Variable current generator chopping/amplitude modulation

[AloyseTech]

Hi,

I'm currently working on a poject involving modulation of a current through an LED. I'm not really

I need to send a variable luminous waveform to a Device Under Test, but the light should only be emitted when the DUT itself is sending light.


I plan to use a photodiode coupled to a LMV7219 comparator to detect when to send the light.

The DUT output light pulse length is around 50µs. The system must send the light at maximum 5µs after the DUT light pulse starts .

There is a small SAMD11 MCU handling the waveform generation. I'm using a filtered PWM signal as a DAC, that feeds a constant current surce amplifier.

To modulate this waveform, I have thought about 2 options :

A. chop the input of the current source by connecting the input to ground
B. chop the output of the current source by disconnecting the current source from ground.

I created a quick simulation using Falstad to illustrate my ideas (interactive link : http://tinyurl.com/24ngzoer) :

Are these 2 options viable ? What would be the best one option ?

My analog electronic skills are quite limited, so thanks in advance for your help.

Théo

[ajb]

Side note: The SAMD11 has a DAC channel, why not use that instead of filtered PWM?

How much current are you trying to drive through the LED?

Option B is liable to hit the LED with a significant current spike when you turn that low side transistor on.  The output of the op-amp will end up somewhere near or above the current setpoint when that transistor is off, so when it turns on it's going to pull the source of your current source FET down and turn it on relatively quickly.  If the current source op amp doesn't respond quickly enough it will be out of regulation for some amount of time and the current applied to the LED could be a lot more than you were aiming for. 

Option A is okay, but it does depend on the slew rate that the current source can achieve.  Stability and response time are sort of two sides of the same coin.  5us isn't that crazy if your drive current is low enough, but will require some care in choosing components and compensation.

Option C would be to use a transistor to shunt the LED to turn it off.  This is commonly done to dim switch-mode constant current drivers, where this method is very efficient, but if you have a single LED at modest current it could be a reasonable choice here, too.  A stepdown converter that provides just a bit more voltage than you need for the LED and current source would help with efficiency but might not be necessary.

[AloyseTech]

Thanks for your quick reply.

Unfortunately I already use the SAMD11 DAC channel for another voltage waveform that I need simultaneously.

The drive current into the LED will be less than 30mA. It will be a single red LED.

For option B, I already spotted in the coarse simulation that a current spike happens at turn on. I thought maybe it would be compensable but my analog skills are limited and I'm not sure how it could be. The spikes is larger if the RC constant of te current source amplifier filter is higher (which I think is better for stability).

For option A I already imagined that I would need a fast op-amp. Beside slew rate, any other parameter that would be of higher importance?

Option C : I like the simplicity of it. I did not thought about it. Efficiency is not really an issue in this project. What would be the keys parameters for the P mosfet selection in this case to avoid glitch when switching it on and off ?

[AloyseTech]

I'm currently selecting the opamp for the current source. Would the Gainsil GS8552/GS8592 be sufficient ? They have a 1.5/4.5MHz bandwidth respectively (so would be fast enough to handle the quick on/off bypassing of the LED), rail-to-rail input/output and are both available with JLCPCB assembly service.