How to build an ultra high speed photo detector? (for detecting red laser pulse)

[Mike Noone]

Hi - For my senior project at my university, I'm working on a project
where we are sending out a pulse of light with a red laser, and timing
how long it takes for the signal to return. Timing the signal is fairly
easily accomplished with an Acam TDC-GPX chip which gives us 10ps
resolution. We are using a high speed laser driver chip which gives us
approximately 40ps rise and fall times on the laser pulse. The issue is
in our detector circuit. Looking at traditional photo diodes, they all
seem to take in the nanosecond range at a bare minimum to react to an
incoming signal. So my question is this: Is it possible to build a
circuit that can detect incoming red laser light that will send out a
3.3 or 5V pulse (differential or single ended) in less than a
nanosecond? Ideally I'd like to see it in the picosecond range - under
100ps would really be ideal. If there is always the exact same delay
(as in a 10ns +- 100ps delay for the output signal) - that would be
just fine. We just need the accuracy.

If it is possible, can anybody give me some pointers as to how to go
about doing this? I should mention that ideally we'd like to keep costs
as low as possible, as this is a student project.

Thanks!

-Mike

 

[Paul Mathews]

Hi - For my senior project at my university, I'm working on a project
where we are sending out a pulse of light with a red laser, and timing
how long it takes for the signal to return. Timing the signal is fairly
easily accomplished with an Acam TDC-GPX chip which gives us 10ps
resolution. We are using a high speed laser driver chip which gives us
approximately 40ps rise and fall times on the laser pulse. The issue is
in our detector circuit. Looking at traditional photo diodes, they all
seem to take in the nanosecond range at a bare minimum to react to an
incoming signal. So my question is this: Is it possible to build a
circuit that can detect incoming red laser light that will send out a
3.3 or 5V pulse (differential or single ended) in less than a
nanosecond? Ideally I'd like to see it in the picosecond range - under
100ps would really be ideal. If there is always the exact same delay
(as in a 10ns +- 100ps delay for the output signal) - that would be
just fine. We just need the accuracy.

If it is possible, can anybody give me some pointers as to how to go
about doing this? I should mention that ideally we'd like to keep costs
as low as possible, as this is a student project.

Thanks!

-Mike

Some hints:
1. Capacitance is the usual problem with photodiode speed. Smaller
photodiodes, PIN type, reverse bias.
2. Reduce load impedance, since T = RC
3. Avalanche photodiodes are yet faster.
4. If you compare differences in timing between 2 optical paths of
different lengths, you can cancel out any delays common between the 2,
which makes it possible to measure c with fairly slow equipment.

Paul Mathews

 

[colin]

Hi - For my senior project at my university, I'm working on a project
where we are sending out a pulse of light with a red laser, and timing
how long it takes for the signal to return. Timing the signal is fairly
easily accomplished with an Acam TDC-GPX chip which gives us 10ps
resolution. We are using a high speed laser driver chip which gives us
approximately 40ps rise and fall times on the laser pulse. The issue is
in our detector circuit. Looking at traditional photo diodes, they all
seem to take in the nanosecond range at a bare minimum to react to an
incoming signal. So my question is this: Is it possible to build a
circuit that can detect incoming red laser light that will send out a
3.3 or 5V pulse (differential or single ended) in less than a
nanosecond? Ideally I'd like to see it in the picosecond range - under
100ps would really be ideal. If there is always the exact same delay
(as in a 10ns +- 100ps delay for the output signal) - that would be
just fine. We just need the accuracy.

If it is possible, can anybody give me some pointers as to how to go
about doing this? I should mention that ideally we'd like to keep costs
as low as possible, as this is a student project.

Yes but it is difficult to do this over a reasonable distance without high
power lasers and avalanche detectors and highly selective optical filter
wich are all quite expensive, however a large lense helps and is cheap.

The amplification needed to recover your very fast pulse edge from the loss
to/from target and the capacitance of the detector will also amplify a lot
of noise.

However there are a couple of other ways of doing this, using a sweep
modulated laser beam and using a synchronous detector to detect the phase
change over a frequency range gives very good results from a low power
laser, using a heterodyne optical receiver improves this method enourmously,
Ive experimented with this with modulation frequencies upto 1ghz.

Another way is to use a tuneable laser and compare the interference from the
target with the interference from a tunable cavity,
to give absolute distance, the resolution is quite remarkable considering
the distance it works over.

Somone was going to look inside one of those Lieca range finders but not
heard anything since.

Colin =^.^=

 

[Ancient_Hacker]

Photodiodes have substantial shunt capacitance which limits their
response time, especially with weak signals. That's why you don't see
any infrared LAN's-- the signals are too weak and low bandwidth to get
very far.

The usual trick is to use FM instead of AM -- this allows you to
extract the echo difference frequency, which is at a manageable rate--
low megahertz instead of several GHz.

 

[Rich Grise]

Hi - For my senior project at my university, I'm working on a project
where we are sending out a pulse of light with a red laser, and timing
how long it takes for the signal to return. Timing the signal is fairly
easily accomplished with an Acam TDC-GPX chip which gives us 10ps
resolution. We are using a high speed laser driver chip which gives us
approximately 40ps rise and fall times on the laser pulse. The issue is
in our detector circuit. Looking at traditional photo diodes, they all
seem to take in the nanosecond range at a bare minimum to react to an
incoming signal. So my question is this: Is it possible to build a
circuit that can detect incoming red laser light that will send out a
3.3 or 5V pulse (differential or single ended) in less than a
nanosecond? Ideally I'd like to see it in the picosecond range - under
100ps would really be ideal. If there is always the exact same delay
(as in a 10ns +- 100ps delay for the output signal) - that would be
just fine. We just need the accuracy.

If it is possible, can anybody give me some pointers as to how to go
about doing this? I should mention that ideally we'd like to keep costs
as low as possible, as this is a student project.

Thanks!

Why not just use a spinning mirror, like Michelson and Morley did in 1879?
;-)

Good Luck!
Rich

 

[John Larkin]

Hi - For my senior project at my university, I'm working on a project
where we are sending out a pulse of light with a red laser, and timing
how long it takes for the signal to return. Timing the signal is fairly
easily accomplished with an Acam TDC-GPX chip which gives us 10ps
resolution. We are using a high speed laser driver chip which gives us
approximately 40ps rise and fall times on the laser pulse. The issue is
in our detector circuit. Looking at traditional photo diodes, they all
seem to take in the nanosecond range at a bare minimum to react to an
incoming signal. So my question is this: Is it possible to build a
circuit that can detect incoming red laser light that will send out a
3.3 or 5V pulse (differential or single ended) in less than a
nanosecond? Ideally I'd like to see it in the picosecond range - under
100ps would really be ideal. If there is always the exact same delay
(as in a 10ns +- 100ps delay for the output signal) - that would be
just fine. We just need the accuracy.

If it is possible, can anybody give me some pointers as to how to go
about doing this? I should mention that ideally we'd like to keep costs
as low as possible, as this is a student project.

Thanks!

-Mike

You'll need a fast silicon PIN detector, or a silicon or germanium
avalanche photodiode. The APD would give a lot more signal, but needs
a high-voltage adjustable supply and protection circuits. A pin
detector will need a lot of gain, a fast TIA or a MMIC or something.

1 ns risetime isn't real hard.

You can buy integrated 150-600 MHz pin-tia widgets, usually in a small
TO-can with a flat or a ball lens, for $20 or some such. Google
"pin-tia". Try Lasermate and Appointech too. Most pin-tia's have AGC,
which trashes prop delay when it kicks in.

The really fast stuff is GaAs, but it's blind at visible wavelengths.

I may have a couple of 850 nm pin-tia gadgets around here that I can
spare... I'll have to look in my box of samples.

Oh, Newark stocks the Optek parts, and their silicon pins are OK.
Apply lots of bias voltage, 24 maybe, to speed them up. Their vcsel
lasers suck.

John

 

[Tom Bruhns]

Photodiodes have substantial shunt capacitance which limits their
response time, especially with weak signals. That's why you don't see
any infrared LAN's-- the signals are too weak and low bandwidth to get
very far.

The usual trick is to use FM instead of AM -- this allows you to
extract the echo difference frequency, which is at a manageable rate--
low megahertz instead of several GHz.

Seems to me what I read in the HP Journal about an HP
distance-measuring transit, back when they were in that business, is
that they used sinusoidal AM modulation at discrete frequencies. You
can then take a sample of the transmitted light locally and compare the
phase of the modulation with the return signal's modulation phase.
There's an ambiguity of the number of cycles, but if you do that at
just a few carefully chosen frequencies, the ambiguity can be resolved.
Then for accurate distance measurement, you must compensate for the
medium through which the light is propagating: humidity and
temperature of the air are important, if it's air, for example. There
are various ways to compare phase accurately to effective resolution of
a few picoseconds, I'm sure, given some time to do averaging, and I
wouldn't be surprised to learn that it can be done quite a bit better
than that. In fact, the averaging gives a pretty big advantage to the
modulated continuous signals, over a pulsed system that can only handle
relatively few pulses per unit time.

Cheers,
Tom

 

[vasile]

Hi - For my senior project at my university, I'm working on a project
where we are sending out a pulse of light with a red laser, and timing
how long it takes for the signal to return. Timing the signal is fairly
easily accomplished with an Acam TDC-GPX chip which gives us 10ps
resolution. We are using a high speed laser driver chip which gives us
approximately 40ps rise and fall times on the laser pulse. The issue is
in our detector circuit. Looking at traditional photo diodes, they all
seem to take in the nanosecond range at a bare minimum to react to an
incoming signal. So my question is this: Is it possible to build a
circuit that can detect incoming red laser light that will send out a
3.3 or 5V pulse (differential or single ended) in less than a
nanosecond? Ideally I'd like to see it in the picosecond range - under
100ps would really be ideal. If there is always the exact same delay
(as in a 10ns +- 100ps delay for the output signal) - that would be
just fine. We just need the accuracy.

If it is possible, can anybody give me some pointers as to how to go
about doing this? I should mention that ideally we'd like to keep costs
as low as possible, as this is a student project.

Nice project. Why don't you use the PIN detector used for laser beam
monitor inside a
laser diode from a CD reader/writer ?
You want to keep costs low, but what tool do you have for measuring 100
pS range ?

success,
Vasile