MOSFET for PWM LED Driver

Thread Starter

newaisa

Joined Dec 6, 2011
12
Hi,

I'm in the process of designing the power part for my PWM Led Driver.

PWM, MCU basically the digital part is done. I am trying to make around 16 Channels of high current PWM output.
Each channel will be driving around 100 pieces of 5V 0.15W single color LED that I got. I think 200Hz should be more than enough for dimming effect. So the mosfet output will be 200Hz/5V/4A(to be safe)

I'm very new to analog design and having a hard time deciding which MOSFET. My PWM is rated at 7V/20mA, of course I can change my PWM generator IC to suit the MOSFET's part if needed.
I have read many articles on how to choose a suitable MOSFET but still can't conclude anything yet. And also, do I need a MOSFET driver? I am confused here.
When I need a MOSFET driver and when I don't?

When a datasheet says a MOSFET can handle 100V. Does that mean it can handle 5V as well? Or I need to get a 5V Mosfet?
 

crutschow

Joined Mar 14, 2008
34,285
Whether you need a MOSFET driver is determined by the MOSFET gate capacitance and the desired MOSFET output risetime. Generally you want to risetime to be no more than a few microseconds to minimize power dissipation. The MOSFET risetime is roughly (V*Cg) / Idr where V is the desired gate voltage, Cg is the gate capacitance and Idr is the gate drive current.

With a 7V drive signal you want to buy a logic-level type MOSFET which can fully turn on with a 5V gate-source signal. A standard MOSFET typically takes 10V to fully turn on.

The MOSFET rating of 100V means it can handle any voltage up to a maximum of 100V.
 

wayneh

Joined Sep 9, 2010
17,496
At 200Hz, my hunch is that you don't need a driver because the incrementally more rapid on/off you gain with that will not be an issue for you at 200Hz.

I use IRF540N MOSFETS for general purpose stuff, mostly because I have a pile of them. It's current rating is few times greater than would be required for your project, but that's OK and I would look for a MOSFET with maybe ~4X whatever your realistic continuous current level is. Low Rds "on" is a good thing, as this is what will influence power dissipation in the MOSFET. Over-specifying your MOSFET adds very little to the cost, but you do pay for bigger/better specs.

As noted, you should look for a logic level fet if your voltage to the gate is not 10V or more. Operating any MOSFET at less-than-full-on, not fully saturated, is what causes heat. They can handle whopping currents if fully on, but not in linear mode.
 

Thread Starter

newaisa

Joined Dec 6, 2011
12
I have tried looking into the datasheet but can't seen to figure out which FET accept logic level as gate voltage and which will require 10V.

Can you give me some guideline on how they are represented in the datasheet. For example IRF540N as mentioned by wayneh. How do I figure out what is the gate voltage?

In the end I have decided to go with TLC5940 for PWM generator as it is easier with 16 channels of them in a single chip.
 

Thread Starter

newaisa

Joined Dec 6, 2011
12
Hi,

I am trying to power few strings of parallel LED by using TLC5940 as the PWM generator.

The chip is active low on the PWM side. Meaning it will be 30% low when the duty cycle is 30%. Now, I am thinking to use a inverting buffer to power the the MOSFET.

TTL LEVEL FROM TLC5940 > INVERTING BUFFER from TTL to MOS LEVEL > MOSFET > LED

The inverting buffer I have in mind right now is SN74LS06 or SN5406

will this method be good? Or do you have any better recommendation on this? Please suggest. I am very new in this.
 

wayneh

Joined Sep 9, 2010
17,496
Here's a figure taken from IR's datasheet for the IRF540n. Notice how the device begins to conduct when the gate voltage is ~3v and is not fully conducting until the gate is at ~8v or more. It is clearly NOT a logic level device.
 

Attachments

crutschow

Joined Mar 14, 2008
34,285
The data sheet will usually give an Rds(on) resistance value. The test value for Vgs used for this measurement is an indication of how much voltage is required to fully turn on the device.

If you look at the measurement for the IRF540n device, for example, Rds(on) is tested at a Vgs = 10V. Thus it is not a logic level device, as wayneh noted.
 

praondevou

Joined Jul 9, 2011
2,942
Why don't you drive the LEDs directly with TLC? Will you be drawing too much current?

In this case, yes you could do what you proposed with an n-channel MOSFET. Or you use a p-channel MOSFET and drive it directly with the TLC output.
In both cases you will need a pullup resistor (e.g. 10k) at the output of the TLC. (see the last picture of figure 1 of THIS datasheet)

If this will be working or not depends also on the PWM frequency. The gate capacitance together with the pull up resistor at either the output of the TLC or the SN5406 will change the slope of the PWM signal.

What is the PWM frequency?
 

Thread Starter

newaisa

Joined Dec 6, 2011
12
Yup the maximum of TLC is 180mA each channel which is too low to drive all my LEDs.

Do you mind elaborate more on the resistor usage on the PWM output? Don't see any resistor used on last picture of figure 1.

The frequency will be set to approximately 500Hz since it's just LED refresh rate. Do you think there would be any problem on the PWM curve after a buffer and a mosfet? I have some delay concern in this.

Either way I will be able to find out when I am done with code.
 

Thread Starter

newaisa

Joined Dec 6, 2011
12
Thanks for the explanation.

Since tlc5940 pwm is active low, I was thinking to use a hex inverter buffer in between out and mosfet gate.

What I have in mind is to use this as the inverting buffer since it has TTL to MOS level output.

Do you think this approach would be feasible? If yes, I still have to do more homework on capacitor/resistor I need to use in between PWM output and MOSFET.

Either way, work in progress. :)
 

praondevou

Joined Jul 9, 2011
2,942
Do you mind elaborate more on the resistor usage on the PWM output? Don't see any resistor used on last picture of figure 1.
The outputs sink current. Therefore if you connect directly the LED from the + rail to the output you won't need a resistor. However, if you connect a MOSFET you will need to make sure to have definite voltage levels at the output pin. Since there is no more LED through which the current flows, you need a pullup resistor. A MOSFET gate is isolated, no continuous current will flow through it.

The frequency will be set to approximately 500Hz since it's just LED refresh rate. Do you think there would be any problem on the PWM curve after a buffer and a mosfet? I have some delay concern in this.

Either way I will be able to find out when I am done with code.
There won't be a problem with 500Hz.
 

Wendy

Joined Mar 24, 2008
23,415
Please do not create a second thread for the same project. One thread is enough.

I have merged the two threads.
 

praondevou

Joined Jul 9, 2011
2,942
The TLC supports LED power supply voltages up to 17V.

That means you can also use a simple CD4049 buffer IC. (if you'd choose the 5406 which has an open collector output it's output needed also a pull-up resistor.)

You could also use a p-channel MOSFET. The 10k resistor at the gate of the p-channel MOSFET could actually be too big, it would slow down the gate discharge. For 200Hz it may work though. If needed it's value could be decreased too.

There is no need for logic level MOSFETs if you choose 15V as the LED power supply.

As stated before, a very common type is the IRF 540 (n-channel) or IRF9540 (p-channel).

 

Attachments

Thread Starter

newaisa

Joined Dec 6, 2011
12
The diagram you posted will give LED 15V right?

The parallel led string I have purchased currently is rated 5V. So I am thinking whether this will work or not?

 

SgtWookie

Joined Jul 17, 2007
22,230
What are you going to be using for a power supply? What voltage?

Several things to pay attention to when selecting a MOSFET:

N-ch or P-ch: It is preferable to use N-channel MOSFETs wherever possible, as otherwise equivalent P-ch will have about 2.5 times the gate charge requirement of the N-ch MOSFET. That is a result of the physics behind the technology; basically it is easier for electrons to flow through the semiconductor than it is for holes to flow.

Vdss: This is the maximum voltage allowed between the drain and source terminals. Select a Vdss rating that is at least 20% more than your supply voltage, and if possible not more than 1.5 times what is required. All other things being equal, a higher Vdss rating increases the Qg (total gate charge) very significantly.

Id: This is the maximum drain current rating. Increasing Id means that Qg (total gate charge) also increases, but not as severely as with increasing Vdss. It's a good idea to select a MOSFET that is rated for several times the current required.

Rds(on) - this is where the Vgs is specified for the rated drain current. If the MOSFET is a logic level MOSFET, you will see the Rds(on) rated at Vgs=4.5v or 5v in addition to 10v. The lower the Rds(on) is, the better - but that increases your Qg.

Qg: This is the total gate charge, as mentioned several times already. The larger that Qg is, the more capable your gate driver will need to be to switch it on and off. If you are using only a couple hundred Hz for a PWM signal, you can probably do without a driver if your Qg is low enough; say 15nC or lower.

For example, take a look at the IRLU8721PBF.
It has a Vdss of 30v, Id=65A @ 10v, Rds(on)=10.1m Ohms @ 4.5v, Qg=13nC @ 4.5vm, and is a compact TO-251 package that is a space saver on a PCB. Since it has such low ON resistance, power dissipation with 4A current will be just 101mW.
http://www.irf.com/product-info/datasheets/data/irlr8721pbf.pdf
If you are supplying your LEDs with 24v or less, this would be a good choice. These MOSFETs are about $1.25/ea

Tip: International Rectifier has a naming convention that can help you find what you are looking for more quickly.
Parts beginning with "IRF" are generally standard level MOSFETs.
Parts beginning with "IRL" are generally logic level MOSFETs.
This is not true in 100% of the cases, but it is for the vast majority.

[eta]
Wow, the scenario changes rapidly when threads are merged!

In this case, P-ch MOSFETs could be warranted due to the control IC in use.

However, the information on MOSFETs that I posted is still generally useful.
 

Thread Starter

newaisa

Joined Dec 6, 2011
12
Hi SgtWookie,

thanks for the reply. Long reply, but I think I understood them generally.:D

So after looking through the datasheet of IRF9540 as suggested I think it is quite suitable in my application.
The Drain to Source voltage will be 5V as my LED are rated 5V 30mA each.

Each of these MOSFET will sink a maximum of 6A in my application. Since IRF9540 can handle 16A @ 100C @ 10V, I think in my case it should work fine.

Will be using LOTS of these LED. The LED count is around 2.3k. Yeah lots of LED. Currently they are powered by 2 x 200W 5V power supply.
I am trying to add some effect onto them with this design.

Now back to my previous question. Do you think my idea of driving my PMOS this way will work?



Please let me know if there is any "dumbness" in my PMOS operation design. I appreciate any help as I progress through my project :)
 

Wendy

Joined Mar 24, 2008
23,415
You need a resistor per LED on that design. I have read people who thought PWM was equivalent, it isn't. Without current limiters like resistors, the MOSFET and LEDs are at risk (as in likely to die a quick and smokey death).
 

praondevou

Joined Jul 9, 2011
2,942
The LEDs are rated for 5V, does that mean they have built-in resistors? Just to make sure.
I didn't draw it in the other post because I was assuming they have them...

If they don't you will need them.

6A for this MOSFET will be ok. Put them on a heatsink though.
 
Top