Logic level FET

P

Paix

Senior Member
I am making a small incubator using a 12V 20W halogen lamp as the element, controlled by a Picaxe 08M.

I have a logic level N channel FET type STP36NF06L 60V 30A, which is a bit of overkill, but a generalised solution that has a little headroom to spare for slightly larger applications.

I am not FET aware but have put a 4k7 ohm resistor in the gate lead and observed that it appears to work just fine. I chose the FET after reading the thread that discussed FET heating when not fully conducting. This device seems simplicity itself, but feel that being voltage driven, that the resistor could easily be 47k ohm or more, but what are the merits of the existing or much greater value resistor?

My question, is there a more appropriate value that I could or should be using when keying/switching the gate via the resistor from an 08M output port?

FET datasheet here Rapid Electronics part no 47-0552. The FET is on the low side of the lamp, which is supplied directly from 12V nominal (13.8V) and the Picaxe 08M powered by a 7805 derived 5V supply. Temperature sensing using readtemp with a DS18B20 chip.
 
Last edited:
J

Jaguarjoe

Senior Member
Without a resistor from the gate to ground to discharge the gate capacitance, that capacitance will take the next best route which would be through your resistor and the Picaxe when its output is off. The larger the value of that resistor the longer it will take to discharge the capacitance which means the longer it spends between full on and full off. Typically, the resistor in series with the digital output and the MOSFET is only 100 ohms or so to squelch ringing caused by the gate capacitance and the inductance of the wire going to the gate. A 4k7 or so from gate to ground will insure the gate goes to zero quickly.
 
B

boriz

Senior Member
"feel that being voltage driven, that the resistor could easily be 47k ohm or more"

Absolutely correct.

The Gate resistor is mainly there only to protect the Picaxe output from excessive current should the MOSFET fail in such a way that Drain current get's routed to the Gate. It's only a precaution. If you are sure the MOSFET will never fail in this mode, then no resistor is required. Though it is good practice and you can never really be sure.

I'm lazy and sometimes use no Gate resistor at all during experiment/prototyping. Then for the finished product, put a resistor in that I know will limit the current to the Picaxe output to 20mA or less should the full Drain voltage be applied, (20mA is generally considered to be the maximum safe current a Picaxe output can sustain.).

EG: Using Ohms law, R=V/I. For a 5v Drain (R=5/0.02), you need at least 250R. And for a 12v Drain (R=12/0.02), at least 600R. In both cases I'd prolly use anything between 1K and 10K, whatever is handy.

I'm assuming this is a simple ON/OFF system where switching speed is not an issue.

If you are using PWM and the MOSFET will be switching many times per second, then the situation is a little different. And a bit more complicated.
 
B

BeanieBots

Moderator
Yep, use a FET driver if you want to PWM a FET.
If you only want piddly little currents and don't mind the FET getting warm then use a resistor based on the calculations shown by Boriz. (which is fine for static switching).

It all depends on your gate capacitance, PWM frequency and whatever current/voltage your motor runs at. (how long is a piece of string?).
 
P

Paix

Senior Member
Thank you Jaguarjoe, Boriz et al. 1k gate to Picaxe output and 4k7 gate to ground close to the FET. Both personally preferred values that I have a stack of.

Switching is simple on/off, the frequency of which is likely to in the tens of seconds, but feel that the advice given is pretty universal.

I'm sure that Dippy's book will become a very popular tomb in everyone's reference library. When is it expected to be published?

Inglewoodpete, please note that the particular FET is described as a logic level FET. Designed to obviate the necessity for FET drivers.

Simple is good :)
 
I

inglewoodpete

Senior Member
Paix said:
Inglewoodpete, please note that the particular FET is described as a logic level FET. Designed to obviate the necessity for FET drivers.
Click to expand...
I think you're confusing a level shifter for driving a non-logic FET from logic levels.

I was answering the question about PWM with a MOSFET (logic level or not) - they require a MOSFET driver to get very fast rise and fall times at the FET's gate. Slow rise and fall times on a gate with (relatively) large gate capacitance become a problem with PWM. Slow transition times result in extended periods where the MOSFET is partially on. If switching a large current with PWM, the MOSFET can easily overheat if it doesn't have a low impedance driver..
 
J

Jaguarjoe

Senior Member
I never made the comparison, but gate capacitance for a logic level part is probably worse than that of a "normal" part because the oxide layer between the gate and the rest of the MOSFET is so much thinner.

Gate drivers can provide more than an amp of gate current in and out of the MOSFET just to deal with charging and discharging the gate capacitance.
 
B

boriz

Senior Member
Paix. Your 4k7 Gate-to-Ground resistor is redundant in this application. It's actually worse than that. It forms a potential divider with the Gate-to-Picaxe resistor, reducing the voltage at the Gate from 5v to just over 4v.

Remember the Picaxe output can Source AND Sink current, so the Gate is driven low just as hard as it's driven high. For a simple switching app like this, a much higher value Gate-to-Ground resistor will be better. Say 100k or more. You could probably omit it all together, but that can potentially result in the MOSFET passing Drain-Source current very briefly when the Picaxe is first switched on with it's output in an indeterminate state.
 
premelec

premelec

Senior Member
Slow virtues...

In regard to switching a tungsten lamp on and off a slow rise can be a virtue to lessen filament shock [they draw about 8x cold current as the positive temperature coefficient comes into play] - so adding the current slowly may cause [in this case] a little more FET heating it will also reduce the spike of current to the power supply and lamp. If you want continuous dimming of the lamp the situation is a bit different... slow on fast off could be best though the fast off could give some power supply spikes also... if you put the FET in the same environment you are trying to heat with the lamp then it's heating will not be wasted...
 
D

Dippy

Moderator
A "popular tomb" ?
Bl**dy hell, that would be big book ;)
Maybe a tome.
Or maybe it would be the death of me....


Boriz makes a good point about the position and value of your resistors.
I would slightly disagree about the non-neccessity of a pull-down (for N chans).

True, in a perfect world the PICaxe o/p would be in 'output' mode and in full control.
But, belt and braces dictates that during startup (as he mentioned) and also in a fail state means that the pull-down could stop your MOSFET dithering. Not 100% foolproof but better.

Your choice of values is, as Boriz says, not ideal.
Either, put the 4K7 pull-down on the non-Gate side of your series Gate resistor.
Or increase the 4K7 to 100K for example and reduce the pot-div effect.

For safety and in prototyping I would include the gate series resistor.
You'd have to be pretty confident to remove it.


I've attached a very rough starter guide for MOSFET switching.
It's intended just to get the old grey-matter ticking.
Start: Routine Excuses;
It is merely meant to be a kick-off point nothing comprehensive.
It was done in a hurry and is probably strewn with errors.
If it is then send a postcard to anyone except me :)
End;
 

Attachments

Last edited:
A

Andrew Cowan

Senior Member
Dippy - that is a fantastic guide, and it really should be put somewhere more prominant than this. At least in one of the 'finished projects' sections (I know this forum isn't big on stickies...).

Well done again :D

Although I'm not sure what 'the BLUE region is meant to be the transistion voltage for the GATE VOLTAGE' is talking about - there's no blue :p
 
D

Dippy

Moderator
Thanks.
The blue bits are those scruffy little blue rectangles on the gate voltage curves.
Aren't they visible? They are there on Adobe Reader. Maybe something odd has happened? C'est la vie.
Actually I've just seen an error..... please listen to the hold-music while I change it....
 
Last edited:
A

Andrew Cowan

Senior Member
I can only see a very faint blue if I look at my screen from about 10 degrees angle - otherwise it just looks white. Could you make the blue a bit darker?

And I'm using Adobe Reader :S...
 
D

Dippy

Moderator
I've reduced the transparency on the blue blobs and changed some things.

Sadly, I've run out of time.... must mow the lawn :(

Attached MK2.
 

Attachments

P

pete20r2

Senior Member
Most of the PWM questions I see end in, "you need a MOSFET driver", can someone more confident/knowledgeable then me post a simple DIY MOSFET driver?
 
I

inglewoodpete

Senior Member
pete20r2 said:
Most of the PWM questions I see end in, "you need a MOSFET driver", can someone more confident/knowledgeable then me post a simple DIY MOSFET driver?
Click to expand...
If you get Silicon Chip magazine, you will see some discreet-component FET drivers in some of their designs. Not alway the best design for PWM, though.
 
P

pete20r2

Senior Member
Unfortunately I do not get that mag, what I am mainly concerned with is the cost involved in a FET driver, are there other options?
 
D

Dippy

Moderator
Certainly looking at discrete circuits give a valuable insight into gate driving and the principles..
But in the real world, in common circuits, I can't see any advantage in avoiding the multitude of driver chips, some of which have been highlighted by Andrew.

Go to the Farnell website and search on MOSFET Driver IGBT driver and you find a shed load with prices.
READ the Data Sheets to see what other bits are need . Make a BOM , add it up and , bingo, you have a cost.

There are also a multitude of Application Notes by various manufacturers.

When you add up the Data Sheets and App Notes there must be dozens of example circuits shown - AND , importantly, advice on component choice and circuit layout. Ignore this at your peril!

Andrew has given you useful link to Microchip's site which has a wealth of information and examples.

You'll find all sorts of drivers and examples; High-side P , Low-Side N , High-side N, Ideal Diode....
If I were you I'd spend an hour hunting and bookmark/save a pile of Data Sheets into your Grimoire.

In some Apps you may choose one of those Smartswitch thingies (MOSFET and driver/protection in a single device) - these can save a lot od hassle and board-space.

There is so much out there - finding it is the biggest problem.
One example of a useful tiddler is the Diodes/Zetex ZXMS6006 , but there are hundreds of similar things bigger and smaller. Just be aware that they tend to respond slower if you want fast PWM.
 
P

Paix

Senior Member
Danger wordslip . . .

Sorry Dippy, it was probably a bit late when I wrote that or else I was having one of those slightly senior moments.

Off to pore over your Rough Guide to MOSFETS and pour a cup of tea.

Sorry about the tomb/tome slip; looking forward to seeing your TOME. :)
 
D

Dippy

Moderator
Piax: haha, I like the puns. I only ever cringe when I see the lose/loose confusion. The trouble is that if you stare at words for too long you can become dickslexic.

Manie , sorry. Next time eh? By the way, I'm designing a dyno at work. Oh, hang on, you were desiging a thing to measure drive-through toe-in weren't you.
 
C

curry87

Senior Member
Sorry for the highjack.

How do you know which fet driver to choose to use with a fet if fet drivers are rated with max output current in mind how does this relate to a 30amp fet for example ?
 
D

Dippy

Moderator
First stab: Gate voltage/capacitance/charge a, gate series resiatance nd , if applicable PWm frequency.
Thus you can have a first guestimate of gate drive peak and average currents.
And, of course, your App's voltage.
 
J

Jaguarjoe

Senior Member
curry87 said:
Sorry for the highjack.

How do you know which fet driver to choose to use with a fet if fet drivers are rated with max output current in mind how does this relate to a 30amp fet for example ?
Click to expand...
Q = I*T or I = Q/T

Q is the gate charge from the datasheet in coulombs.
I is current in amps
T is time in seconds

You determine how quick you want to switch, the MOSFET vendor determined the gate charge, and the resultant current determines one of the parameters for the driver.
 
You must log in or register to reply here.
Top