Yes in certain application we need a lot of gain and dynamic range.

For example, the diffusive-eflective optic sensor uses a photo-diode to 
recive the light reflected by the diffusive target. The emitter is often Red
LED or IR LED. Depends on the distance and the color of the target, 
the photo current can be from 10nAs to 10uA or even wider range and
need to be detected. The sensing distance can be several meters.

So the overall trans-impedance will need to be up to 30Meg Ohm and 
the pre-amplifier may have a gain of up to 1Meg Ohm and need to 
have a dynamic range of 60db or higher. And it is a pulse with width as 
low as 1us (typical 1.5us to 4us depending on the trans-impedance 
amplifier used). A simple transistor or normal opamp based 
trans-impedance will only work with pulse width above 3us and their
settling time is often the problem. 

This high gain will also post big problem for EMC. Normally we need
metal-can to protect the pre-amplifier. Ambient light will also cause
the pre-amplifier to be saturated.

Regards,
Xiaofan


On 9/16/05, Bob Blick <bblick@sonic.net> wrote:
> Xiaofan writes:
> > Actually the speed limit of normal application is more on
> > the trans-impedance amplifier and Photo-detector. We never
> > need to worry about the LED speed (pulsing at 4us or less,
> > even less than 1us at high current >100mA and low duty ratio).
> > Photo-transistor is pretty slow but photo-diode is much
> > faster (typical rise and fall time 20ns). It will be
> > quite hard to develop a low cost >1MHz **TRANSIMPEDANCE
> > BANDWIDTH** trans-impedance amplifier though. Just take
> > note that TRANSIMPEDANCE BANDWIDTH << GAIN BANDWIDTH.
> 
> Sounds like you're reading from an Analog Device's spec sheet. However,
> you might have noticed how slow the IR LED response was, and it was not
> the photodiode in that case(see the results for the bluish-white LED,
> which has similar amplitude). In fast optocouplers I'm sure a different
> type of IR LED is used, but this is a wireless-remote type of LED I was
> measuring.
> 
> Transimpedance amplifiers for photodiodes are easy enough and fast enough
> unless you need a lot of gain and dynamic range. I designed a cascoded one
> a while back that was good in all those respects, but like most high
> frequency amps it wasn't so much expensive as sensitive to layout.
> 
> But I have peaked my own interest, and sometime I'll do another round,
> this time with an amp, an expanded horizontal scale, and normalize the
> output amplitude.
> 
> I'd also like to quickly move the white LED and monitor it with a
> spectroscope to see how time affects the spectrum. But it would be nice to
> produce some metrics and I don't really have accurate tools for that.
> 
> Thanks everyone for all the feedback.
> 
> -Bob
>

-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist