Optical Sources and Optical System Efficiency

Optical Sources and Optical System Efficiency

Understand the power you are delivering

Optical Sources and Optical System Efficiency


High power optic sources are potentially dangerous to the researcher and care should be taken when dealing with them. There is the possibility of damage to tissue in the form of burning or irreparable eye damage. There are many variables that affect the risk level (including wavelength, duration of exposure). For light in the visible spectrum, it is generally a good idea to keep the total optical power below 15 mW. It is always a good idea to wear eye protection when working with high-powered optic sources. We recommend laser safety glasses from Thorlabs. Still, never look directly at your light source!

Lasers and LEDS

There are two major types of light sources used in labs today. They both output a set amount of light, and the efficiency throughout the entire optical system is affected by mechanical loss as well as timing characteristics. Power meters (ex: Thorlabs S120C) are great for measuring steady state power, but timing characteristics are better understood with photodetectors (ex: Thorlabs DET10A2; however, the photodetector does not output an optical power measurement. To bridge this gap, correlate the measurements by measuring the steady state of the optical source with the power meter, then measure the same power with the photodetector (in volts). Assume a linear comparison between the two (example: if the optical source produces 10 mW per the power meter, and 1 Volt on the photodetector, it is safe to assume that by changing the power on the optical source such that it produces 0.5 Volts, you are now at 5 mW).


Light Emitting Diodes (LEDs) aare cheap, small, have great on/off characteristics, and are available at many wavelengths. But they have two key issues when it comes to optogenetics. First, the light LEDs create gets scattered over a wide area. Even the best collimation systems only capture about 20% of the light into optogenetic patch cords. The second drawback is that LEDs generate light of a relatively wide range of wavelengths. A “473nm” LED may actually be producing light ranging from 450nm to 500nm. Opsin performance in optogenetics is highly wavelength dependent, so ensuring optimum opsin performance can be tricky with LEDS (you can get bandpass filters from suppliers like Chroma to tune the light). Some more basic info on LEDs may be found here.

Lasers are bulkier and more expensive than LEDs, but they get around the two main drawbacks of LEDs for optogenetics. Laser light is highly focused, so it can be directed into a patch cord easily and efficiently. Lasers are also significantly better at delivering light of a specific wavelength (generally, +/- 2 nm), making it much easier to calculate opsin activation levels. Lasers are therefore our usual light source of choice for optogenetics.


It is critical for optogenetic researchers to understand the optical power they are delivering to the tissue they are researching. Consistent, repeatable, and well understood optogenetic results are only obtained through careful optic consideration. It is an insufficient assumption to measure the power at the light source and assume it is the same at the tissue. Light is lost or gained through mechanical means and timing characteristics (pulsing). To deliver the intended optic stimulation, care should be taken to understand these gains or losses. Any losses of optical power can be thought of as inefficiencies.

Examples of sources of inefficiency: · The laser’s collimator has not been properly aligned · The tip of the patch cord is dirty, cracked, or not sufficiently polished · The optical fiber within the patch cord has been damaged · The laser uses TTL input pulses to turn on and off

These (in)efficiencies compound upon one another. If there is a 50% loss between the laser and the patch cord along with another 50% loss between the patch cord and the cannula, this would translate to a total system efficiency of only 25%.

Measure Efficiency from Mechanical Loss

In order to define the efficiency of a light delivery system, first define the starting point. With the light source constantly on, measure all the light output after collimation. Use an optical power meter (example: Thorlabs PM100D) to measure the power with a sensor (example: Thorlabs S120C) placed close to the light such that all light is collected on the sensor. But don’t let the laser barrel make contact with the sensor, as it will easily scratch and damage the delicate sensor.

Some of this initial optical power will be lost when it is passed through a patch cord. Connect the patch cord to the light source and measure the optic power out from the patch cord. Again, be careful not to scratch the power sensor with the tip of the patch cord. The ratio of power coming out of the patch cord to the power at your starting point defines the efficiency of the patch cord. Note that two lasers may collimate light differently enough such that one patch cord could have a different efficiency with each laser. So whenever possible, measure the efficiency of each patch cord with the laser you intend to pair it with. Any changes to the light delivery system can significantly change system efficiency.

Researchers should also measure the efficiency of each of their cannulas before surgical implantation as there is no way to measure its efficiency once it has been surgically implanted.

Regular cleaning helps to maintain system efficiency. Cleaning can be done with a Kimwipe wetted with isopropyl alcohol. If a cleaned patch cord still isn’t performing well, then it may need to be freshly polished.

These mechanical inefficiencies can be assumed to remain the same over different powers and wavelengths. A patch cord with 80% efficiency will output 8 mW when 10mW is input and 80 mW if 100mW is input, and you can assume this will be true for green or blue light.

  • Example: I wish to deliver 10 mW of power to some tissue using one laser, one patch cord, and one of two different cannulas (labelled Cannula 1 and 2). I set up my system and measure 10 mW of power out of the laser, 8 mW out the patch cord, 6 mW through Cannula 1, and 7 mW from Cannula 2. I can therefore calculate an efficiency of 8/10 = 80% for this patch cord on this laser. Cannula 1 has an efficiency of 6/8 = 75% and Cannula 2’s efficiency is 7/8 = 87.5%. I mark which is which, implant them, and wait for recovery. On the day of experimentation, I calculate in reverse to determine what the power output from the laser should be. I take my desired output (10 mW) and multiply by the inverse of each efficiency measurement. If I’m using the Cannula 1, that calculation is 10mW x (1/0.75) x (1/0.80) = 16.7mW. So, I would need to have my laser output at 16.7mW of light to deliver 10mW out of Cannula 1. If I’m using Cannula 2, I will achieve an output of 10mW from the cannula when the laser is turned up to 10mW x (1/0.875) x (1/0.80) = 14.3mW.

Measure Efficiency from Timing Characteristics

Optogenetics typically utilizes short duration, pulsed optic light to achieve opsin activation. Mechanical shutters achieve very fast on/off characteristics, but are bulky, noisy, and require open optics. Therefore the typical method to achieve optic pulsing is through a TTL signal sent directly to the optic source. This method is imperfect, resulting in odd on/off timing characteristics that we will want to account for. These characteristics are too fast to measure reliably with an optic power meter; instead, a photodetector (such as the Thorlabs DET10A2) and an oscilloscope allow for extremely precise measurements of timing characteristics


Note that photodetectors like the DET10A2 should output a voltage linearly proportional to the amount of light hitting the sensor. But the slope of this linear relationship is wavelength dependent. To translate photodetector measurements to optical power, you will want to first measure the optical power of a laser at steady state. You will then want to pass the light into the photodetector and note the output voltage on the oscilloscope. The ratio of optical power to photodetector voltage output allows you to better quantify the relationship between timing characteristics and power levels.

Compensating for Timing Characteristics

If we graph the laser output power over time, the area under these curves illustrates the total amount of optic power delivered in each pulse.Typically, the light source is slow to reach its steady state, but some sources overshoot it, delivering more optic power than expected.

Timing Characteristics Red outline has two ideal pulses, while other colors represent imperfect pulses frequently seen

The red lines in the figure above illustrate an idealized optical pulse. It instantly rises to the appropriate power level before instantly dropping back down to zero output when the pulse is finished.

The pulse shown in green takes a notable amount of time to rise before plateauing to the desired power level. The pulse also gradually drops back to zero when finished. When compared to the red curve, the green curve has less area under it. This equates to less power being delivered. It is therefore recommended to increase the power of this light source, compensating for the initial loss of optic power due to pulsing.

The blue curve actually overshoots the steady state power level before settling back down to it. These pulses are delivering more power than the ideal pulse seen in red. It is therefore recommended to decrease the power of this light source, compensating for the initial increase of optic power due to pulsing.

  • Example: I wish to input a 10 msec, 10mW single pulse to stimulate an opsin. I first need to find the ratio of optical power (mW) to millivolts (mV), which is the metric used by the readout on the oscilloscope. I turn my laser on at steady state and measure 10mW with a power meter. I then pass the steady light into a photodetector and see the traces below. The image on the left shows a steady state laser hovering at 10mV on the oscilloscope. So in this simplified example, we now know that each millivolt on the oscilloscope is equal to one milliwatt of optic power (1mW = 1mV).


When we pulse the laser for 10 milliseconds, we see the image in the middle. Even though we haven’t changed the power settings, the average optical power during the pulse is much closer to 5mW than 10mW. The intensity also has a slight rise during the duration, slowly working its way up to steady state power output. In this example, we can just adjust the laser’s power level up until it is around the desired 10mW. The rightmost image shows this adjusted pulse. Note that the beginning of the pulse is below our desired power level while the end of the pulse is above it. We would ideally want these undershoots and overshoots to “average out” to 10mW, ensuring that we are delivering the desired amount of total optical power with each pulse.

Factors that Impact Timing Characteristics

The on/off timing characteristics of optic sources are often dependent upon chosen optic power. Optic sources typically operate best at their highest power, but these powers are typically far too high for optogenetics. The on/off characteristics will change as the power is lowered. Measure these characteristics at your desired power and compensate for gains/losses. The figure below demonstrates two 10 millisecond laser pulses from the same laser at two different power levels. You can clearly see that the 0.1mW pulse is far more erratic than the 5mW pulse.


One option to achieve consistent timing characteristics at lower power levels is to set the laser to a higher power level and introduce controlled inefficiency to the system. This will lower power delivered at the end while maintaining desirable timing characteristics. A variable optical attenuator example: Thorlabs VAO MMF allows fine control of introduced inefficiencies. The ONE Core also has also designed a convenient holder for the VOA MMF part available from Thorabs.

  • CAUTION: Increasing the light source’s power also increases the safety risk to the researcher. Although the power out the cannula may be low, the power out of the light source may be excessively high. If the patch cord breaks open or someone forgets to turn off the laser before disconnecting the patch cord, there is a serious risk of eye damage.

The duration of light pulses used for optogenetics is usually set by experimental parameters and not the other way around. This means that changing the pulse duration is probably not an option for many researchers. Instead, it may be desirable to adjust for any significant fluctuations by adjusting the power level of the laser, as discussed earlier.


The length of the light pulses for optogenetics are usually set by experimental parameters and not the other way around. This means that changing the pulse length is probably not an option for many researchers. Instead, it may be desirable to adjust for any significant fluctuations by adjusting the power level of the laser as discussed earlier.

ONE Core acknowledgement

Please acknowledge the ONE Core facility in your publications. An appropriate wording would be:

“The Optogenetics and Neural Engineering (ONE) Core at the University of Colorado School of Medicine provided engineering support for this research. The ONE Core is part of the NeuroTechnology Center, funded in part by the School of Medicine and by the National Institute of Neurological Disorders and Stroke of the National Institutes of Health under award number P30NS048154.”