Need Statement
(Not a lot of experience with power electronics, apologies). I need to power a low-voltage halogen lamp (OSRAM 64328 HLX, 65W, Nominal voltage 9.8V, 6.6A) with a 230V AC source. Additionally, I need to have the capability to digitally control the powering of this lamp, which I will probably do with a National Instruments DAQ (so with 0 to 12VDC, and a current of mA).

I need the light source to be as stable as possible. Based on some research, I've seen that an interesting approach is to make use of an LDR to quantify the illumination output of the lamp, and if a deviation or drift is captured, to compensate making use of a PID + digital control to change the voltage provided to the lamp. In my case, in contrast with the paper, I plan to implement the PID directly in the DAQ (and I'm working with a much powerful lamp).

Question
What is the most efficient solution to approach this?

Extra Input
An idea that I have been looking into is to make use of a PWM that can handle the 230V AC and with PWM regulation move around the nominal voltage of the lamp. However I am not really sure how a hardware implementation of this looks like, if possible.

I also imagine that a weakness of this approach might be that the LDR could saturate easily with such a powerful lamp, so hopefully I will find one that can work with this illumination without saturating. I am mainly interested in the NIR output of the lamp, so a InGaAs LDR seems to be the way to go. The LDR would be located in a dark container, were the lamp is the only source.

Thank you all in advance for your time and input!

 

How are we defining efficiency and why do you think your approach makes any sense. Using mains power to directly derive 9.8 VDC is a dubious approach and has serious safety concerns. The duty cycle would be constrained to a very small range (3% ± 0.5%) making effective control all but impossible. The reason we distribute AC power is because transformers can be used to adjust the mains power to the voltages we need. Using a 230VAC to 12.6VAC transformer would allow a very efficient SMPS to produce 9.8 VDC at approximately a 50% duty cycle with great adjustability and control.

 

I also imagine that a weakness of this approach might be that the LDR could saturate easily with such a powerful lamp, so hopefully I will find one that can work with this illumination without saturating.

There's a simple solution to this: attenuate the light reaching the sensor, either by a filter or an aperature of some sort. There's no need to expose the sensor directly to the lamp's full brightness.

I am mainly interested in the NIR output of the lamp, so a InGaAs LDR seems to be the way to go.

Rather than a light-dependent resistor as a sensor, you might consider an integrated device such as TI's OPT101. With no external components, it will give you a voltage output proportional to light intensity that you can feed directly into your data acquisition system.

I also second @Papabravo's comments, above.

 

Since this is a light source for a research experiment, using PWM to control the optical output would be unwise.

All incandescent lamps have finite warm-up and cool-down times and properties. You need to use a constant output power supply to power the lamp. Be aware that the spectral output will be dependent on the temperature and hence the electrical power applied to the lamp.

For pulsed output, use a mechanical shutter.

 

You can‘t use a LDR because the EU has banned them due to the tiny amount of cadmium that they contain, so you’ve got to use a photodiode which is 55% arsenic. You know it makes sense.

Don’t LDRs vary a lot with temperature and drift badly over time?
You might be better off using a halogen lamp to calibrate your LDR!
I’d go for the OPT101 because it is a better solution, not just to avoid the RoHS police.

But seriously, how about the good old fashioned phase fired triac dimmer? A good design will happily drive a transformer with the lamp on the secondary, avoiding some very hot rectifiers with 6.6A going through them. I suppose that In a way it IS PWM.

Why the halogen lamp and not a high power infrared LED?

 

For sensor, I would be looking at a Light-to-Digital Sensor such as OPT3002.

 

For sensor, I would be looking at a Light-to-Digital Sensor such as OPT3002.

The only drawback I see to that part is its spectral response, which peaks at 500 nm (blue) and falls off rapidly in the NIR region. OTOH, the OPT101 peaks in the infrared at 850 nm.

 

The only drawback I see to that part is its spectral response, which peaks at 500 nm (blue) and falls off rapidly in the NIR region. OTOH, the OPT101 peaks in the infrared at 850 nm.

Not to mention that you have now got to implement the control loop digitally, and keep it stable, bearing in mind all the delay times such as the time the lamp takes to respond to a change in power. I’m sure flickering isn‘t going to be acceptable. This looks like a problem in need of an analogue solution.

 

Not to mention that you have now got to implement the control loop digitally, and keep it stable, bearing in mind all the delay times such as the time the lamp takes to respond to a change in power. I’m sure flickering isn‘t going to be acceptable. This looks like a problem in need of an analogue solution.

Or perhaps a solution in search of a problem.