to 1.5R||could work with 1.8V||File:NDS331N-D.PDF|
|5LN01C||1.0V||2.0V||50V||6R to 10R||high RdsOn||File:5LN01C.PDF|
|AO3400||1.05V||2.5V||30V||120mA @ 3V, 105mA @ 5V||0.18R to 0.54R||don't use below 2.5Vds||File:AO3400.pdf|
|BSS138||1.3V||2.5V||50V||100mA @ 3.3V, 70mA @ 5V||1.2R to 2.3R||don't use below 3.3V||File:BSS138-D.PDF|
|BSS123||1.7V||2.5V||100V||100mA @ 3.3Vds, 70mA @ 5Vds||0.8R to 1.2R||good for 3.3V and up||File:BSS123-D.PDF|
|BSH108||1.5V||2.6V||30V||75mA @ 3V, 45mA @5V||0.10R to 0.18R||good for 3V and up||File:BSH108.pdf|
|NTR4003||1.1V||2.8V||30V||0.9R to 1.2R||File:NTR4003N-D.PDF|
|2N7002||2.5V||5V||60V||?||3.2R @ 5V||High RdsOn, don't use below 5Vds||File:2N7002.pdf|
P-Channel, Enhancement-mode Logic-Level MOSFETs SOT-23
Fortunately, for all these P-Channel FETs the pinout is the same:
|BSH203||-1.1V||-2.5V||-30V||-80mA @ 5Vds, -105mA @ 3.3Vds||0.8R to 1.2R||could work with 1.8Vgs||File:BSH203.pdf|
|SSM3J332R||-1.2V @ -3Vds||-2.5V||-30V||-110 mA @ 5Vds, -200mA @ 3.3Vds||0.05R @ 3.3V||could work with 1.8Vgs||File:SSM3J332R.pdf|
|NDS352AP||-2.5V @ -10Vds||-3.5V||-30V||-100mA @ 5Vds, -105mA @ 3.3Vds||0.9R to 1.4R||don't use below 3Vgs, low cost||File:NDS352AP-D.PDF|
|BSS84||-1.9V @ -10Vds||-4.0V||-50V||-140mA @ -5Vds||8R to 12R||High RdsOn, don't use below 4Vgs||File:BSS84.pdf|
|Si2303BDS||-2.0V||-4.5V||-30V||90mA @ 5Vds||0.18R to 0.5R||don't use below 5Vgs||File:Si2303BDS.pdf|
Note: if you're after TO-92 package FETs I can mention:
- P-Channel: BS250P, VP0109N3-G, ZVP2106ASTZ
- N-Channel: BS107A, BS170, 2N7000