LED Switching

[col_implant]

Hi,

I'm about to begin a project on optical detection. We are stimulating a sample with light and examining the output. The phase between output and input is most important property and we need to measure it down to 0.1 degree.

For our light input we will use an LED. Since the phase is so important we need a circuit that can switch the led in a perfect square wave, upto about 100KHz. So need to turn the LED from fully off to fully on in a minimum of time.

Can anyone suggest a circuit to acomplish this? Most conventional LED circuits have quite a long transient so are unacceptable. Also is there a particular type of LED which should be used?

Cheers

C

 

[Boomslang]

Would the PWM output from a PIC or other mcu not be square enough?

 

[ericgibbs]

Hi,

I'm about to begin a project on optical detection. We are stimulating a sample with light and examining the output. The phase between output and input is most important property and we need to measure it down to 0.1 degree.

For our light input we will use an LED. Since the phase is so important we need a circuit that can switch the led in a perfect square wave, upto about 100KHz. So need to turn the LED from fully off to fully on in a minimum of time.

Can anyone suggest a circuit to acomplish this? Most conventional LED circuits have quite a long transient so are unacceptable. Also is there a particular type of LED which should be used?

Cheers

C

hi,
As the light spectrum from a standard LED is not coherent I would think its impossible to measure a 0.1% phase difference.

You need a coherent light source, ie; a laser emitter.

 

[Roff]

Col_implant:
0.1 degree of what? Your modulation waveform?

 

[col_implant]

Would the PWM output from a PIC or other mcu not be square enough?

- Its not just the electric waveform itself but more so the actual switching of the LED which has an undesirable transient

hi,
As the light spectrum from a standard LED is not coherent I would think its impossible to measure a 0.1% phase difference.

You need a coherent light source, ie; a laser emitter.

Col_implant:
0.1 degree of what? Your modulation waveform?

- The phase measurement well be the difference in the waveform input into the LED and that measured at a photodiode. Not the difference in the phase of the light waves

 

[ste5442]

Won't the phase difference simply be related to how far away the emitter is from the detector...or is that your point?
With a few more details we may be able to help.

There is no such thing as a perfect square wave and the LED will not instantaneously turn on and off.

I think you must be measuring the distance between the emitter and 'sample' since that is the only thing that would alter the phase?
Actually, the 'sample' would need to be moving pretty fast to make any measurable change in the phase of the light reflected!

*More details required ;-)

 

[ericgibbs]

- The phase measurement well be the difference in the waveform input into the LED and that measured at a photodiode. Not the difference in the phase of the light waves

hi,
As 'ste' says you measuring the response delay of the emitter/detector and transit time of the pulse.

What sort of values are you expecting to see and what will they represent.?

 

[ste5442]

It sounds like an interesting project but I think you may need to describe it again without using the word 'phase'

 

[col_implant]

Ok, i will try again.

We plan to investigate the sample by stimulating it with high intensity light from an LED. We expect only a fraction of this light to make it through the sample, and we expect there to be a very slight delay in the transmitted signal related mostly to distance but also to the refractive properties of the sample.

I understand that a perfect square wave does not exist, and if it did the harmonics probably be a big problem. My goal mainly is to have an LED swith on and off symmetrically, and with a minimum rise/fall time. I dont think there is a huge difficulty with the switch on, but it is my understanding that when the 'logic high' is removed from the LED there is a slight delay in the extinguishing of the light emitted. The junction "slowly" loses its bias and the light emitted falls off, due to the discharge of the junction capacitance.

I was thinking if there may be some nice circuit to switch the LED off. Perhapps to apply a negative bias, discharge the junction, and extinguish the light quickly.

To give an idea of rise/fall times: say the switching signal is 100KHz, which we want to measure to an accuracy of 0.01 %. 100k/0.01 = 10M. This would be our required sampling frequency, and it would be desirable to have the LED go from fully on to fully off in one sample time; 100 ns

 

[ericgibbs]

hi col,
For nanosec switching you should be considering a low power pulsed laser diode.

You will find as well as the refractive index of the sample causing a 'delay', a more dominant delay effect
will be caused by the strength [amplitude] of the receiver signal.

The rise time of your detectors will be affected by the received signal strength, the 'weaker' the signal
the longer the rise/switching times in the detector.

Laser diodes can be pulsed in the 25nSec region, giving coherent light, but they are also prone to producing
non coherent light at the end of the driving pulse, which can be microsecs to decay.

 

[Roff]

To give an idea of rise/fall times: say the switching signal is 100KHz, which we want to measure to an accuracy of 0.01 %. 100k/0.01 = 10M. This would be our required sampling frequency, and it would be desirable to have the LED go from fully on to fully off in one sample time; 100 ns

0.01% is one part in ten thousand, not one part in 100. Your sampling rate would need to be 1GHz, and your LED switching time would have to be around 1ns.

 

[col_implant]

Sorry I ment 1% or 0.01

So does anyone know a nice circuit to switch off an LED quickly?? Or at least to make the on/off light response symetric?

 

[Roff]

Sorry I ment 1% or 0.01

So does anyone know a nice circuit to switch off an LED quickly?? Or at least to make the on/off light response symetric?

Maybe. How much current do you plan to drive through the LED?
Why do you need the precise 50% duty cycle?

 

[col_implant]

Maybe. How much current do you plan to drive through the LED?
Why do you need the precise 50% duty cycle?

Since we are in the early stages we dont yet know about intensity so we will be going with the the more the better approach. We would start with a circuit capable of a number of different current levels, with a max of 100mA.

I dont want a perfect 50% duty cycle... i just dont want the light emitted from the LED to fall off slowly when the signal is removed, since after all it is the electrical signal input to the LED that will be compared to the electrical signal output from the photodiode on the recieve side

 

[ericgibbs]

Sorry I ment 1% or 0.01

So does anyone know a nice circuit to switch off an LED quickly?? Or at least to make the on/off light response symetric?

hi,
What circuit are you going to use to switch ON the LED.?

 

[col_implant]

Ha, hopefully the same circuitry... havent designed anything yet. Tryin to get some ideas before we do

 

[ericgibbs]

Ha, hopefully the same circuitry... havent designed anything yet. Tryin to get some ideas before we do

hi,
While you are doing the feasability study, look at the following:

Use a low power laser diode, pulsed, using an avalanching transistor.

Between the sample and the detector use a Schottky optical bandpass filter.

Designed to pass the coherent laser frequency.

 

[Roff]

Col_implant wrote:

Ha, hopefully the same circuitry... havent designed anything yet. Tryin to get some ideas before we do

Same circuitry as what?

I can show you a circuit that should switch LED currents in 5ns or less, but within that time range, the rise and fall times might not be symmetrical, e.g., one might be 3ns, the other 5ns. Should I pursue it?

 

[col_implant]

Yea please do... cheers

 

[ericgibbs]

Col_implant wrote:Same circuitry as what?

I can show you a circuit that should switch LED currents in 5ns or less, but within that time range, the rise and fall times might not be symmetrical, e.g., one might be 3ns, the other 5ns. Should I pursue it?

hi Ron,
I would be interested.