I had an idea to measure the speed of light. Instead of using a laser and say bouncing 100s of times off two parallel mirrors to get the needed distance...I thought maybe getting 100m of optical cable (which may be more like 150m of light path...not checked yet!).
I want to generate a very fast pulse. Seeming as light is 3E8ms-1, in a 100m (for errors sake) length of optical cable, id need a pulse of t=100/3E8 = 33E-8 seconds...this is 3.3Mhz.
So, is it possible to make a pulse only in the range of 0.4 microseconds? Ideally 0.2microseconds so half on half off...and there maybe be counters that can resolve down to 0.2 microseconds...
I was then thinking of using a splitter so a small portion of the beam is fed out of the loop on each pass of the beam. A counter with an amplifier can count the pulses...once say 100 passes have happened (about 15km of optical pass) then the counter stops the timer.
15km would give me 3E8/15E3 = 50 microseconds.
I have read that off the primary coloured shelf LEDs can be pulsed even up to 50Mhz which gives me some hope.
A 5% feed off beam splitter may give me the needed signal for an amplifier to trigger a binary counter.
I have not overly looked at components yet...but I know of counters in the 10s of MHz and OPAMPS can be in the MHz range...
challenging experiment, might work. note that pulses of light will stretch
Note that the speed of light was determined long ago before people had optic cable and computers with short pulses etc. I find the old experiment with a fast rotating gear (and mirrors) real genius.
See - Foucault's measurements of the speed of light - Wikipedia
robtillaart:
challenging experiment, might work. note that pulses of light will stretch
Note that the speed of light was determined long ago before people had optic cable and computers with short pulses etc. I find the old experiment with a fast rotating gear (and mirrors) real genius.
See - Fizeau–Foucault apparatus - Wikipedia
Looks interesting, till I read the 8km between source and mirror part ha! To scale that down to say 8m class room length, the hundreds of revs per second would be more like 200,000revs a second! I have no idea if there is even a gear possible and a motor/gearbox capable of that...
I have read that off the primary coloured shelf LEDs can be pulsed even up to 50Mhz
It looks like you can find a source for optical pulses with short rise time. What loss can you expect in 100m of fiber optic cable? What will you use for a detector?
Does pulse length really matter? I think you would want to start a timer when you trigger the pulse, and stop it when it arrives a the other end of the fiber optic cable. At 16Mhz .33 us is about 5 clock cycles. You might want to used a Due with 84Mhz clock. You may not need to use multi pass reflections within the cable if you are willing to allow +/1 one count errors of approximately 4% with the Due.
Maybe I can use an LED and flash it at a stupid (the 50Mhz?) which is a pulse period of 20 nanoseconds.
Over 100m, light would take 333ns to go from source -> desto.
Time = distance/speed = 100/3E8 = 3.3E-7 seconds = 333ns.
Could use 2 counters. Id need counters that can have such a low resolution...talking 100MHz, to increment the counters.
I could feed the carry out of one counter to the "hold" pin of the other that is being fed by the fibre cable.
As it takes 333ns for every 100m of cable, there will be a 333ns delay between the counters being incremented, so I would expect 333/20 = 16.5 counts less on the one that was incremented via the fibre.
The counter from the clear path source will, when topped out, will send the carry-out pin high and trigger the fibre's counter to hold its current value. The value can be read.
cattledog:
It looks like you can find a source for optical pulses with short rise time. What loss can you expect in 100m of fiber optic cable? What will you use for a detector?
Does pulse length really matter? I think you would want to start a timer when you trigger the pulse, and stop it when it arrives a the other end of the fiber optic cable. At 16Mhz .33 us is about 5 clock cycles. You might want to used a Due with 84Mhz clock. You may not need to use multi pass reflections within the cable if you are willing to allow +/1 one count errors of approximately 4% with the Due.
Sounds a bit of a plan. I have hit a small "hitch". May have a way around it.
A sensor, such as a phototransistor, has a pretty large fall time...well more than 50Mhz worth of resolution.
I wanted a short pulse length so I could have multiple passes through the 100m cable.
If the pulses were too long, they would interfere with each other.
100/3E8 = 333ns. A pulse any longer than that would mean the fibre would be one continuous loop of light. With a 20ns pulse, that would be only 1/16th of the fibre would have light in it. This means the remaining 313ns of fibre is "dark" so a counter could be incremented.
Your idea is a good idea! I could have the sensor time removed. The sensor lag could be measured I guess...and the value subtracted.
EDIT: On a quick search, someone managed to overclock a DUE to 114MHz.
I don't think you would use a timer as such. You would just spin counting instructions til you saw your signal come back. I'm not sure how well this would work though. How many instructions could the due execute in the 300ns?
Still worth a go though- even if it's a total failure you'll know why and measuring the speed of light is a pretty cool thing to fail at.
Actually my mind boggles at the idea that our processors are fast enough that you could start a pulse and actually execute some code before it got 100m.
Grumpy_Mike:
Do you know the speed of light in a fiber optic cable is different from that in air which in turn is different from that in a vacuum?
What computer do you have? Almost certainly the answer is no. You are not in Arduino country here.
Yes. The difference between air and a vacuum is far well below any other possible accuracy and standard deviation a device like this could fathom.
EDIT: So forgot to mention* the refractive index of the optical cable can be taken in to account. The experiment is more a show of using distance and time to calculate the speed of light rather than say using diffraction/wavelength.
As air =1.0002 times that of c.
Further, what would be the minimum time for a DUE at 114Mhz to make a pin high and then repeatedly read a digital in pin?
This kinda hangs on the using a DUE.
If not, maybe the high MHz scale counters are my only option.
There are some instructions on how to perform an educational experiment to measure the speed of light here.
You can also do a similar experiment, but using electricity rather than light,by sending narrow electrical pulses down a long length of coaxial cable, and measuring the time delay for the pulse to reach the far end. (Use a 100m reel of cable).
This is actually simpler to do, as you don't need any detectors to convert the light to an electrical signal.
Ok, here is an idea, but it requires some means to pulse LED's on an off in at most nanoseconds. But maybe it doesn't even require a microprocessor. Arrange a collection LED's on a circular arc using as big a diameter as your experiment will allow. Now put a photoresistor at the circle center, hooked up as half of a voltage divider. When it receives light, its resistance drops, and you can measure a sudden jump at the midpoint of the voltage divider. If you have a relatively fast oscilloscope, say 100 Mhz., you can measure pulses, or intervals between pulses, down to 10 ns.
Now, set up a flash sequence on the LED's (this is where a fast timer chip comes in, but they do exist), so that each one is flashed a certain number of nanoseconds later then the last one. If you have a 10 ft. radius, and you pulse these 10 ns. apart, then (roughly), each pulse will arrive at your photoresistor just as the last one is ending. Of course, you do have to adjust this idea for rise and fall time of the pulses, but LED's have rise and fall times in the low ns. But if you reach the timing between pulses where the photoresistor is lit for the whole duration of the flash sequence, that interpulse timing (accounting also for each pulse length) should be an indicator of the fact that the light took just that long to travel from each LED to the center.
I admit this is not an exact description but I think it has some potential to do this experiment with reasonably simple equipment. I might try it myself.
Delta_G:
I think you got something mixed up. Speed of light in a vacuum is the fastest. There's no way it's faster in air than a vacuum. That number has to be less than 1.
The refractive index. This is proportional to time.
jrdoner:
Ok, here is an idea, but it requires some means to pulse LED's on an off in at most nanoseconds. But maybe it doesn't even require a microprocessor. Arrange a collection LED's on a circular arc using as big a diameter as your experiment will allow. Now put a photoresistor at the circle center, hooked up as half of a voltage divider. When it receives light, its resistance drops, and you can measure a sudden jump at the midpoint of the voltage divider. If you have a relatively fast oscilloscope, say 100 Mhz., you can measure pulses, or intervals between pulses, down to 10 ns.
Now, set up a flash sequence on the LED's (this is where a fast timer chip comes in, but they do exist), so that each one is flashed a certain number of nanoseconds later then the last one. If you have a 10 ft. radius, and you pulse these 10 ns. apart, then (roughly), each pulse will arrive at your photoresistor just as the last one is ending. Of course, you do have to adjust this idea for rise and fall time of the pulses, but LED's have rise and fall times in the low ns. But if you reach the timing between pulses where the photoresistor is lit for the whole duration of the flash sequence, that interpulse timing (accounting also for each pulse length) should be an indicator of the fact that the light took just that long to travel from each LED to the center.
I admit this is not an exact description but I think it has some potential to do this experiment with reasonably simple equipment. I might try it myself.
This is a nice idea. Cables must be of a similar length! Electricity is about 0.01c you could end up accidently measuring the factor of cable length vs c. Just as forewarning :).
Measuring the speed of electrical pulses does not have the same "wow" factor for the students as measuring the speed of light.
Although, explaining the "speed of light" part of electricity and the propagation of electromagnetic fields vs the slow moving electrons with kinetic energy in response to that field may be a more interesting teaching task.