[Laser] Pulse laser test

[James Whitfield]

I don't think that I am following all of the discussion on pulse
communications.  I suggest a thought experiment to clarify what the ideas
relate to.  Please forgive me if I have slowed the topic.

Suppose that you want to send a BPSK31 signal at a "carrier" frequency of
312.5 Hz.   This was chosen because it is easily generated by a computer
sound card and is 10X the bit rate of the system.  Therefore you should get
ten full cycles of the carrier frequency for each element sent.  Or said
another way, the output tone can reverse its phase no more frequently than
every ten full cycles.  The classic ( for transmission on an HF
transceiver ) narrow band filtering has the amplitude of the phase reversal
point attenuated to zero.  Call this our baseline case.

Send a test signal of all ones or all zeros.  The spectrum will be a single
spike at the carrier frequency. Send a test signal of alternating ones and
zeros.  The spectrum of this signal will be two spikes, one above and one
below the carrier at the baud rate.  Minimum bandwidth, no harmonics of the
carrier, no harmonics of the baud rate.

Suppose that we have a version optimized for driving LED emitters, still a
linear device but not bandwidth limited, so our example has constant
amplitude sine waves that can reverse phase at the zero crossing after ten
full cycles.  Call this Case 1. Our all ones or all zeros test signal is the
same.  The alternating test signal now has sidebands at one third amplitude
for the third harmonic of the baud rate, one fifth amplitude for the fifth
harmonic,  and so on for each odd harmonic.

Now we modify the signal for pulses.  Compare the Case 1 sine wave amplitude
to a reference.  At a certain reference point, the comparitor output will be
High 50% of the time and Low for the other 50%.  Call this Case 2.  The
spectrum of the all ones or all zeros test signal will be the fundamental
plus odd harmonics.  The alternating ones and zeros test signal will give a
spectrum of all of the odd sideband harmonics of all the odd carrier
harmonics.  It would be a complex spectrum, but it should be relatively easy
to tabulate the frequency and amplitude of the stronger ones, say greater
than 10% of the strongest.

Now for Case 3, choose a reference point that gives you 5% High and 95% Low.
I won't attempt to describe the spectrum, I just want to ask is this the
type of pulse you are trying to use instead of a sine wave or 50/50 duty
cycle square wave?

Are you trying to compare the ability to copy signals that are Case 3 ( 5% )
to Case 2 ( 50% ), or are you trying to compare a Case 3 pulse to Case 1 (
100% modulated signal of peak amplitude equal to the pulses )?

Or are you trying to compare short pulses of higher amplitude than the
longer pulses, or sine wave?


Now I have questions about how you are going to try to demodulate the
pulses.  And another thought experiment.  Suppose that your light sensor is
in a camera will a cycling shutter.  To be able to capture the signal you
need to open the shutter when the pulse is present.  One way is to leave the
shutter open continuously.  Suppose that you know that the pulse rate is
312.5 Hz.  If the duty cycle is known to be 50%, can you setup two identical
cameras with alternating shutters, then tweak the shutter synchronization so
that the output of the two sensors is always complimentary?

How about a known 5% pulse?  Are you going to open the shutter for 50% on
two cameras, or use 20 cameras, each open for 5%.  Seems to me, that a 5%
pulse in a 5% shutter time is going to be 10X stronger than the same pulse
viewed for 50%.

OK, I know you can't use 20 cameras, but can you take twenty strings of
samples?  If the sound card is taking 44K samples per second, we get about
140 samples per cycle of the 312.5 Hz tone, and therefore 1400 samples per
bit.  If the bit consists of ten 5% pulses ( 7 samples out of 140 ) in one
of two phase points, how do you average the samples to converge on the
synchronization of the system?  Once you synchronize, what do you do to
transmit the message?  Is it sufficient to simply repeat the message, or is
it better to send a synchronizing "message" then a payload "message".

I can imagine a test bench for these ideas.  Baseline case uses an LED
biased at half brighness, then the signal is added with enough amplitude
that on the lows the LED goes out ( or at least non-linear ).  Case 1 uses
the same, but the software may not really exist to create the signal.  Case
2 and Case 3 use a clipped high gain amplifier to "square" the analog BPSK
signal, but may drop one or two pulses at the zero crossings.  The output
may be directly useable for Case 2, but I suggest a one-shot adjusted to the
correct pulse width for each of the two cases.  To add noise into the
system, add it into the same LED, or perhaps similar LEDs mounted within the
sensor field of view.


James
N5GUI