[Laser] Cloud bounce new optical receiver.

[Art]

Thanks Yves,


>Hello  all
>After many comparing measurements we have optimised the K3PGP RX  type
>especially for cloud bounce.The K3PGP RX is a no feed  back front 
>end and the photo
>response shows a decremented gain versus frequency  with no define flat bandw
>idth area.

This is important. Throughout the years, as I have looked at the PGP 
front end, I've always wondered where the feedback was. This comes 
from digesting op amp based photodiode technical material. Now I have 
the answer, there is no feedback path.

My understanding was that the feedback element in an op amp based 
photodiode front end determines the bandwidth because the feedback 
resistor limits controls the gain by modifying the gain-bandwidth 
product. More gain = less bandwidth and visa versa.



>For example, the native schematic with a  simple photodiode BPW34 and a
>MPF102 FET gives -18 dB gain reducing when the  frequency changes 
>from 10 Hz to 2
>KHz.
>However gain and noise density follow  the same curve profiles up to 100 Hz
>so the signal/noise stays quite good even  at more than 3 KHz.
>The extra gain at less than 100 Hz enhances strongly the  scintillation and
>is a problem for L.O.S. voice com but not in  N.L.O.S.
> >From 100Hz to DC most of FETs give a strong 1/F noise  enhancement.
>New audio FETs can give ultra low noise in VLF.

Interesting.....so, with an op amp, we look for low input bias 
current (which is effectively noise input current) and high frequency 
response. The high frequency response is tamed by using a feedback 
resistor to limit the bandwidth. Since our PGP type front end doesn't 
have feedback, we have to instead search for FET's with low noise 
figures in the audio range (because we don't have feedback, so we 
don't need/want high frequency gain)?

>
>The  new front end schematic remains extremely simple.
>Only some resistors values  change, an output emitter follower stage is added
>but the essential of noise  reduction comes from the transistors choice.
>All these transistors have an  equivalent input noise lower than 1 nV /
>s.r.Hertz from 10Hz to 10  KHz.
>Because the 30 pF FET input capacitance, this RX is particularly well
>matched for the 50 to 300 pF PN or PIN silicon photodiodes and for 
>the low speed
>VLF FFTDSP communication programs like JASON, WOLF-GUI or K0SM Laser  Scatter.
>All the tests have been done to 75 Hz with a very low current  modulated LED
>beacon (without optics on receivers).
>An 20 dB signal/noise  with Spectran (at 0,084 Hz resolution) with this
>receiver gives less than 5 dB  signal/noise on a VK7MJ RX in the 
>same conditions
>(at 75 Hz).
>The schematic  is in Photo Album "F1AVY F8DO cloud bounce" on
>Optical_DX at yahoogroups or can be  loaded at:
>http://pageperso.aol.fr/F1AVYopto/opto/ultimepgp7.jpg
>The  2SK170 and the SSM2210 are from CONRAD and RADIOSPARE in France.


If we were serious about preserving every single partial db of noise 
figure (without concern for $$), would these transistors be the ones 
to use???? Or are there better choices, although more pricey?

Yves, did you spice model this?? Also, if you did, did you by any 
chance use LTSpice?

I wonder if you have done any comparison between the OPT301 and your 
modified PGP rx?

This looks like the type of rx that would lend itself to smt 
fabrication techniques, shorter leads and a smaller construction make 
it easier to shield and improves the performance. I bet this would be 
a popular do it yourself project if someone did a PCB layout or offered a kit.

Regards,

Art