ard(a)p850ug1.demon.co.uk (Tony Duell) wrote: Hi There was one thing I remember about using these things 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

Oops, that was supposed to be a direct E-mail. Sorry about that. At 10:15 AM 7/3/99, you wrote: <snip>