> -----Original Message-----
> From: Jeethu Rao [SMTP:jeethur@SANCHARNET.IN]
> Sent: Tuesday, October 22, 2002 9:22 AM
> To:   PICLIST@MITVMA.MIT.EDU
> Subject:      [EE]: Maximizing ADC Range
>
> Hi,
>         I'm working on improving(actually reimplementing) an existing
> analog
>         Circuit using an MCU. Its  for a grain color sorter(used for
> sorting
>         rice) I'm using an AVR 8515 for this. That does not exactly
> constitute
>         the problem.
>
>         The problem is the application. It is to detect the color of a
> singular stream of rice, which passes through an illuminated chamber.
>         Two photodiodes with a dichroic beam splitter detect colors of
> the
>         grains against a        distinctly colored background plate, In
> the analog
>         ckt, lighter shades     give a higher voltage and darker shades
> lower
>         the signal voltage. It  works well in analog, with some 14
> OPAMPS
>         (LM725s and 741s)       For each        photodiode. So, I have
> an average signal
>         on the photodiodes (for instance, 2.5 volts) with all the lamps
> lit
>         up. A black grain       gives a lower signal (say 2.2 volts),
> while a
>         normal grain will not   get as low as that, might get a little
> higher
>         if it's a little lighter        than other grains (In the coffee
>
>         industry, we call such grains Bleaches, but there's no
>         nomenclature or purpose for that in the rice industry), but
>         never will the signal rise above 3 volts or so.
>
>         Now, if I set the ADCs ref to 3.0 volts so that it reads 1023
> (AVR    AT90S8535 has a 10 bit adc like a PIC16f877), I'll be wasting
> 2/3rds  of the ADCs range       below 2 volts. And the ADC on 8535
> doesn't have a  VRef- input. And even if I use an adc with both Vref+
> and  Vref-, The         brightness of the CFLs used in the illuminated
> chamber tend to         deteriorate over time, so the mid point will not
> always be the same,
>         it shifts over time. And another problem is that the ADC
>         readings initially will be low, since it takes a little time for
> the
>         lamps to attain full brightness. The solution In the analog
> circuit is
>         to compensate the value with a multi turn pot for zero setting.
> But
>         this is inelegant. My idea is to use two DACs to set the Vref+
> and
>         Vref- once the mid point has been determined. I'm not sure if
> this
>         Approach works well. A simpler approach would be to use a higher
> res
>         ADC (~16 bits), but I need an SPI ADC,(low pin count) and I'm
> using
>         Microchip MCP3201 (12bits) for this. I guess there is a pure
> analog  solution to this        problem , just using OPAMPs. Could you
> folks please    enlighten me in this regard ?
>
>
This is a simple dynamic range problem, you either cover a large range with
poor resolution, or a smaller range with better resolution.  What you need
to define is what resolution is acceptable for your purpose.  You can get
16bit ADC's with an SPI interface, but they are not at all common and most
of them use a DSP type interface rather than plain SPI.  Take a look at the
Linear Technology LTC1865.  Alternatively, you could consider oversampling
and averaging to give you more effective bits.  You need to sample 4^n times
where n is the number of extra bits you want.

The beam splitter arrangment sounds very simmilar to an arrangement we use
to wavelength lock tuneable lasers in the fibre optic communications
industry.  The difference bwteen the two diodes gives wavelength information
(i.e. colour), and the sum of the two gives us optical power.  If this is
the way your system works, then you can automaticaly compenstae for
differening light levels by using the sum term to scale the difference term.
This would mean no manual reset is required.

Regards

Mike

--
http://www.piclist.com#nomail Going offline? Don't AutoReply us!
email listserv@mitvma.mit.edu with SET PICList DIGEST in the body