Basic mosfet switch question.

[JimmyM]

Question:

I understand that biasing the gate + turns the mosfet on. the 4.7 resistor to knock the voltage down makes sense because the gate can only take so much voltage, but what's the 10k for?

In the drawing the OP posted, it's to pull the gate down when the switch is opened. The 4.7K acts with the 10K to form a voltage divider, to lower the gate voltage. When the switch is opened, the gate charge dumps through the 10K.

 

[JimmyM]

Not necessarily. The short answer is that there is no real point reducing the voltage applied to the gate unless it is too close to the maximum permitted. So for instance, if we look at the data sheet for the IRF540N we see under "absolute maximum ratings" that the maximum gate-to-source voltage is +/-20. So unless there is a danger of approaching or exceeding 20 V (for this specific MOSFET) there is no need to reduce the gate voltage to a lower value than the battery voltage.

Exactly. You would omit the 4.7K resistor, and use just the 10K.

 

[Oznog]

There are plenty of teeny tiny MOSFETs that can switch many amps due to their spectacularly low rds-on. The lower rds-on means less heat and thus the smaller package.

You need to look at rds-on vs Vgs (battery voltage if no divider is used) and Id vs Vgs. See at a particular Vgs, it allows a specific gate current to flow. If the load doesn't draw that much, then it acts like a closed switch except for a small rds-on. Higher Vgs and rds-on goes down slightly.

But, say your Vgs=6v when the batts are low, and the spec sheet says Id=1.5A for Vgs=6v but your driver- a switching driver- is supposed to draw current surges up to 2.5A due to its PWM circuit. Well, this is bad news because the MOSFET will instantly raise its resistance when current goes above 1.5A to limit the current to only 1.5A. Not only does that prevent the light from operating properly but will probably incinerate the MOSFET.

Note that a lot of garden-variety MOSFETs require >8v or even >10v to turn on fully. There are also ones which turn on fully at 5v, 3.3v, or even 2.5v. Also note that there are limits on Vgs, maybe 20v for the higher voltage gates but the low voltage gates may draw the line at like 8v.

Gates are very sensitive to voltage spikes. If you were to knock the case so the battery bounced off the contact for a moment, the inductive spike from the driver could create a positive or negative voltage that would destroy the gate. A zener diode on the gate (or some MOSFETs have internally protected gates) will keep it 100% protected.

 

[JimmyM]

Without having to do a lot of research to find the FET you want...
I've found a few FETs that work very well for flashlight purposes.

For Battery voltage 6V<VBATT< 30V and Amps <10, the IRLR7843.
For Battery voltage 10V<VBATT<24 and amps <25-ish the IRF1324S.
For Battery voltage 10V<VBATT<40 and amps <20 the IRF2804.
Generally though, the IRF2804 is best if you have the room and it has a great voltage range. If you want absolutely the lowest Rds(on) (0.6 mOhm) and will never exceed 24V then go for the IRF1324.
For smaller installs the IRLR7843 or IRLU7843 will do 8-9 amps with no problems. This is the better choice for lower voltage lights like the 5761, 64275, etc. Anything from 5V-10V. The 7843 has better gate performance at voltages under 10V than the others.

 

[Greg G]

Thanks for sharing that Jimmy. I'm saving that list for future use. I unfortunately don't have a lot of time to R&D everything I'd like to, and posts like this really help.

 

[Greg G]

John, I received the IRL540.

Thank you very much for helping me out with this, and all the information you've posted in this thread!

 

[Illum]

Does Radio Shack carry the IRF540N? It's a very common MOSFET and will only need to dissipate about 40mW when passing 1 amp. No heat sink needed.

IIRC last time I checked they only have the IRF510N:candle:

 

[JohnMuchow]

John, I received the IRL540.

Thank you very much for helping me out with this, and all the information you've posted in this thread!

Excellent! Glad I could help out. :)

John

 

[The Dane]

Took the original diagram and modified it a little.
I needed something for 20V max but with a softstart to minimize bulb cold-blow.

The 4.7kOhm is changed to 2.2kOhm
The 10kOhm is changed to 4.7kOhm
A 10microF capacitor is paralleled with the 4.7kOhm resistor

This means that the gate will start at 0V and power the fet up in @ 0.5 sec.
This will eleminate the tendency og blowing (way) overdriven bulbs at cold start.

To further beef the circuit up i used a IRFS3206 FET
840A pulse, 120A continous and Rds is 2.4milliOhm.

 

[JimmyM]

This approach has been discussed in the past. Using a cap to slow down gate rise. It will overheat the FET. 0.5 seconds is a long time to dissipate all that power. 0.1 seconds is a long time for this approach to work. That's why I went with PWM for the design of the JM-SST and the new voltage regulators.
It may be turning on in far less than 0.5 seconds though. The FET is FULLY on at Vgs = 10V. It may take 0.5 seconds for the gate voltage to rise close to the voltage divider max, but far less to get to 10 V.

Took the original diagram and modified it a little.
I needed something for 20V max but with a softstart to minimize bulb cold-blow.

The 4.7kOhm is changed to 2.2kOhm
The 10kOhm is changed to 4.7kOhm
A 10microF capacitor is paralleled with the 4.7kOhm resistor

This means that the gate will start at 0V and power the fet up in @ 0.5 sec.
This will eleminate the tendency og blowing (way) overdriven bulbs at cold start.

To further beef the circuit up i used a IRFS3206 FET
840A pulse, 120A continous and Rds is 2.4milliOhm.

 

[wquiles]

This approach has been discussed in the past. Using a cap to slow down gate rise. It will overheat the FET. 0.5 seconds is a long time to dissipate all that power. 0.1 seconds is a long time for this approach to work. That's why I went with PWM for the design of the JM-SST and the new voltage regulators.
It may be turning on in far less than 0.5 seconds though. The FET is FULLY on at Vgs = 10V. It may take 0.5 seconds for the gate voltage to rise close to the voltage divider max, but far less to get to 10 V.

+1

In fact I did pretty much the same circuit and ran bench testing on it back in 2005:
bench testing of incandecent soft start ckt ...


Like Jimmy states, this design does not work that well (it is more of a delayed start, than a true soft start), but you can always do like I did and go ahead and built it and test it. It is the best way to learn about these circuits ;)

Will

 

[JimmyM]

In that thread you were testing an MN61 bulb. Not a 64623 (or similar) monster. A 0.5 second start with a 100W bulb over driven will overheat the FET in short order. And you NEED to figure out a way to shut it off quickly. You can pretty much forget flashing this thing on and off.

What battery/bulb combo are you going to use?

+1

In fact I did pretty much the same circuit and ran bench testing on it back in 2005:
bench testing of incandecent soft start ckt ...


Like Jimmy states, this design does not work that well (it is more of a delayed start, than a true soft start), but you can always do like I did and go ahead and built it and test it. It is the best way to learn about these circuits ;)

Will

 

[JimmyM]

To further beef the circuit up i used a IRFS3206 FET
840A pulse, 120A continous and Rds is 2.4milliOhm.

The IRF2804 and IRF1324S-7P are both better for <20V operation.
The 1324 has 0.8 mOhms Rds(on) and a Pulse current rating of 1640A.

 

[The Dane]

Me oh my, i must have a Midas touch:D
I just flicked this one on/off 50 times is as many seconds, with my thumb on the FET. No heat at all!
12V 50W bulb overdriven to 20.8V 5.5A (all my bench PSU can muster!)

So for my purposes and uses it works flawlessly :poke:

 

[The Dane]

Can some of you knowledgeable guys please try to help me here.

I have put the above mentioned circuit together and tested it til i almost got a blister on my thumb. Not from FET heat but from flicking the contact on and off.

I am simply not able to get any heat build up in the FET, so:
What am i doing wrong?
Several comments mentioned FET meltdown when using my soft start approach, and it's simply not happening, what gives?

Bulb 64623 overdriven to @175W and thats it.

 

[wquiles]

Can some of you knowledgeable guys please try to help me here.

I have put the above mentioned circuit together and tested it til i almost got a blister on my thumb. Not from FET heat but from flicking the contact on and off.

I am simply not able to get any heat build up in the FET, so:
What am i doing wrong?
Several comments mentioned FET meltdown when using my soft start approach, and it's simply not happening, what gives?

Bulb 64623 overdriven to @175W and thats it.

Do you have some data (like a scope trace) that you can share with us showing what the circuit is actually doing?

 

[paetzi]

Hi, i want to direct drive a sst 90 in a 3d mag.
What kind of mosfet is the best for this setting and what resistor do i need?
Hope you´ll help me.

 

[JimmyM]

What's an sst 90? I'm assuming it's an LED. But what are it's specs. Voltage, amperage.
Thanks.

Hi, i want to direct drive a sst 90 in a 3d mag.
What kind of mosfet is the best for this setting and what resistor do i need?
Hope you´ll help me.

 

[jtr1962]

Which mosfet would work for this application? Are the IRF510's from Radio Shack up to this?

Lots of MOSFETs available through electronic supply houses which are way better than the IRF510. For example, I just received these for a project. On resistance is 0.014 ohms with Vgs of 10V, or 0.021 ohm with Vgs of 4.5V. They use a TO-252 SMD case (smaller than the IRF510 TO-220). You can solder the tab to a copper or brass plate for heatsinking although truth is you won't need it if your current is less than about about 5 amps. Best of all, they only cost me $0.189 in 250s. I can send you one or two to play with if you want.

 

[paetzi]

Sorry. The sst- 90 is an led.
It runs with 3,7 Volt and up to 9A.
So around 33W.