Hi Speed LED Driver Question

[boldinventions]

Hi All,
I'm interesting in experimenting with sending data via IR LED's. I've picked out **broken link removed** from Panasonic.

If possible I would like to drive it with a digital signal coming from an FPGA at approximately 10 MHz, with an average duty cycle near 50%.

Since it looks like the max rating on the forward current is 100mA, I would like drive it a near that or higher if I could.

However, the MOSFETS I was looking at are pretty slow. Digikey has over 10,000 parts, and you can't search by speed. Does anybody have any suggestions as to how to drive this LED and what parts to use?

Thanks in advance,
-Kevin

 

[crutschow]

A small MOSFET, such as the 2N7000, has a rise and fall time of 10ns, which should be sufficient at 10MHz. But you would need to drive it with a high speed, low output impedance MOSFET driver circuit to charge and discharge the high gate capacitance of the MOSFET.

Detecting a 10MHz IR signal is also not a trivial task.

 

[boldinventions]

Thanks, Carl! You got me pointed in the right direction.

I searched a bit more and found the Fairchild FDV301N. I'm hoping I can get away without a MOSFET driver with this one. Do you think I'd be able to switch 100 mA to my LED on and off at 10Mhz with a 3.3v FPGA output?

Detecting a 10MHz IR signal is also not a trivial task.

My plan was to use an OSRAM SFH 2400-Z photodiode. It claims a switching time of 5ns. From there to a simple high-pass filter, then to a comparator, and into an FPGA input. I've never used a photodiode before, but they seemed much faster than phototransistors. I'm mostly worried about how small the signal will be and noise.

-Kevin

 

[blueroomelectronics]

Is 10MHz the modulation or data rate?

 

[boldinventions]

Is 10MHz the modulation or data rate?

10MHz would be the carrier, the data rate would be much lower.

-Kevin

 

[nike6]

the FDV301 and FDV303 are excellent parts, highly recommend.

i am not so sure about the max. speed, I've built small oscillators with them, using 4MHz crystal, and also 10MHz ceramic resonator.

so they can handle signals within MHz range.

the turn-on point is 2.5 volts or so, you'd have to study the datasheet, I have the one for FDV303, but I am not so good interpreting them.

as for n-channel you'd have to line in the LED from Vcc towards the MOSFET, and connect the drain to the ground (correct me if I am wrong, i have made some circuits using this MOSFET, just memory).

 

[crutschow]

I searched a bit more and found the Fairchild FDV301N. I'm hoping I can get away without a MOSFET driver with this one. Do you think I'd be able to switch 100 mA to my LED on and off at 10Mhz with a 3.3v FPGA output?

That transistor has 0.7nC of gate charge that has to be switched. To transfer that amount of charge in 10ns requires 70mA of drive current. I doubt that a FPGA output can provide that much current so you will need to add a high current digital buffer circuit.

My plan was to use an OSRAM SFH 2400-Z photodiode. It claims a switching time of 5ns. From there to a simple high-pass filter, then to a comparator, and into an FPGA input. I've never used a photodiode before, but they seemed much faster than phototransistors. I'm mostly worried about how small the signal will be and noise.

Depends a lot on how far the signal has to go. Chances are you will have to add a high frequency amp to the detector (see Electronic Design Welcome for an example).

 

[boldinventions]

That transistor has 0.7nC of gate charge that has to be switched. To transfer that amount of charge in 10ns requires 70mA of drive current.

Hmm. So there doesn't seem much point in using the MOSFET if I need 70mA to drive it so I can sink 100mA to light up my LED. I guess I can understand that the 100mA could be continous, were the 70mA would only be needed for for the 10ns during switching.

I looked at a few MOSFET driver chips, but I didn't find a fast one that could run on a 3.3v supply, which probably makes sense given the properties of typical MOSFETs.

So I'm thinking now of just using a bunch of digital buffer gates hooked up parallel, each with their own current limiting resistor on it. The 74AC125 apparently can be run at 10 MHz, and by using all four gate I should be able to get my 100 mA. And I can power it with 3.3v. The data sheet does not specify the rise and fall times, just the propagation delay.

Depends a lot on how far the signal has to go. Chances are you will have to add a high frequency amp to the detector (see Electronic Design Welcome for an example).

That was an excellent link! Thanks very much. The part they use in that circuit is available from Digikey, and is in a reasonable package too. The only downside is that I can't run it with a +3.3v supply. So I looked around on the web, and there are lots of articles on photodiode amplifiers, although the other examples aren't designed for 10MHz. Then I saw there are several entire books on photodiode amplifiers. (sigh) What I'm really interested in is doing the FPGA code, not the analog electronics. Oh well.

Thanks again for your help, I think you've improved my chances for success dramatically.

-Kevin