# Selecting an Opamp for photodiode current conversion

#### abhaymv

Joined Aug 6, 2011
105
Hello everyone,
I'm trying to make a circuit to detect the variations in optical power from a source. I need very accurate results and the source I'm using is an array of about 36 high power LEDs. The Irradiance level (level of light power) in the area of illumination is about 700 Watts/meter squared. (That's nearly the intensity of sunlight.)

Here's what I have to do. I'm measuring the irradiance level (level of light power) of the light falling on the photodiode over an interval of 1 second. As the current output of the photodiode is proportional to light intensity, I'm hoping to pick up the variations in the output of the photodiode. To do this, I'm thinking of
1>Converting the photodiode current output to 0 to 5V voltage range.
2>Feed this voltage to a 16 bit ADC.
4>Store the readings of current in a variable in MATLAB
5>Find the minimum and maximum values in the variable and thus calculate light source instability. ((max-min)/(max+min))

I need help with step 1. I understand that photodiode current will be in microamperes range. I understand that a current to voltage conversion circuit (attached) is used for this.

I need help selecting an opamp for this circuit. As the application demands high accuracy, I need a low noise opamp with output range of 0 to 5 V so that I can feed it into a microcontroller. Can you help me by suggesting an opamp for my application? Will a regular 741 do?

Should I use a phototransistor instead of a photodiode? My light is in a spectral range of 400 to 1100 nm.

#### Attachments

• 17.1 KB Views: 38

#### wayneh

Joined Sep 9, 2010
17,504
I need very accurate results...
If that's the case, you should consider buying instead of building. You need tremendous stability of response across a changing wavelength spectrum and, likely, varied temperatures. Without that, any variations you observe could well be noise and not signal. Getting a large signal to noise ratio will be difficult in a DIY project.

I think if you truly want to measure irradiance with some accuracy, you need to scan for irradiance versus wavelength and then integrate under the curve. And the scanner needs to be calibrated to be accurate at every wavelength. This sounds quite challenging but I'm sure it's all been done before and that there are commercially available solutions.

#### John P

Joined Oct 14, 2008
2,027
What you have there is a basic transimpedance amplifer (hey, my spell checker has never heard that word). Here's what the late and great Bob Pease had to say about the circuit:
http://electronicdesign.com/analog/whats-all-transimpedance-amplifier-stuff-anyhow-part-1

I wouldn't be as cautious as Wayneh, but you definitely need to look at the expected spectrum of the LED light versus the characteristics of the diode. You could be putting out a lot of light and failing to measure it if the photodiode just doesn't respond well to the peak wavelength.

#### wayneh

Joined Sep 9, 2010
17,504
I wouldn't be as cautious as Wayneh...
He did say "very accurate". If he mis-used the word and really needs precision to detect fluctuations, especially high frequency ("high" means >~1Hz in this context) noise as opposed to very slow (hours, days, months) drift, then everything changes in his favor. But, for instance, the spectral output of his LED source will shift in color depending on the current it sees, which is presumably fluctuating. Maintaining a "very accurate" measurement will be a challenge.

#### abhaymv

Joined Aug 6, 2011
105
Thank you for the links, guys. I'll go through them ASAP.
@Wayneh, I'm sorry for the confusion my wordings created. All I need is a relative measure of the intensity of light (power), irrespective of the spectral content.

#### wayneh

Joined Sep 9, 2010
17,504
Well, there is no such thing - they are intimately related - but I get your point.

The excellent articles provided by Bertus will take you a long way.

#### abhaymv

Joined Aug 6, 2011
105
Well, there is no such thing - they are intimately related - but I get your point.
Yeah, I know there isn't a measure of relative intensity if you don't consider the wavelength. But to measure Spectral Irradiance, I'd need a spectroradiometer or a monochromator and some photodiode arrangement. Both are very costly. Well, I might get access to a spectroradiometer at some point, but I'm hoping to measure the temporal instability before that.

Thank you for your help. Those links will also give me an idea of what kind of opamp I'd need, I hope...

#### wayneh

Joined Sep 9, 2010
17,504
Yeah, the short message is to forget the 741. It's literally the oldest op-amp you can find and every one since is better in some regard if not all regards.

Pay attention to the range of voltages in which the inputs can be sensed and the output voltage that can be achieved, relative to the power supply rails. Some op-amps are "rail-to-rail" but many have narrow ranges if operated from a single, low voltage supply. For instance the 741 on 5V has a useful input range not much more than a single volt or so.

#### abhaymv

Joined Aug 6, 2011
105
... 741 on 5V has a useful input range not much more than a single volt or so.
Ouch.

I'll give those articles a careful read. I glanced through them and they're just what I need. A thousand thanks to Bertus and John P! And you too Wayneh! I really appreciate all the help you've given me in my last project, the 'solar simulator'. This one is an extension.

Last edited: