I'm quite sure Jinx solution is the simplest one. You need a negative
supply voltage.
Even if you find a rail to rail input-output OA, near 0mV input will
gave you some errors.
The weird solution is reference at 4.5V, PIC supply at 5V and AD range 0-3V.
What if you'll use internal reference, and a stable 3V supply for the PIC ?

On 5/5/07, Rolf <learr@rogers.com> wrote:
> Hi all, I have an EE problem that I would like help solving. I have a
> Li-Ion charging chip (LTC1733) that, as one of it's outputs, provides a
> voltage proportional to the charge current. In essence, in my
> application, it is a 1V/A ratio. It will put 1V on the output if it is
> charging @ 1A, and 10mV for 10mA charge, etc. The full-range output of
> the device is 1.5V indicating the maximum 1.5A charge current. This is
> practically a DC signal, there are no bandwidth concerns.
>
> In addition to the charger, I have a high-precision 4.5V reference
> (REF194 from Analog Devices). This is used as the VRef+ of the PIC's A/D
> peripheral. The rest of the system runs at 5V
>
> Currently I am using a MCP6044 op-amp to multiply the signal by 2, to
> produce a better full-range 0-3V). I am considering a 3x multiplication
> to use the full range of the A/D, but the 3V range is sufficient for now.
>
> My problem is my lack of experience with op-amps at low input voltages,
> combined with a general lack of experience with electronics... I don't
> know the right tools for the job...
>
> The problem: at low signal voltages the OpAmp's limitations cause very
> inaccurate readings. It seems that anything less than about 50mV (a
> charge current of 50mA) seems to be below the OpAmp's 'low rail'.
> Setting it up as a voltage follower I put 10.6mV on the non-inverting
> side, I get 4.3mV on the output. Obviously, my DMM may be of dubious
> accuracy at this range, but it is consistent.
>
> I tried to move the signal out of the low-rail area of the opamp by
> 'subtracting' it from the 4.5Reference voltage using a 'classic'
> difference op-amp layout. But, because the signal impedance is
> relatively high, neither summing nor difference mechanisms were able to
> get it out of the rail areas of the opamp without impacting the signal
> itself.
>
> Fundamentally the charge currents below 50mV have very substantial
> errors. Further, it seems that neither amplification, summing,
> difference, not any other op-amp tricks I can find on the web or H7H AoE
> can help with this (using the MCP6044 opamp). One option would be to
> just ignore anything less than 50mA/50mV as 'trickle' charge, and go
> from there, but I do want to try to get this right (it is a hobby, and
> learning is a major motivator).
>
> One solution may be to use a negative voltage on the op-amp, but I
> currently do not have one available, and I am reluctant to add one.
> Another option may be to use a chip that is more successful in this range.
>
> So, two questions.... what would be a good way to sample a DC voltage in
> the 0-1.5V range from a high impedance source paying particular
> attention to the areas between 0 and 50mV?
>
> Are there better "rail to rail" opamps for this task than the MCP604x
> series? (I am using other op-amp units in the 6044 package for other
> purposes, not just the charge measurements).
>
> Thanks for any hints, tips, pointers, and corrections.
>
> Rolf
> --
> http://www.piclist.com PIC/SX FAQ & list archive
> View/change your membership options at
> http://mailman.mit.edu/mailman/listinfo/piclist
>
-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist