can you add a 3rd stage , to the simulation that takes the 20mv out to say to 200mv for a signal suitable for a/d use, and see what the noise factor looks like , the capacitance loading effect on the 2nd stage ??

the larger diode is mainly in active area , as some that i have looked at , this add capacitance as well , which slows down the signal it can react to the aim is to have a large surface area to ensure as much of the available light hits it , also extra optical lens help both increase the active area and increase the avaiable signal ,

i will need to improve the optics as well , as much as size can allow,

 

re reading this app note
http://www.linear.com/solutions/1785

" at increasing frequencies the capacitance of the photodiode comes into play and the circuit noise gain rises as the 1MΩ feedback looks back into lower and lower impedance. But the 6.8nF comes to the rescue. In addition to the obvious quenching of noise source R3, the 6.8nF capacitor increases the JFET gain to about 30 at high frequency effectively attenuating the downstream noise contributions of R2 and the op amp input noise. Thus the circuit achieves low input voltage noise at high frequency where it is most needed."

i am wondering adding the 6.8nF cap to the the cct " R2 in the simulted cct " over the Jfet should increase the sensitivity of the jfet if the freq is required to be increased ???

 

Also added an offset adjustment.
Is there information in the amplitude of the signal or only the timing of it?

 

at the moment i just have general timing info , i have ordered the parts to see what it does for real , layout / housing etc is critical even for breadboarding , i need to detect the signal to confirm the results

amplitude will vary with the signal strength applied on the diode , but i am expecting the worst as a start point
can you add a 6.8n cap to R2 on jfet and see what it does to the pulses ??
also increase the freq to as fast as it can till you cant see the pulses at 5nA , it be nice to see the fastest pulses this can see
the Jfet is garden type so not sure if that a large limitation as the speed is increased
pm me if you like

cheers

Sheldon

 

The pulse looks pretty go down to 1.5 usec. After that it starts to look a bit like a saw tooth. I haven't been able to get the second circuit to run, but I think they are basically the same. It is just that one takes the voltage off the source with the cap across the resistor in the drain and the other takes it off the drain with the cap across the source resistor, I didn't run the noise because I don't have a diode model with noise. It is probably the biggie.

 

Yep, still 1 mv / na.

 

ok thanks , i think the noise level will be the key , as their is little signal to play with ,
a really nice strong signal to start would be good , but is unlikely
you can see why APD's that well setup can be a big advantage with high gain , over SI photodiodes , but cost a fortune , plus the HV issues that come with them
a lot of real testing to be able to get this to work at 1us without it looking very saw tooth or osc ,

even with the online tool model i cant get it go lower than 1.3us with a 5nA input ,
thats with a output from the 2nd stage at 50ma , noise floor of 5mV , adding an extra 3rd stage is required but the noise is going to be bad

cheers

sheldon

 

You should get 5 mv out of the first stage with 5na of diode current. Did the tool include the jfet?

 

no the tool only allows for a 2 stage amp approach,

most of the op amps have a jfet as part of the input stage to achieve the low noise ,

 

No, My bad. I didn't count the jfet as a stage. I should, but didn't. So yes, you need 3 stages like post 23.

 

well i got the cct working but the noise is high , mainly the layout is on breadboard , i will continue to play with it further

I would also like to try and a the jfet to the OPA380 as a test , wondering if you can model it with the Jfet added to the front end

cheers

sheldon

 

Do you mean the noise?
Let me see if I can find and OPA380 model - or one close.
Do you have any idea what the noise is for the diode?

 

no , i dont think they spec that

i have tried several diodes but the most consistent signal so far seems to be from SFH2400FA

 

I haven't been able to get the 380 to work, but it has about 5X the noise of the LT part and is only 5 volts.
I'm surprised you could get it to run on a breadboard.
You really have to have clean supplies the noise almost passes straight thru.
Can you scope the noise? If it is the same as the supply we can maybe do something.
I added a regular diode to the model to get some noise for it. The result was still only 10nv/sqrthz, which I think means we should only see 5 mv of noise on the 200mv signal

 

yes the 380 is a 5v only unit , as it only needs a 1 resistor , and cap + whole lot of decoupling from the supply it was very easy to breadboard for scope test / measurement ,

ill rebuilt the LT parts onto 2 smd pcb breakout boards today for further testing

i am still waiting on other photdiodes for testing , more so the larger surface area ones

 

How much noise did you have with the current one? Can you post a picture?

 

ill redo the measuments over the weekend and tyr to get a picture from the scope

 

i am seeing a noise from the neg rail , even after smothing caps , unfortunatly i need to use a 12v single supply source only and the negative rail is produced by a neg pump generator cct , apart from that i am seeing that 2nd stage is sentive to high osc until the pot is pulled either to close gnd rail .

testing progressing ,

 

Contact Hamamatsu 617-536-6441 and request a copy of the Photonics Technology Seminar 10/2007. This seminar solves the problem for you.

 

can you elaberate what is covered in the document this or is there a link you can advise , i am talking to hamamtatsu in japan regular on the optic sensors i have requested