D. L. Davis wrote:
>
> I built an altimeter with a PIC14000 which I have in bench test now.  I am
> using the 16 bit A/D of the PIC14000 to measure voltage from an air pressure
> sensor.  It is used to control an autopilot, and the PIC runs the autopilot
> servo too.
>
> Everything works great with one big exception.... the PIC is extraordinarily
> sensitive to light!  The logic all functions fine and is unaffected by
> light, but the A/D measurement is strongly influenced by ambient light on
> the chip.  I'm using a JW part now, with tape over the window.  But even
> with tape over the window, I can effect the A/D count by an order of
> magnitude just by shining a flashlight on the PIC from across the room.  I
> have replaced the little red sticker tape that came with the part from
> Microchip with double layers of black tape, which improves the A/D
> stability, but only to about 6 or 7 bits, nowhere close to the 16 bit
> resolution.  If I control ambient light carefully, it still moves around
> randomly in the least significant 9 or 10 bits.
>
> I have noticed that the chip measures its own internal reference with good
> accuracy and stability and is nearly insensitive to ambient light (stable to
> about 14 bits).  This much works great, but the measurement of external
> voltages at the pins is hugely affected by light???? (Yes I have checked
> that the voltages at the pins are stable).
>
> I am hoping that the OTP version without a window will not have these
> problems.  But I am concerned that I will not even be able to get a good
> test done with the JW part.  I haven't yet found a way to get more than
> about 7 bits of stability (much less accuracy) out of this JW part, which is
> totally inadequate for an altimeter or autopilot application.
>
> I have heard of EPROMS having light sensitivity, but the program and data
> memory in this PIC seems to work just fine.  Only the A/D is affected.  Has
> anyone else seen this light sensitivity problem?  Has anyone used this part
> successfully with stability of 12 or more bits?  Any ideas out there?
>
>                                         Dewey Davis


This doesn't surprise me at all.  Silicon is photoconductive, regardless
of whether you are talking about junctions or MOS capacitors (i.e.,
gates).  The degree to which a given function is affected by light would
depend on several factors:

1. Metallization: the Al layer is opaque.
2. Impedance: lower impedance circuits are less affected by leakage
3. Location: circuits nearer the center of the die have a wider field of
view

Whether the A/D is the charge balancing type, current sources, or
resistor ladders, it's bound to have some relatively high impedance
sections.

BTW, ordinary black electrical tape is not opaque to near IR.  The old
silver write-protect labels for 5.25" floppies work well, or, you can
get 3M opaque black crepe tape or something similar.
--

Paul Mathews, consulting engineer
AEngineering Co.
optoeng@whidbey.com
non-contact sensing and optoelectronics specialists