Understanding MOSFET datasheets

Hello.

I am planning on purchasing some mosfets for a number of projects I have on the go at the moment. I want to make sure that I am buying the right mosfets for the job, because these things can be rather expensive.

I want to make sure that I am reading and understanding the datasheets, and was hoping some of you people could clarify things for me.

The basics that I need to know are the following:
Maximum continuous voltage the mosfet will allow from Source to Drain
Maximum continuous current the mosfet will allow from Source to Drain
The voltage / current needed to fully turn the mosfet ON
The turn on / turn off time

Do mosfets work strictly fully ON or fully off? Or do they operate like a transister, where there is a region that it is only slightly on, depending on the gate current / voltage?

Also, will the mosfet turn off when the gate voltage is dropped to 0? Or does the source voltage have to be 0 as well for the mosfet to turn off? I remember there is some type of device that only needs a pulse at the gate to be turned on, and won’t turn off until the source voltage = 0… I just don’t remember if this is how a mosfet works too.

Here is a mosfet I was thinking of buying… If anyone could tell me the vital information that I need to know about it, that would be very helpful.

IRFP460 mosfet

Here is a link to the DATA SHEET

Thanks again!

mosfet datasheet.pdf (103 KB)

The maximum specifications are just that, maximum that can't be exceeded. They are considerably higher than what it will run with continuously.

MOSFETS are like transistors in that they have part on areas. Read the data to see what the series resistance and possible current flow at a particular gate voltage.

These are curves. EG, at 2 volts on the gate the current is very low. At 5 volts it may be 5A while at 10 volts it may be 50A. It varies according to the device.

With the arduino, you need is a logic level gate mosfet. The gate needs to be fully on or the resistance of the drain/source is not at its least so causes heat.

My favourite small mosfet is NTD5867NL. It works with 5v on the gate and handles a few amps with ease. They are cheap. I bought mine from RS Components.

There are others but make sure they are fully on at 5v on the gate.

When you remove the gate voltage, there is a time taken to discharge the gate capacitor (internal). Therefore it is best to have a discharge resistor from gate to Gnd. This is important especially when you use PWM as it will get excessively hot without the R.

Weedpharma

localbroadcast: Do mosfets ... operate like a transister

weedpharma: MOSFETS are like transistors

Not to be too picky, but mosfets are transistors ;)

Yea.. I've been doing some reading, and I think for my purposes I'll be needing a mosfet driver. There's so many different drivers to choose from.. it's hard to decide! Any suggestions on a driver that works well with the arduino outputs?

JimboZA: Not to be too picky, but mosfets are transistors ;)

Good to meet a fellow pedant!! ;)

Weedpharma

JimboZA: Not to be too picky, but mosfets are transistors ;)

yea.. yur right.. but I hope you all understood what I meant by what I said.. I was just comparing mosfets to yur standard sized bipolar transistors..

So what value in the datasheet shows the voltage that needs to be applied to the gate in order to turn the mosfet completely ON?

After doing some reading.. it sems like the gate pin operates like a capacitor.. and the mosfet turn on time is dependent on how quickly the gate is charged.. so a higher current applied to the gate results in a quicker turn on time. This is the Q value in the datasheet, given as a value measured in nC. Q = IT, where T is the turn on time in seconds, I is the current applied to the gate, and Q is the value in the datasheet given by Qg or Qtot.. Some datasheets give more than one value for Q, which is confusing.. Qtot Qgs and Qgd... I'm not sure what the purpose of all 3 is yet.. still need to figure that out..

That capacitance stuff matters mostly when you are going to use a very high PWM frequency - like megahertz.

It's less important at lower frequencies typically used for driving robot motors and that the arduino can practically generate.

So what value in the datasheet shows the voltage that needs to be applied to the gate in order to turn the mosfet completely ON?

That mosfet is not a great choice. Look at Fig 1 in the first post you linked. If your gate voltage is 4.5 volts, you will only get 4 amps through the circuit, even for large Vds. This implies a fairly high apparent resistance and therefore heating. To turn that on, as far as it will go, you need a gate voltage of around 7 or 8 volts.

If you look at Fig 8, that fet looks good for controlling voltages around 100-200 volts but not so good for controlling 12 V or 24 V to a robot because the on-resistance is rather high.

The Rds resistance of your mosfet - even with 10 volts on the gate voltage, is 270 milliohms at 12 A load current.

The Rds for the one weedpharma prefers, with only 4.5 V on the gate and a 10 A load current, is only 50 milliohms.

That's 80% less heat you are generating, right there. And the performance of your mosfet will be even worse than that, if you can only drive the gate to 5 volts.

michinyon: That capacitance stuff matters mostly when you are going to use a very high PWM frequency - like megahertz.

It's less important at lower frequencies typically used for driving robot motors and that the arduino can practically generate.

Except that it matters from a few kHz upwards... Switching losses for a high power load will dominate, and to keep them under control you normally want the switching time to be < 1% of the PWM cycle time, so for 8kHz PWM that's 1us or better. That usually means needing 10's to 1000's of mA drive to charge the gate capacitance (for instance a gate capacitange of 2nF charges to 10V in 1us with 20mA or more, but a large MOSFET with 20nF needs 0.2A

The key parameters you need to look at in a datasheet are:

Vds(max) Rds(on) and the Vgs at which this is achieved Qg (total gate charge) for the gate voltage you use.

You may also be interested in the body-diode's performance and the Vgs(max) and the plateau voltage (usually only hinted at in graphs). Its best to use a gate drive voltage about twice the plateau voltage as this gives roughly symmetrical switching times.

You don't need to look at the max current because if you do your sums right with Rds(on) you'll never be anywhere near this, its an extreme thermal limit. And if you have an Internation Rectifier datasheet they lie blatantly about max current anyway.

well.. I'm not using the mosfets to drive 24 volt robot motors.. I'm using them for a load at 100v - 150v at 10amps. So I need the bigger mosfets.

I am thinking of driving the mosfet gate using a push-pull setup with npn pnp BJT transistors.. This way the arduino can trigger the transistors, which can provide a high current to the mosfet gate, allowing for fast gate charge / discharge of the mosfet.

I am trying to find a 3 or 4 amp transistor that the arduino can turn on.. and from what I'm reading, the arduino can only provide about 5 volts @ 20mA to do this..

I was looking at the MJE243 MJE253 npn pnp transistor set.. but I'm not sure the arduino can provide enough current to completely turn it ON.. Here's the datasheet stats for those transistors:

Collector−Emitter Saturation Voltage = 0.3v , 0.6v (IC= 500 mAdc, IB= 50 mAdc) (IC= 1.0 Adc, IB= 100 mAdc)

Base−Emitter Saturation Voltage = 1.8V (IC= 2.0 Adc, IB= 200 mAdc)

Base−Emitter On Voltage = 1.5V (IC= 500 mAdc, VCE= 1.0 Vdc)

From this information.. what is the current required to fully turn this transistor on?

I’m sure we’ve been here before with this, but I’ll suggest these drivers again to you from that other thread we were having discussions.

And to drive them nice and cleanly with these UCC27424P MOSFET drivers
The drivers are high speed dual drivers with a paralleled MOSFET and bipolar transistor complimentary output for good sourcing and sinking of main power MOSFET gate charge.

This driver can be driven directly by logic, I’ll be connecting some of these to a DUE soon.

They also have a nice feature, with an enable pin, which you can connect back into some other safety cut-out system if you need.

I think I paid AUD$1.35 each from element14 and recieved another tube of them today. UCC27424P

These drivers have an output stage that has both bipolar and MOSFET devices in parallel.
Check out the datasheet for the specs.

Surely, it’s got to be easier to use such a device as this that is designed and proven to work with power MOSFETS, than to mess about with discrete components.


Paul

rockwallaby.. yes I ordered a handful of those drivers. So they work fine with the pwm from the arduino? Also.. I hope they work with the mosfets I ordered.

IRFP460 20A 500V N Channel MOSFET Transistor

Any advice how to wire this driver up with my arduino and this mosfet?

localbroadcast: well.. I'm not using the mosfets to drive 24 volt robot motors.. I'm using them for a load at 100v - 150v at 10amps. So I need the bigger mosfets.

Then you should seriously consider changing over to IGBTs as a possibility.

localbroadcast: I am thinking of driving the mosfet gate using a push-pull setup with npn pnp BJT transistors.. This way the arduino can trigger the transistors, which can provide a high current to the mosfet gate, allowing for fast gate charge / discharge of the mosfet.

No, use proper MOSFET driver chips, they work :)

I am trying to find a 3 or 4 amp transistor that the arduino can turn on.. and from what I'm reading, the arduino can only provide about 5 volts @ 20mA to do this..

Not likely, you need fast switching so darlingtons are out of the question, and single BJTs don't have the gain. The only option is a pair of MOSFETs to drive the gate on the big MOSFET. Funnily enough that's exactly what you get in a MOSFET driver chip....

I was looking at the MJE243 MJE253 npn pnp transistor set.. but I'm not sure the arduino can provide enough current to completely turn it ON.. Here's the datasheet stats for those transistors:

Collector−Emitter Saturation Voltage = 0.3v , 0.6v (IC= 500 mAdc, IB= 50 mAdc) (IC= 1.0 Adc, IB= 100 mAdc)

Base−Emitter Saturation Voltage = 1.8V (IC= 2.0 Adc, IB= 200 mAdc)

Base−Emitter On Voltage = 1.5V (IC= 500 mAdc, VCE= 1.0 Vdc)

From this information.. what is the current required to fully turn this transistor on?

Undefined, you haven't said what collector current or what power loss you can tolerate.

Just get a MIC4422 and some good ceramic decoupling for it. At that power level (> 1kW load) you need really clean fast switching - during switching transients the peak power dissipation in your MOSFETs will be similar to the load. You keep the switching time down to 100ns or so to prevent nasty things happening. The MIC4422 can drive many amps if needed.

Use a fast opto-coupler on the input to the MIC4422 (with that power level if you blow a MOSFET the MIC4422 will explode too, you don't want to take out everything else and give yourself a lethal shock either). Something like ACSL64000 series?

If you are PWM'ing the load you'll need to calculate your losses back-of-the-envelope style...

MarkT, why suggest using IGBT without giving us a valid reason over using MOSFETS.
I think we all have some agreeance with a MOSFET being a suitable device for the project.
As with standard transistors, you can can’t simply parallel IGBTs as you do MOSFETS.

Also, MarkT, it appears that you haven’t read the post prior to yours, where I mention the use of a MOSFET driver, specifically the UCC27424P, which is based on the basic variety of XX4422 style drivers, but is way better.

The one I specified has the extra driver enable pin as well, it can take logic inputs direct from as low as 3.3v.

localbroadcast, yes with what I have seen, this driver will work fine with your MOSFET.
What I suggest you try to do is to use a drive voltage of higher than 5.0 volts, say something closer to 10v or 12v. The reason is that, even though you may have a nice chunky 5.0v source, any MOSFET driver will not swing its output rail to rail. It will be some point below 5.0v, and too close to the gate threshold voltage of the power MOSFET.

If you have as part of your design a 12v supply, then use this as the Vcc of the driver chip to give the a good gate voltage on the main power MOSFET. The driver chip will handle the higher charge and discharge of the power MOSFET quite well.

Using 12 volts on the driver chip will still allow you to connect the Aruino directly to the input pin.
These drivers have two separate drivers within them, so you can use two power MOSFETS, or, if you want, you can parallel the two drivers stages foe even greater current source/sinking for really demanding power MOSFETS.

MarkT suggests the use of an opto-coupler, I wouldn’t suggest this, just connect your Arduino output pin directly to the input pin of the driver you have. Adding more circuitry for extra protection can be worthwhile sometimes, but can actually cause more nuisance other times with extra complexity and reducing performance.

Do put a fuse or some form of safety circuit in the high voltage side.

With any MOSFET, be careful with static when handling the device before placing it in circuit, the gate insulation layer is susceptible to damage. So don’t walk around in synthetic clothing rubbing your feet on the ground.


Paul

The voltage the MOSFET will be switching on / off will be about 115Vdc. The load current will be about 6 amps. I want the load to see 36Vdc, so the 115Vdc output pulses from the MOSFET will be filtered, giving hopefully a smooth 36Vdc.

So now I understand what the Max power dissipation means… It’s basically the maximum allowable power loss due to the Drain-source resistance Rds(on).

Obviously I want this loss to be as low as possible. From what I can tell, this loss can be lowered by choosing a MOSFET with a lower Rds(on). Also, keeping the MOSFET as cool as possible during operation will help because the Rds(on) increases with temperature. I can’t think of any other ways to minimize this loss… anyone have any other ways to lower this loss?

The gate threshold voltage VGS(TH) and the gate source Voltage for ON state VGS are different values.
For the MOSFET 2SK3502-01MR:
VGS(TH) = 5V
VGS(ON) = 10V

For example:
if VGS = 7.5V
and the full current potential form the drain to source = 6A
and the full voltage potential from the drain to source = 50V

So if VGS is between 5V and 10V, then the Drain-source current / voltage is affected…
Will the full 50V still be applied to the load, and the load current is reduced to something like 3 Amps? Or will the full 6 Amps make it through the load and the voltage is reduced to something like 25 Volts?
Or will both the current and the voltage be reduced by some amount proportional to VGS?

Since I want the MOSFETs to work strictly as switches, I want the gate voltage to always be 10V. Does the output of the MOSFET driver MIC4421 you mentioned provide this 10V? Some MOSFETs have 15v gates… Does the same driver work for those as well? I guess my question is… How do you get the proper gate voltage from the driver? Do you need to provide a separate voltage source for the driver in order to provide the voltage and current needed? I know that the arduino PWM outputs will not provide enough power to do the job.

Also… do you know the part number for an equivalent HIGH-side mosfet driver?? I prefer to have the MOSFET before the load…

Sounds like you've been doing a lot of reading and making good sense of it all. ;D

As far as the driver output voltage, like I mentioned, you need to supply this, as a good solid supply, so, if you have 12Vdc available, and it has good current capacity, then let this be the supply to the Vcc of the driver chips, which will then be the output swing of the driver, ~0V to ~12V, which will be near to ideal for the gate on the power MOSFET.

The Arduino will interface nicely into the input of those UCC27424 driver chips you have bought.

If you want high side MOSFET, then you need to get P-Channel MOSFETs and also high side drivers. Is there any technical reason you need to have high side control, or is it simply a personal preference? Low side driver and N-Channel is more typically used.

Remember, the electrons flow from the negative potential to the more positive potential, so, what we often think as 'hot side' being the more positive end of the wire is the end where electrons flow into. Not that it really matters in this design. :D


Paul

Please have something capturing video when you fire this up the first time.

polymorph wrote:

when you fire this up the first time

was the word 'fire' used consciously I wonder, do you have no faith? :smiling_imp:

Actually, we should have a section in this forum for such things, we have the gallery where people proudly show off their projects. We can still learn a lot when projects go sideways and smoke is let loose.

I even learn from these threads where others have a different point of view or experience.

I think localbroadcast will make this work, he's reading up a lot and asking the right questions.


Paul