Measuring flash intensity using a photodiode

[joe25]

Hi everybody,

I want to measure the flash intensity of my flash lamp using a photodiode.

The maximum repetition rate of the flash lamp is 500 Hz, the pulse duration ~ 2 µs.

At the moment I'm using a PCI 6602 counter/timer card, therefore I have to buy new hardware (max 2000$) anyway.

The accuracy should be as high as possible because the signal is used to normalize the flash fluctuations in my measurement.

I'm not sure which NI hardware meets my requirements (USB 6009, NI PCIe-6351,...) because there are so many options.

Any recommendations?

Thank you!

Johannes

[johnsold]

Johannes,

Before you can choose a device, you must convert your measurement requirements into values which can be compared to the device specifications.

What you have specified about your measurements:

1. Signal source is photodiode.

2. Light source is flash lamp with frequency <= 500 Hz and pulse duration ~ 2 us.

3. High accuracy.

4. <= $2000.

Comments:

1. A photodiode is not used by itself. What other circuit components are used with it? What is the voltage or current you get from your photdiode circuit at maximum flash lamp intensity? At minimum intensity? What is dark current (or voltage at output of circuit)? How much stray light interference will there be?

2. What is the shape of the optical pulse? Do you need to record the shape, the peak amplitude, or the area under the curve? If you want the peak, what is the duration of the peak (for example the time the signal is > 90% if the instantaneous peak)?

3.a. Do you need high accuracy or high resolution? 

3.b. How much accuracy (or resolution) do you actually need? What is the limit of accuracy on other parts of your system? You probably do not need more accuracy than you have in those other parts.

4. What else will you be measuring? In particular how many channels of data do you need to measure? Do you have triggering signals availabel? Are they digital or analog?

Lynn

[joe25]

Lynn,

Thank you for your support!

1. At the moment no photodiode has been purchased and it is quit difficult to tell what the maximum/minumum voltage will be. But we expect a very low intensity (scattered light on the outside of a fiber should be measured) und the stray light interference of course will be as low as possible. Sorry that I have no more informations but on the other hand I'm very flexible with the other circuit components because they have to be bought anyway.

2. I have attached the shape of such a pulse. To have an information about the total flash intensity the area under the curve should be recorded.

3a. Well, I need to caputure every flash individually. Therefore the high resolution is not so important (500 Hz max) but to get the most accurate area information I would say that high accuracy and high resolution are equal.

3b. Each flash is utilized for an UV-VIS-absorption measurement and due to the flash fluctuations the noise is quit high (2-6 % depending on the flash lamp model). The noise should be reduced by using the photodiode. Therefore the noise of the photodiode and the readout should be as low as possible (< 0.1 %).

4. The PCI 6602 measures frequency and generates TTL triggers for the flash lamp and spectrometer. One more digital trigger can be used for the photodiode-readout.

I could change the PCI 6602, too. But then I would need at least 6 counters (< 3000$) and equal performance.

Does this help?

[johnsold · Accepted solution]

Johannes,

Yes. The additional information was very helpful.

1. You may want to consider a photomultiplier tube as opposed to a photodiode.  That will spend most of your budget there rather than on daq acquisition equipment but may give you better optical performance. PMTs can be very sensitive (single photon responses), fast, and low noise.

2. and 3. For an area under the curve measurement I would consider a high speed integrator (fast op amp with capacitor feedback).  The output is a voltage proportional to the area under the curve.  Since you only need to read it at 500 Hz, a relatively slow and less expensive DAQ device could be used. You also need to reset the integrator after each read.

Consider the comparison between direct digitization of the curves in your image to the integrator I proposed:

A digitizer operating at 10 MHz would capture 3 or 4 non-zero samples on the narrowest peaks and probably not sample at the highest part of the pulse.  Thus, the calculated area could have significant errors. Certainly larger than your target < 0.1%. Faster digitizers with >= 12-bit resolution (required for accuracy) are very expensive.

The integrator op amp costs ~ $10 US.  The other components required would increase that slightly. Even if you needed to buy a power supply and an enclosure for the integrator, the cost could be < $200-300 US.  The sampling rate of the A/D conversion is ~ 500 Hz.   Even a USB-6008 (<$200 US) is fast enough, although its triggering and accuracy may not be good enough for your application. Certainly there are several devices in the < $1000 US range which are more than adequate.

4. Stay with the 6602 for your frequency measurements and timing generation. It works well, has a proven track record, and you already have it. Add a suitable analog input device and you should be ready.

I do not want to make this sound overly simplistic, because it is not. What you are trying to measure requires some care and some effort to assure that you get meaningful results.  For example the discharge that powers the flash lamp may produce strong electrical transient signals.  These could be much larger than the photodiode/PMT outputs and they occur simultaneously.  Preventing those transients from corrupting the measurements might require a substantial amount of testing and design.

Lynn

[joe25]

Lynn,

thank you, this is very helpful, especially the hint with the high speed integrator!

Can I use a photodiode instead of a photomultiplier just for testing the proposed assembly? What device (PCI-(E)-Card or USB would you recommend for this application (e.g. PCIe-6321)?

The final assembly might get tricky as you have already mentioned because all parts have to be placed below an electric motor, so we'll see.

Johannes

[johnsold]

Johannes,

I think a photodiode for preliminary testing should be fine.  You might find that it is sensitive enough and you can avoid the expense of the PMT.

I would probably use another PCI device since you have the counter in that form (assuming that you have another slot available). The X sereise devices seem to be good ones.  the PCIe-6321 has a maximum sampling rate fo 250 kS/s so you would definitely need the integrator or some external circuitry with it.

Lynn

[joe25]

Lynn,

which sampling rate would be sufficient for an option without an integrator (integrator is fine, just as a general question or a second option)?

Johannes

[johnsold]

Johannes,

Using the images of the flash lamp intensity you posted earlier as a guide, I estimate that sampling at 10 MHz would result in errors ~5-10% for the area under the curve (neglecting any errors in the A/D conversion). If I were digitizing those signals I would want at least 25 MHz sampling rate.

Lynn