app examples of 4 stage op-amp

Specifically, I am interested in the TLE2064, a single IC
with 4 op-amps. The application is to amplify and quantify a weak
signal from an IR PIN photodiode using 0.82 pF caps as feedback
(i.e., signal treated as current source). My single stage
works electronically in testing with a stronger light source,
but does not sufficiently amplify my weak signal.

I was thinking someone might have examples of (or warnings against)
circuits where the 4 stages are wired together. Do I risk losing
linearity or some other problem?

Alternatively, maybe someone knows of specific IR PIN with peak
sensitivity near 880 nm with integrated amplifier I could
experiment with for comparison. The ones I have seen are on/off
switches, which I doubt would work.

Dominic

 

Before giving any advice, I need to understand better what you are
trying to measure. How did you arrive at the integrating capacitance
of .82 pf (a horribly small value)?

The main problem with cascading several opamps is the amplification of
the noise and errors of the first stage, along with the signal.

 

I am trying to measure IR from faint stars. The use of 0.82 pF for the
feedback resistance is rather well known and common replacement for the
very uncommon high resistance values that would otherwise be used, like
around 1-10 GigoOhm.

Perhaps this offers better insights into my question?

Dominc

 

For this application, the single most important opamp spec is input
bias current, since this error adds directly to the signal you are
trying ot measure. The TLE2064 is not a bad opamp in this respect
(see page 9 of the data sheet at
http://focus.ti.com/lit/ds/symlink/tle2064a.pdf ) with a typical bias
current of 3 picoamps. But that current integrated with a .82 pf
capacitance will give a ramp speed of about 3.7 volts per second with
no photo current. You also need a very low leakage method to reset
the capacitor voltage to zero before each integration, and the switch
must have an off state leakage current less than the opamp bias
current if it is not to be an even bigger source of error. It must
also have a lot less than .82 pf of capacitance in order to not
dominate the integration.

The offset voltage is probably the next most important specification
for the opamp, because the diode voltage has to be held very close to
zero volts in order to keep its leakage current below the photo
current being integrated. This is a less impressive spec of as high
as .8 millivolt at 25 degrees C.
Cooling the opamp improves the bias current, but may degrade the
offset voltage to 3 millivolts.

An OPA627B might make a better front end.
http://focus.ti.com/lit/ds/symlink/opa627.pdf
with an input bias current typically 1 pA, and input offset voltage
typically less than 100 uV. But you don't its speed capability, so a
very low bias current amplifier such as an LMC6081
http://cache.national.com/ds/LM/LMC6081.pdf
with typical bias current of .01 pA and offset voltage of 150 uV,
or LMC6061
http://cache.national.com/ds/LM/LMC6061.pdf
with bias current of .01 pA and offset voltage of 100 uV.

Tell me more about your layout, reset mechanism and how you provided a
precise .82 pf of integration capacitance. When we have the front end
as good as you can get it, we will be ready to think about additional
amplification.

 

As I move along, I am working on the web page which describes
this circuit...

Main page is http://www.megspace.com/science/stp/

and this specific project is at...

http://www.megspace.com/science/stp/pmt/sfh2030f.html

I mainly chose the TLE2061 because it is cheap and pin compatible
with some of the other ICs you mention.

Perhaps you know of an easy alternative so as to not re-invent the
wheel, like an IR PIN with an integrated op-amp?


Dominic

 

I would like to give some specific recommendations, but the jpg files
on the second site are not valid, so I cannot see how it is supposed
to be built.

 

I can recommend an alternative to the SFH203FA that has a larger die
area and is cheap ($1) and available to US builders from Digikey. It
is a side view instead of end view part but has a 3X3 mM die instead
of a 1X1 mM die.

http://rocky.digikey.com/WebLib/Photonic Detectors/Web Data/pdbc158.pdf

There are also a couple 2X2 mM end view version more like the SFH203FA
With lens:
http://rocky.digikey.com/WebLib/Photonic Detectors/Web Data/pdbc142.pdf
and with flat front:
http://rocky.digikey.com/WebLib/Photonic Detectors/Web Data/pdbc139.pdf

These bigger dies make it easier to get a poorly focussed and guided
star completely on the die.

They also have much larger area units, but most do not have the
integral IR filter.

Digikey sells several other brands of IR PIN photo diodes, including:
http://rocky.digikey.com/WebLib/Sharp/Web Data/tec_datasheet_pd49pi_481pi.pdf
http://rocky.digikey.com/WebLib/Panasonic/Web data/PNZ323B.pdf
http://rocky.digikey.com/WebLib/Panasonic/Web data/PNZ313B.pdf

 

I am still working on the web page as I develop this project. The
first version, which needs improvement, couples the PIN directly
to the opamp, which presently uses an LMC6001, having replaced
the TLE6401 once it was clear the latter component was detecting
light and not burning up, etc. The LMC6001 has input bias of 25 fA.
I feel OK with this opamp, as far as I can tell, but am not
sufficiently picking up starlight. I think we could use the star
Vega as an example since most people in the world can easily pick
out this star in the night sky, and it is in good position for most.

For the moment, it looks like I can just swap these PINs out as
needed, I chose the SF1020 becuase its spectral response pretty
much exactly suits the design requirements for the I region of
the UBVRI system used in astronomical photometry (peak is 880 nm).
The others I have seen thus far do not have as narrow bandwidth
as the SF1020.

The 0.82 pf feedback cap is the only other component other than
the power supply at this time. I took this from the Jung IC
opamp book I reference on my web page.

Dominic