Transimpedance bandpass w/o inductors?

Discussion in 'Electronic Design' started by Asa Cannell, Sep 25, 2004.

1. Asa CannellGuest

I am trying to design a transimpedance circuit for very high
sensitivity. I have pulled a circuit directly out of the OPA129
datasheet. Its a basic transimpedance configuration using a 1Gohm
resistor for feedback. I have built the circuit and am very impressed
by how sensitive it is. However, it is useless in any kind of ambient
light whether static (sunlight) or dynamic (flourescents at 60hz). I
need a way to block low frequency (DC to 60Hz) signals.

I thought about putting an inductor in the feedback loop, but I don't
think they make inductors with an impedance equal to a 1Gohm resistor
at around 1khz.

I tried putting a capacitor in series with the input, and this blocks
DC, but the photodiode will easily saturate in sunlight or other
static light sources.

I was thinking there might be some way to make a bandpass
transimpedance amplifier and maintain my 1Gohm feedback resistance,
without using inductors. Anyone done this?

I am using this for outdoors flouroscopy (of oil), it will be facing
concrete and/or fairly level water in bright sunlight, so it needs to
be very immune to strong DC signals.

The signal will be at ~1khz. (its a pulsed uv light)

Asa

2. John PopelishGuest

I would try putting a noninverting integrator or band reject filter in
parallel with the feedback resistor, perhaps with a current output.

3. Jim ThompsonGuest

1GOhm? I have a couple of those between my ears ;-)

[snip]
The only way to avoid sunlight type biasing is to utilize a (low-pass)
current feedback loop to null the "DC" component.

Do you really need the 1GOhm effective transconductance? Maybe do it
in several stages separated by high-pass (active) filters.

...Jim Thompson

4. John LarkinGuest

Why no inductors? I'd think that dumping the pd current into a
parallel-tuned LC would be ideal; that would extract the signal but
let arbitrary amounts of DC flow without saturating any signal stages.

OK, no inductor: you don't need 1G feedback. Go with a value that
sunlight won't saturate, then highpass or bandpass filter to let the 1
KHz through, then more gain as needed.

A UV-passing, narrowband optical filter would be helpful, and a
synchronous detector later on will optimize s/n.

John

5. Rene TschaggelarGuest

The whole setup is called lock-in and is based on extremely
high gain for the ~1khz. Meaning you rather have multiple
filter stages, ahem bandpass filters with 1kHz passband
each. Make sure, the ~1khz are stable and accurate and
you can narrow the noise with steep filters. A common
Q value for the ~1khz bandpass of 1000 up seems reasonable.

A common practise in physics, BTW.

Rene

If you are unfamiliar with lock-in amplification google for lockin amp
explanation

That turned up a microcontroller implementation for measuring low
reistance and the manual for a commerical lock-n amplifier. Both had
reasonable outlines of the process.

Another thought, if the frequency of light coming from the flouroscopy is
reasonably narrow (I would expect it to be) an optical badpass filter in
front of the detector should also improve your signal to noise ratio.

Robert

7. Product developerGuest

Flourescents lights also put out a lot of 120 hz
Forget all the filtering and just a Schott or similar pass band
optical filter centered on your source's wavelength or use a high
quality low pass optical to cut off everything above your source's
maximum wavelength of interest.

If you are pulsing your source are you also using a reference from
your modulator or chopper to provide a reference for lock-in or
synchronous detection and recovery of the incident light? If so you
will inherently remove most all of any residual artifacts from indoor
or outdoor sources.

I have designed an oil contamination system for a client years ago but
it was a transmissometer-based design looking at a specimen sheet with
the oil under test in assigned locations on a sheet of absorptive
analytical paper. Based on absorption at proprietary wavelengths the
oil's level of contamination was measured.

There is considerable prior art regarding devices that look at oil
using spectrophotometry in both transmissive and reflective
geometries.