My concerns came up when I was attempting to make some image-planed holograms on PFG-3c plates. PFG-3c is slow, 3000 microjoule/cm^2, and I was getting multi-minute exposures. Parts of the image was coming out, but generally there was a big black blob in the middle of the plate, with rings like an interferometer would make.
So, first thing is vibration, and I re-did all my baffling, sealed an air duct which passes over my table (ignored up till now because I normally just turn the AC off when making holograms), and spent a lot of time staring at the output of an interferometer.
Then I saw the laser blink. Oh Joy, my relatively difficult to acquire wonderfully long coherence length 315M is unhappy.
So, I plugged a photo-transistor into the microphone input of my computer, and wrote a quick-and-dirty program to watch the audio, and when a sample greater than a given amplitude is read, dump the surrounding second of samples. One problem with this setup is that the mic input is AC-coupled. So a pulse looks like two pulses, one going positive and one going negative. Block the beam momentarily and you see:
FIXME: sample data here
Then I listened to my florescent lights. 120 Hz. Which is odd because I bought ones with electronic ballasts, I was expecting 400 Hz. Then I listened to my LED flash light's PWM dimmer. Then I was finished convincing myself the apparatus worked and got back to it...
I placed the photo-transistor on the stand which my beam splitter is mounted on, and had it watch the light which spills out of the beam splitter (its a polarizing cube beam splitter, but its not AR coated for 532nm...).
The laser is surprisingly noisy. Unless its my mic input... I was getting pulses with an amplitude of 500 (on a scale of +-32768) in bursts, several for a minute and then nothing for five or ten of minutes. Very irregular.
Then I started tapping on things. Vibration causes events. Then I noticed that what was really happening was the vibration was causing the photo-transistor to move. I just taped it to things, and left too big of a cantilever.
So I was getting false positives. OK, tape the sensor down more securely, put all the baffling back on, and treat it like making a hologram. Don't enter the room while measuring (air currents), turn off the AC. That reduced events dramatically. I had none for two hours. And then two in one minute...
Then I assembled an interferometer, and put the photo-transistor in part of the interference pattern. That was VERY sensitive. Unusably sensitive. I could watch the interference pattern and see it was vibration, and not mode-hops. A mode-hop is more or less instant, the interference pattern goes *boink* to a new position, instead of perceptibly moving there like vibration. More false positives.
I placed the photo-transistor in the rings reflected from the lens in my spatial filter, about 1.5 feet away so the rings are relatively large, a period of maybe 1/4 inch. When the laser is calibrating itself, you can watch the rings jump as the laser mode-hops. However once the laser finishes, calibrating the rings do not move without significant effort. The interference is between reflections off the front and rear surfaces of the lens, it will take some serious vibration to get those to move relative to each other.
This picture shows the photo transistor (taped to an angle bracket, in turn taped to a baffle), and the interference pattern its observing. Note that the camera is having some moire pattern issues with the rings.
I then had the laser do its calibration dance. I got 73 events in 511 seconds. I then observed the distribution of samples for each event. X is event number, Y is sample value, or standard deviation.
Note how it abruptly calms down on the right side. Thats where it finished its calibration. Also note that when its really mode hopping, its pegging the ADC.
My definition of an event as +- 500 ADC units is way too low. A mode-hop is more like 10,000 ADC units, or more likely a peg.
This graph shows events after the calibration is complete. Note the Y axis is much smaller, these events are nowhere near the magnitude of the hops during calibration. There are 115 events, over about 7.5 hours.
I believe that observing the interference pattern between two surfaces of the same lens detects mode hops and ignores most vibration events. A minimum amplitude of 500 ADC units causes vibration events to be detected as false positives. If the lowest amplitude event during the calibration is considered to be a mode hop, events below that amplitude (IE: all of them) can be discarded.
So I believe my 315M is not mode-hopping at this time. I need to make some long exposure holograms to verify that conclusion.
The software I used can be found in hop1.0.tgz