ID this PC card

From: Dwight Elvey <elvey_at_hal.com>
Date: Tue Jun 8 18:22:40 1999

ard_at_p850ug1.demon.co.uk (Tony Duell) wrote:
> >
> > The 7109 is an adc, of the dual-slope integrating variety more or less
> like
> > the panel meter IC's Intersil developed and everyone else copied. I
> think

Hi
 There was one thing I remember about using these things
from years back ( 7109's ). This was a problem that depended
on the chip version. It was fixed by some manufactures on some
versions but I don't recall which. You most likely won't
find it in the spec sheets either, I spent several hours
talking to designers to find this one out.
 If you over range, it messes up the value in the autozero
capacitor. You have to wait a long time for this to
settle out again after an over range reading. It was
a real pain for me because I was running a mux on my
input and expected to sample a lot of inputs in a short
time. It was really bad also because even a single LSB
of over range was enough to upset the auto zero capacitor
by as much as 20% of full scale on the next reading.
 The problem was that on an over range, they didn't
continue to de-integrate the main measurement capacitor.
This meant that when it went to the auto-zero phase,
there was still a large signal left on this capacitor.
It was then transfered with gain onto the auto zero
capacitor.
 The fix was to short out the measurement capacitor
on an over range but this still had a small effect on
the auto zero because of offsets in the system.
It would have been better to continue the de-integrate
phase until the circuit detected the zero. This wouldn't
have extended the time much because full scale was usually
close to clipping in most cases, except when one was
doing a wide range of ratiometric measurements. On
a later system, I used a part from Teledyne that was
something like T??500A. It was the analog front end
like the 7109's had but no digital part. I was able
to make a programable part do the controlling and
avoided this problem as well as stretched it to 14 bits.
 Any how, it is best to make sure that no input
exceeds the full scale, even slightly, unless you are willing
to do a number of dummy reads on a in range signal to
let the auto zero capacitor settle ( I needed 20 reads on
my setup to get it below 3 LSB's! It depends on the
resistor and capacitor values used. Larger auto zero
capacitors took longer but made more stable auto zero
functions otherwise).
Dwight
Received on Tue Jun 08 1999 - 18:22:40 BST