Very fast (100+ MHz) LED or laser diode switching

[MatthiasH]

Hi,

I am looking for a way to switch a strong light source (0.5W LED or 200 mW LD, but laser is probably faster) with a rectangle function at up to 150 MHz. There are a bunch of LED and laser diode driver ICs out there that claim to do up to 266 or 300 MHz (I'm using the HG-1D from iC-Haus) , but I haven't quite succeeded yet: whether LED or LD, above 50MHz modulation depth goes down and they just don't seem to switch off any more (inductance?). Does anyone in here experience at these very high modulation speeds? Somehow they must be able to do it in DVD writers and optical communication....

Thanks,
Matthias

[VDX]

... for such high frequencies you'll need special diodes, optimized for pulsing, not for CW operation.

AFAIK most high power light sources are limited to some ten MHz max. - maybe modulating the beam with an AOM-crystal (AcustoOpticalModulator) will allow higher frequencies?

Viktor

[mixedgas]

Normal LDs have a sort of resonance function at 500 Mhz to 1 Ghz that is the upper limit. The higher power the LD, the lower the problem occurs. 50 Mhz is not that low. It may be how you are biasing the LD to conduction.

In many cases they are just fine for pulsing if average energy is well below their CW limit.

What duty cycle or pulse width do you need for your pulses? At what specific rate?

The usual lab scheme is to apply a DC bias to just below threshold, and then couple the drive pulse in from a pulse generator or RF oscillator using a low cost device called a "Bias Tee".

I need more specific data to help you.

For very short pulses, special LDs are used, and they actually are available off the shelf at near IR wavelengths.

Ic_Haus parts are very nice, but at 1 watt you need something a little more specialized and capable.

Take a look at www.avtechpulse.com/ for commercial pulsers. You can get a idea of what is possible from their web site, and some free advice.

A simple method for a ~50% duty cycle is DC bias plus a RF generator, this is not expensive to home build.


Steve

[MatthiasH]

Hi Steve,

I already have a RF generator that, according to our lab's 500MHz 'scope, produces a fairly clean rectangle signal from 10 - 120 MHz at 50% duty cycle (which covers the range we need). Thermal dissipation in LD or driver shouldn't be a big deal, since we only need very short trains of pulses (5-10 ms of signal every second or so). Doesn't matter if the first and last 1-2 ms are not overly clean, as long as we get some nice rectangle somewhere in between.

Maybe the DC bias technique won't be all that hard to implement using the iC-HG driver chip at hand, with its 6 channels that could be bridged in 3 pairs (as foreseen on the evaluation board). All it should take is the unfiltered signal on channel 1 providing the RF modulation, and a lowpass filtered copy on channel 2 providing an envelope for the DC bias.

[mixedgas]

No, Get a bias T. Run the IC Haus as the constant current and add in the RF from a dedicated sine wave RF amp.

You need the low impedance RF source with sine wave. The Ic-Haus chips are going to clip the drive. I'm not at liberty to discuss, but I have some time working on Ic_Haus based projects.

Great chip for small diodes.

I can help with this, I do RF as part of making my living. I'm a professional university technician. Really the term is "Research Associate". I'm well above the technician rating.

Thorlabs sells bias Ts.

If your at a university, search the journal Rev. Sci. Inst for some insperation.

Kind Regards,

Steve Roberts

[VDX]

... hmm, I'm using high power IR-LD's with threshold currents of 400 to 600mA and max. currents of 5 or 9 Amps, where I would be happy to recive PWM frequenzies of more than 50 MHz.

But then probably the power source is the main problem, not the LD's, correct?

Viktor

[chad]

Just a note. Those IC-Haus HG Aren't rated very high for cw operation so watch your power dissipation. They are easy to bake in cw.

chad

[MatthiasH]

Those IC-Haus HG Aren't rated very high for cw operation so watch your power dissipation. They are easy to bake in cw.

Shouldn't be a big deal at the 0.5% (or so) duty cycle we're shooting for. As a prolific baker, I've learned to be extra careful during the adjustment phase.

[chad]

Just watch the dc bias

Let us know how it goes.

chad

[-bart-]

Are you using the ichaus evaluation board, or did you design your own application?
Needless to say that these frequencies require pretty strict HF design topology.
How do you determine this modulation dept ? Optically or voltage at the diode?
How fast is your photodiode ? Do you use ttl or LVDS ?

First convince yourself that your LVDS/ttl input signal remains full integrity all the way upto the driver-pins.
Then drive an ohmic dummy load, and make sure sure you achieve full bandwidth.
Then use led or ld
Only resort to dc bias T if you have fully validated your driver.