I am trying to design a turbidity detector and this has to be sensitive to tiniest amounts of contamination present in a sample fluid. My setup consists of a 780nm laser whose current I control with an SMU. Light from this laser passes through a cuvette which contains the fluid and the intensity of the transmitted light is detected by a photodetector using an electrometer. The laser has an in-built photodiode and I use that to monitor any change in the laser's intensity, using another electrometer built by a different manufacturer. My intent here is to normalize the transmitted intensity by the laser's input intensity, to factor out laser's intensity variation. However, the laser is pretty shabby(intensity varies a lot) and therefore I need the two instruments to be precisely time-synced when they make a measurement. What's the best way to achieve this?

What I have thought of:

Send "trigger" signal to both the devices when I want to make a measurement. The problem here is that the electrometers are manufactured by two different manufacturers and there might be a delay-to-measurement difference between the two.

  • \$\begingroup\$ You wrote "the laser is pretty shabby(intensity varies a lot)": please quantify frequency (or better spectrum) of intensity variation. \$\endgroup\$ – Curd Mar 31 '18 at 6:43
  • \$\begingroup\$ That variation - how much error does it cause? \$\endgroup\$ – Solar Mike Mar 31 '18 at 6:59
  • 1
    \$\begingroup\$ I do instrumentation. It's my life. So I've been in similar situations. A solution is that you hire someone (or, better, do it yourself so you know it is done right) to develop a crafted electronic system as well as software that provides precision synchronized values with the same averaging/sampling period (electronically, or otherwise.) Or go to a dual-wavelength approach, perhaps, so that careful timing is less important. If you persist in using equipment from two manufacturers who aren't solving your specifics, you will pretty much get what you deserve. Sad to say. But truth. \$\endgroup\$ – jonk Mar 31 '18 at 7:04
  • \$\begingroup\$ @Curd It's mostly drift, so DC to low Hz is where the variation lies. \$\endgroup\$ – user1155386 Mar 31 '18 at 15:40
  • \$\begingroup\$ @SolarMike Pretty significant, sometimes the reading isn't repeatable at all. \$\endgroup\$ – user1155386 Mar 31 '18 at 15:41

A better laser regulates intensity by current with a PD in a feedback loop detecting a weak sample of the beam internal to the laser. Location and stability of the optics for the PD are critical to stability and are best done internally unless you have a large budget.

PD's are very stable (PT's are not) but temperature can also be regulated to improve this laser stability.

I would be more likely to measure turbidity by ;

  • return loss (reflection) which would be far more sensitive but requires special design of apparatus from stray contain reflections and use a very small aperture to measure a linear reflection rather than a radial reflection. This would apply to low torbidity clear liquids, otherwise ...
  • scattering by diffusion around the container with the beam diffused to multiple detectors.
  • then transmission loss for heavy turbidity (really dirty)
  • multiple wavelengths for sensitivity to particle size from IR to UV depnding on clarity and particle types for photoemission or reflectivity of contaminants.
  • if the non-turbid fluid is not clear, then I would use a dual beam approach with filtered fluid.

I was able to find another crude way to synchronize data(to some extent). I modulated the laser's intensity to transition from completely off to on state during the measurement. This change in intensity being of a large magnitude, created a strong 'edge' in the data collected by both the PD-s. I then aligned that edge in post-processing (both the electrometers have between setup with the same NPLC and sampling rate). This transition in intensity also helped me zero out offset current. Maybe the next step is to continuously modulate the intensity at a certain frequency, edge align and then demodulate.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.