Solutions for reducing noise from various sources

Dark Noise (CardioPDA)

The place to begin is to examine the noise in the dark. This noise should not be larger than a certain size and should have a flat FFT spectrum. With the CardioPDA system, measure the dark noise with the following settings:

1. Second stage amplifier low-pass filter: 1 kHz;
2. Second stage amplifier high-pass filter: AC2 on;
3. Use fastest acquisition rate: 1.3 usec;
4. Number of points: 1024;
5. First stage gain: high; and
6. Second stage gain: 50.

Determine the size of the dark-noise by measuring its RMS value under Output, determine its frequency content by using the FFT option under Trace. A flat spectrum with an absolute RMS value of less than 50 is good.

Extraneous Dark Noise (CardioPDA)

There are a number of recognized causes of a not-flat FFT; all of them can be eliminated. A peak at 60 (50) Hz (or harmonics) indicates AC noise in the system; 120 or 180 Hz may come from the room light; peaks around 200 - 300 Hz may come from acoustic noise (e.g. from the air handling system); a peak near 1000 Hz may come from oscillation of the amplifier system.

Extraneous Light Noise

After the extraneous dark noise is minimized, a second class of noise sources, termed extraneous or technical noise, may be present after illuminating the preparation. Again, there are a number of recognized causes, all of which can be eliminated.

Fluctuations in the output of the light source:

Tungsten filament lamps. It is not difficult to provide a power supply stable enough so that the output of a tungsten filament bulb fluctuates by less than 1 part in 105. Tungsten filament sources are almost always used in absorption measurements and are also preferable in those fluorescence measurements where relatively large light intensities are obtained (i.e. a photocurrent > 10-8 amps). LEDs. Ultra Low Noise LEDs with detached fan from Prizmatix (www.goldstonescientific.com). Low Noise LEDs from Cairn Research (Cairn Research Ltd). Laser Diode Illuminator (LDI). Powerful, low noise and narrow bandwidth LDI from 89North and Cairn Research. You can test for light source noise by looking at the noise with an evenly illiuminated object with no contrast. If the noise is still present, it is probably in the light source.

Vibrational Noise

Another source of extraneous light noise is vibrations in the light path. This noise tends to be in the frequency range of 1-50 Hz.

Shutters. The opening of a shutter attached to the microscope can cause a very large vibration signal. We found that this noise damps out over a period of about 1 second. Vibrations from the shutter can be greatly reduced by mounting the shutter so that it is not touching the microscope (i.e. on another platform that is vibrationally isolated from the microscope). Floor Vibrations. A number of precautions for reducing the effects of this vibrational source include eliminating air-water interfaces, eliminating convection currents in the lamp housing, filtering the saline to remove particles, and mounting the apparatus on a platform made from two optical table-tops separated by (and mounted on) air-filled rubber tubes (obtained from Newport : isolator, air cushion, #5274). Embedding ganglia in 1-3% agar further reduced vibrational noise. Minus K Technology sells Biscuit bench top vibration isolation tables with very low resonant frequencies. They provide outstanding vibration isolation. (Choose a model which will support >25 pounds more than the weight of your apparatus because you may have to add weight to balance the table.)

Shot Noise

When all of the extraneous noises are reduced enough, the remaining noise will either be dark noise (very low light intensity) or shot noise (higher intensity). In the CardioPDA apparatus this crossover occurs at a resting light photocurrent of about 5 x 10 -10 amps. In the CardioCCD system the crossover occurs at an intensity about 10 times lower. Shot noise arises from the statistical nature of photon emission and detection. In a shot-noise limited measurement, the signal-to-noise ratio is proportional to the square root of the number of measured photons and the bandwidth of the photodetection system. In this situation, increasing the signal-to-noise ratio can only be achieved by 1) increasing the illumination intensity, 2) improving the light-gathering efficiency (e.g. larger numerical apertures) of the measuring system or 3) reducing the system bandwidth by lowering the low-pass filter cut-off frequency.