Leica laser range finder , someone opened it?

[Yannick]

Did someone opened a Leica laser range finder (one of their new
models)??? , i wonder if they use optoelectronic mixing and what kind
of APD they use for detection. They get +-1.5mm resolution over
100meter. I am thinking of buying one just to see and measure how they
do it...

I managed to do the optoelectronic mixing with good results, the
limiting factor seems to be the thermal noise of the resistor used to
convert the photocurrent into voltage before the voltage amplification.
I use the C30902E APD agt 217V reverse bias, with 5K resistor (its a
smd pot and 5K gave me the best S/N ratio), after this there a couple
cap to reject daylight dc current wereafter there is an OPA656
amplifier with a gain of 2600 (non inverting amp) wereafter an OPA177
with an additional gain of 5 wereafter a LP filter.

In theory the resistor at the input should be as high as possible to
have highest S/R ratio (noise increases with root and signal lineair
with R value) but then biassing of the APD becomes a problem...

 

[colin]

Did someone opened a Leica laser range finder (one of their new
models)??? , i wonder if they use optoelectronic mixing and what kind
of APD they use for detection. They get +-1.5mm resolution over
100meter. I am thinking of buying one just to see and measure how they
do it...

I managed to do the optoelectronic mixing with good results, the
limiting factor seems to be the thermal noise of the resistor used to
convert the photocurrent into voltage before the voltage amplification.
I use the C30902E APD agt 217V reverse bias, with 5K resistor (its a
smd pot and 5K gave me the best S/N ratio), after this there a couple
cap to reject daylight dc current wereafter there is an OPA656
amplifier with a gain of 2600 (non inverting amp) wereafter an OPA177
with an additional gain of 5 wereafter a LP filter.

In theory the resistor at the input should be as high as possible to
have highest S/R ratio (noise increases with root and signal lineair
with R value) but then biassing of the APD becomes a problem...

Hi,
I used the same APD device but I find it works best as a mixer closer to or
above vbr (~250v for mine) where the gain/voltage curve is much steeper,
of course you need a bit of careful control (and current limit), the amount
of extra noise from the apd at low light is a good indication of the
operating point on the curve.
what frequency and level of modulation have you got on the apd ?
and what is the IF frequency ?
low frequency conversion is problomatic becuase the mixing doesnt reject low
frequency light changes.
the main problem I am finding now is that at >1ghz I cant fully modulate the
cheap laser diode I have and so the constant light cuases more shot noise.
Im trying to get it to go at ~2ghz (and looking for a better laser)

I use a discrete DG mosfet to amplify the signal from the APD with a gain of
~20 (using a pnp current source and 1k load impedance), a sheilded inductor
provides a DC path and tunes out the capacitance - you actually need a
capacitor to ground to provide an ac path for the APD bias modulation
current, a small amount of resistive feedback could flaten out the
phase/frequency response but actually this is not critical.
a std 455khz bandpass filter filters out the IF and a subsequent op amp
gives x100 wich gives a very good level to a quadrature demodulator, this
measures to have aproximatly 1nv/hz^-2 noise when the input is shorted, with
the APD removed the noise is mostly interference, it easily saturates with
an unfocused laser shone at the APD from a short distance away.

I find the noise is purly shot noise from the apd wich increases with
increasing light, when totaly dark the output is very quiet.
Therefore I think something is wrong if your amplifier/resister noise is
greater than your apd noise, I had a 10k resistor initialy but like you
found DC bias current was too high for this, an inductor needs to be
shielded well, a good alternative is to use a common base npn transistor to
provide a DC feedback path.

I have seen a different type of laser range finder wich uses interference
yet still measures absolute distance, it uses a dual cavity arangement where
the light from the target is reflected back into the laser and an external
cavity is modulated by a piezo element. The internal cavity is also affected
by the distance to target and as the variable cavity length is changed it
also affects the operating wavelength of the laser and so alters the
interference cuased by the reflected light, the resultant change in
brightness due to inteference is actually detected by the photodiode in the
laser diode itself and some correlation method comparing brightness to
external cavity length is used to determine absoulute actual distance down
to a few nm.

Ive no idea wich method the Leica uses, it sounds very usefull.

I still find it interesting enough to drive me to see how fast I can get it
going with optoelectronic mixing.

Colin =^.^=

 

[Yannick]

Hi,

Hi Colin

I used the same APD device but I find it works best as a mixer closer to or
above vbr (~250v for mine) where the gain/voltage curve is much steeper,
of course you need a bit of careful control (and current limit), the amount
of extra noise from the apd at low light is a good indication of the
operating point on the curve.

Yes i was planning to do this but first wanted to get rid of the noise
somehow...

what frequency and level of modulation have you got on the apd ?
and what is the IF frequency ?

50Mhz and also 100Mhz worked fine, after this my laser modulator wasnt
fast enough (used a transistor with ft of 100Mhz) , i have build a new
one which will go to 500Mhz. Will try it soon.

The IF is between 10Khz and 100Khz. The amplitude between 0.5V and 3V
(after this not more signal gain seen, about 2V seems best for S/R
ratio) at 217V reverse bias.

the main problem I am finding now is that at >1ghz I cant fully modulate the
cheap laser diode I have and so the constant light cuases more shot noise.
Im trying to get it to go at ~2ghz (and looking for a better laser)

Were can you find lasers with datasheets, i am still using sone
laserdiode from a cheap laser pen

I use a discrete DG mosfet to amplify the signal from the APD with a gain of
~20 (using a pnp current source and 1k load impedance), a sheilded inductor
provides a DC path and tunes out the capacitance - you actually need a
capacitor to ground to provide an ac path for the APD bias modulation
current,

I dont really understand your schematic and how you bias it but an
inductor must be really really large to be high impedant at these low
IF frequencies...

the filtering of the bias frequency happens mostly due to the large
gain and GBP , but i was thinking of just adding a capacitor at the
input to reject it before... dont know if it is going to make a lot
difference...

I find the noise is purly shot noise from the apd wich increases with
increasing light, when totaly dark the output is very quiet.

Ok, thats also the case, when its dark the noise is a factor 2 less but
still about 100mV after a total amplification of 5K * 5000 voltage
gain...

My signal amplitude on a white wall at about 3 meters is 3.5V pk-pk.
but AC laser power is still not optimal... maybe a factor 2-3 more is
possible

If i short input, noise is all gone...(mostly ) if i remove
photodiode it doesnt do much,just a factor 2...

Therefore I think something is wrong if your amplifier/resister noise is
greater than your apd noise, I had a 10k resistor initialy but like you
found DC bias current was too high for this, an inductor needs to be
shielded well, a good alternative is to use a common base npn transistor to
provide a DC feedback path.

A common base is very low impedant , the signal would source away, i
think i do not understand the schematic

I have seen a different type of laser range finder wich uses interference
yet still measures absolute distance, it uses a dual cavity arangement where
the light from the target is reflected back into the laser and an external
cavity is modulated by a piezo element. The internal cavity is also affected
by the distance to target and as the variable cavity length is changed it
also affects the operating wavelength of the laser and so alters the
interference cuased by the reflected light, the resultant change in
brightness due to inteference is actually detected by the photodiode in the
laser diode itself and some correlation method comparing brightness to
external cavity length is used to determine absoulute actual distance down
to a few nm.

Ive no idea wich method the Leica uses, it sounds very usefull.

we will see, will buy one in the coming weeks...

I still find it interesting enough to drive me to see how fast I can get it
going with optoelectronic mixing.

Me too, think 2Ghz is the max due to diffusion current speed is limited
to these periods...

Yanick

 

[colin]

Hi Colin


Yes i was planning to do this but first wanted to get rid of the noise
somehow...

I'd just increase the apd gain untill your resistor noise gets lost below
the apd noise.

50Mhz and also 100Mhz worked fine, after this my laser modulator wasnt
fast enough (used a transistor with ft of 100Mhz) , i have build a new
one which will go to 500Mhz. Will try it soon.

500mhz gets interesting as even very small wires have significant inductive
impedance to be a nusiance.
you can even use a TV to see what frequency is !
im using several 3ghz+ MMICs to drive the laser and will probably use them
for the APD bias too.

The IF is between 10Khz and 100Khz.

Thats quite low, high eficiency lamps emit lots of noise in this range wich
will go straight through the mixer.

at 500mhz its quite dificult to synthesize the LO and RF frequencies so
close together, DDS will goto ~ 200mhz

I had a lot of problem with 2ghz VCO <1mhz apart.

The amplitude between 0.5V and 3V
(after this not more signal gain seen, about 2V seems best for S/R
ratio) at 217V reverse bias.

that sounds about right, maybe your diode has a much lower vbr than mine.

Were can you find lasers with datasheets, i am still using sone
laserdiode from a cheap laser pen

I cant find good datasheets, im stil using a cheap laser pointer type diode
too.
I might go for one of the VCSEL types.
I use http://www.roithner-laser.com/

I dont really understand your schematic and how you bias it but an
inductor must be really really large to be high impedant at these low
IF frequencies...

2mh for 455khz wich is quite large, hard to find a well sheilded one.
10khz becomes unpractical.
Im probably switching to 6mhz IF so a much smaller inductor and use a TV IF
sound trap BP filter.

the filtering of the bias frequency happens mostly due to the large
gain and GBP , but i was thinking of just adding a capacitor at the
input to reject it before... dont know if it is going to make a lot
difference...

ah no what I meant was that you need a path for the bias frequency to go
through apd and down to ground, otherwise you end up with no bias frequency
voltage accross the APD.

Ok, thats also the case, when its dark the noise is a factor 2 less but
still about 100mV after a total amplification of 5K * 5000 voltage
gain...

My signal amplitude on a white wall at about 3 meters is 3.5V pk-pk.
but AC laser power is still not optimal... maybe a factor 2-3 more is
possible

If i short input, noise is all gone...(mostly ) if i remove
photodiode it doesnt do much,just a factor 2...

sounds like the noise from your resistor is quite high, what bandwidth ?
maybe its rf pickup.
does it agree with calculations ? if its a factor of 2 below APD noise then
its reasonable.
the calculations for apd noise at a given bias voltage, temp AND incident
light are quite complicated but given on the datasheet,
I managed to make my measured noise agree with it. I get close to maximum
demodulation efficiency.
the resistor noise is negligble by comparison.
If you use synchronous demodulation then the noise disapears through the
floor anyway.

A common base is very low impedant , the signal would source away, i
think i do not understand the schematic

With the transistor biasing method it just acts like a current source with a
low frequency negative feedback to keep the gate voltage constant. the
collector goes to the APD and (not the emitter) so is high impedance, the
apd DC current is taken away through the emitter with a resistor as part of
the low freq feedback path.

another simpler alternative just for removing dc bias is

apd
___|____to amp
| |
100k |c
b|_|/ npn
| |\
| |e
C=10n R=1k
|__|
gnd

it looks like a ~100k resistor at AC but 1k at DC

we will see, will buy one in the coming weeks...

Me too, think 2Ghz is the max due to diffusion current speed is limited
to these periods...

Yanick

My vco goes from 800mhz to 2.2ghz, gets hard to make a simple VCO with such
a wide range above this.
the limit of the APD is not difusion current limited but RC time constant
wich is very small.
as for the laser diode im not sure, as its effectivly an oscilator that has
to start up.

Colin =^.^=

 

[Yannick]

I'd just increase the apd gain untill your resistor noise gets lost below

the apd noise.

Good idea, but i hope it doesnt really brake down, will first meausure
dc current through it, maximum is 1mA i believe...

500mhz gets interesting as even very small wires have significant inductive
impedance to be a nusiance.
you can even use a TV to see what frequency is !
im using several 3ghz+ MMICs to drive the laser and will probably use them
for the APD bias too.


Thats quite low, high eficiency lamps emit lots of noise in this range wich
will go straight through the mixer.

Yes i know, 455Khz should probably be the best...

I am now using 2 old Shwarz(i dont remember the name exactly)
frequency generators who can go up to 1Ghz with 1.5V pk-pk amplitude

that sounds about right, maybe your diode has a much lower vbr than mine.

That would be strange but who knows...i will try up to 250V , first
have to check datasheet again but i believe it was 250V, espescially
the c30902S APD, it can operate in geiger mode, but i only have the E
type... i also saw some interesting APDs from hammatsu with much less
noise , less reverse bias voltage and same gain... (S9717 APD). Also
the C30902S-TC is interesting as it is cooled , but i don't know the
prize, will be too high probably!

thats indeed the main criteria, as much AC power transmit as
possible..

I cant find good datasheets, im stil using a cheap laser pointer type diode
too.
I might go for one of the VCSEL types.
I use http://www.roithner-laser.com/

aha , interesting, will look into them

2mh for 455khz wich is quite large, hard to find a well sheilded one.
10khz becomes unpractical.
Im probably switching to 6mhz IF so a much smaller inductor and use a TV IF
sound trap BP filter.

yes but then your amplification will drop... hows your sensitivity, can
you measure up to 100 meters...? i am intending to use the DSP blackfin
from analog devices to do all the measurements and averaging soo i can
easily average out the noise...

ah no what I meant was that you need a path for the bias frequency to go
through apd and down to ground, otherwise you end up with no bias frequency
voltage accross the APD.

Ofcourse Colin, but i meant at the input of the amplifier, i bias the
diode at the other side, with a 56 ohm resistor between the 217V DC
voltage and the AC modulation signal (with high voltage smd cap in
between).. then the other node of the APD is at the input of my amp
with a resistor to gnd... here i am going to put a cap in parrallel
with this resistor to filter out the high modulation frequency..( but
the other node still has this freq soo the mixing still hapens).I was
wondering if i could use a 10Meg resistor then the parasitic cap would
be high enough to filter everything out above the IF freq... but then
the dc biasing becomes a problem... but still if you could then bias
the diode at 500+ voltage and then have a 300 V on this resistor (due
to the dc current through the diode) , the S/R ratio would be much
higher than now... (if resistor noise is the dominant noise factor
ofcourse)

sounds like the noise from your resistor is quite high, what bandwidth ?
maybe its rf pickup.

the bandwidth is just white noise... it follows exactly my filter... i
did a FFT (fast fourier transformation with my digital scope) and it
looked like white noise)... i also put a speaker at my output so i
could hear the signal (till 14Khzmy ears could handle) and there
was no Rf pick up, it sounded like white noise and only my signal was a
beep...

does it agree with calculations ? if its a factor of 2 below APD noise then
its reasonable.
the calculations for apd noise at a given bias voltage, temp AND incident
light are quite complicated but given on the datasheet,
I managed to make my measured noise agree with it. I get close to maximum
demodulation efficiency.
the resistor noise is negligble by comparison.
If you use synchronous demodulation then the noise disapears through the
floor anyway.

I have to do the calculations for shot noise, but measurements dont
lie... but upping the reverse bias voltage can maybe help, and then
lower the gain of my amplifier...

synchronous demodulation ?

With the transistor biasing method it just acts like a current source with a
low frequency negative feedback to keep the gate voltage constant. the
collector goes to the APD and (not the emitter) so is high impedance, the
apd DC current is taken away through the emitter with a resistor as part of
the low freq feedback path.

another simpler alternative just for removing dc bias is

apd
___|____to amp
| |
100k |c
b|_|/ npn
| |\
| |e
C=10n R=1k
|__|
gnd

it looks like a ~100k resistor at AC but 1k at DC

Damn, This is a very good idea, its actually the same as a current
mirror, were the APD Dc current determines the set point of the
transistor and because the input impedance of the common collector
(seen from the base) is much higher then the 100K, the Collector
current is the same as the APD's DC current... and for AC the gate
voltage is 0 ... very good idea...i wonder if it is not better to use a
NMOS in stead of the bipolar transistor as the input impedance is much
higher and the need for the 1K is not nessecary anymore... (hmmm if vds
does not decrease too much ofcourse due to signal swing) which improves
the noise (no thermal noise of resistor but probably not an issue
because it is 100 times smaller)...or am i missing something here?

I thaught of something like this before but then with a cascoded
current mirror (to increase output impedance even further) but then i
biased this current mirror with an external resistor which gave APD
bias problems when the DC current of the APD was much different, didnt
think about the feedback to the drain/collector soo the APD's dc
current is used.

My vco goes from 800mhz to 2.2ghz, gets hard to make a simple VCO with such
a wide range above this.

Which VCO ic do you use? Does it have sweep capabilities, i am planning
to try FMCW principle also (frequency modulated continious wave) , then
good resolution qnd sensitivity can even be possible with normal PIN
photodiode..

the limit of the APD is not difusion current limited but RC time constant

Ah thats possible... but i remember that diffusion current speed is
also limited not much above these frequencies...

as for the laser diode im not sure, as its effectivly an oscilator that has
to start up.

indeed... i was also planning todo speed measurements with doppler
effect, in theory much simpler then distance... but the problem is that
the frequency counting must be really precise and modulation frequency
very high...(at 1Ghz ist about 1Hz for every 0.5feet/sec)
soo its a pity

thanks for the good info again!

Yannick

 

[colin]

Good idea, but i hope it doesnt really brake down, will first meausure
dc current through it, maximum is 1mA i believe... .....
That would be strange but who knows...i will try up to 250V , first
have to check datasheet again but i believe it was 250V, espescially
the c30902S APD, it can operate in geiger mode, but i only have the E
type... i also saw some interesting APDs from hammatsu with much less
noise , less reverse bias voltage and same gain... (S9717 APD). Also
the C30902S-TC is interesting as it is cooled , but i don't know the
prize, will be too high probably!

As long as the current is limited to the maximum it will be ok.
mine has taken quite a bit of abuse and survived.
too much voltage and it just latches on for a while.

How many times have you accidently touched the hot end of the apd ?
its good to have <<1ma current limit.

I think the one we are using wil be fine, at least its readily avaialable if
expensive.

yes but then your amplification will drop... hows your sensitivity, can
you measure up to 100 meters...? i am intending to use the DSP blackfin
from analog devices to do all the measurements and averaging soo i can
easily average out the noise...

I have no optics at all atm, just shining the unfocused laser to the apd
from a about a foot away.
the inductor should keep the same sensitivity even at 6mhz, although it wil
end up being more narrowly tuned.

Ofcourse Colin, but i meant at the input of the amplifier, i bias the
diode at the other side, with a 56 ohm resistor between the 217V DC
voltage and the AC modulation signal (with high voltage smd cap in
between).. then the other node of the APD is at the input of my amp
with a resistor to gnd... here i am going to put a cap in parrallel
with this resistor to filter out the high modulation frequency..( but
the other node still has this freq soo the mixing still hapens).I was
wondering if i could use a 10Meg resistor then the parasitic cap would
be high enough to filter everything out above the IF freq... but then
the dc biasing becomes a problem... but still if you could then bias
the diode at 500+ voltage and then have a 300 V on this resistor (due
to the dc current through the diode) , the S/R ratio would be much
higher than now... (if resistor noise is the dominant noise factor
ofcourse)

you need to have a low impedance path to ground at the modulation frequency
at the cold end of the APD otherwise you end up with no RF voltage acros the
apd, its no use to have the modulation voltage at both ends of the apd it
needs to be accross it.

synchronous demodulation ?

best look it up, its the best way to recover a signal buried well below a
noise floor.
you just modulate it with a signal of the same (IF) frequency, the output is
then a dc voltage proprtional to phase.
the noise reduction arises becuase you can reduce the bandwidth of the
subsequent filter to as low as you want.
(but response time falls)
if you use quadrature demodulation it gives you full 360' and also gives you
the signal strength.

Damn, This is a very good idea, its actually the same as a current
mirror, were the APD Dc current determines the set point of the
transistor and because the input impedance of the common collector
(seen from the base) is much higher then the 100K, the Collector
current is the same as the APD's DC current... and for AC the gate
voltage is 0 ... very good idea...i wonder if it is not better to use a
NMOS in stead of the bipolar transistor as the input impedance is much
higher and the need for the 1K is not nessecary anymore... (hmmm if vds
does not decrease too much ofcourse due to signal swing) which improves
the noise (no thermal noise of resistor but probably not an issue
because it is 100 times smaller)...or am i missing something here?

Its actually more like a very slow voltage clamp I think. ie clamped to vbe+
a little bit.
also it actualy looks like an inductor.

Winfield Hill has written a bit on lowest noise current sources,... worth
reading.
If im not mistaken the 1k resistor reduces the effect of the noise of the
transistor, the higher the better
although it will stil work without it.

with a mosfet the output impedance will actualy be lower than a bjt,
although the input impedance is higher
but mosfet noise is higher too.
100k is quite reasonable, you could of course use a darlington or super beta
device.

Of course the common base configuration does away with the base resistor
altogether.

this circuit may interest you ...
http://www.imagineeringezine.com/PDF-FILES/40krvr2.pdf#search="40krvr2.pdf"

Which VCO ic do you use? Does it have sweep capabilities, i am planning
to try FMCW principle also (frequency modulated continious wave) , then
good resolution qnd sensitivity can even be possible with normal PIN
photodiode..

Its a varactor controlled colpits oscillator pushed to the extreme. using
mmic and very low capacitance varicaps and the inductor is almost purely
parasitic just a very very short arch of flattened wire. VCO=sweepy.

constant

Ah thats possible... but i remember that diffusion current speed is
also limited not much above these frequencies...

no AFAIK the bias voltage doesnt rely on diffusion current at all to acheive
its effect.

thanks for the good info again!

welcome

Colin =^.^=