> > ----------------------------------------------------------------- > >From the terminal of: David R. Hatch > dhatch@gray.cscwc.pima.edu Now sporting 105!!! MEGS of hard > disk space!!! > ----------------------------------------------------------------- > > > I think what I would do is stay with the 8 bit a/d and stick an > instrumentation amp - a very accurate one, BTW - in between it and the > bridge. A gain of say 20000 would give your dV of 2.44E-6 an apparent > range of 48.8 mV to the a/d. That translates to about a five value > range for your dV. Accuracy could be easily increased at the loss of range. > A simple window comparator could be made for 'gross' over- or undertemp > range. > This would be my method mostly because I can't see myself paying for a > 24 bit a/d when an 8 bit will do. > > David Hatch > Hi David, Thanks for your reply! I'm not sure I follow it exactly. Let's see ... a gain of 2E4 would put an LSB of the converter at 48.8 mV, giving about 51 steps in 2.5 volts. Any more steps than that and we'd be into the rail. This gives us a dynamic range of 5.1E-3 Kelvin and would only require a 6-bit converter (minimum) to do the job. So in order for the loop to operate in its linear range, we have to be within 5.1 millikelvin of the set point. I'm wondering about the loop damping being a problem when the outside temperature (ambient) is being perturbed. Let's say the outside temperature is stepped by a few degrees ... will the loop over/undershoot more than 5.1 millidegrees while it compensates for the step change in ambient? If it does, it could be bad news, depending upon how the over/under ranging is done. But, it might be OK. When we cause our environmental chamber to change temp by 30 degrees in five minutes, we do see more than 5 millidegrees overshoot in the response of the crystal oven controller. Anyway, from a cost standpoint, your approach is definitely worth considering. Right now, we're more worried about physical space available and less sensitive to cost. So having the AD7714 doing dual duty (measuring the thermistor bridge and the output current of a PWM DC-DC converter) may be worth the cost in comparison to the space it consumes. But I'll definitely take a closer look at your idea. Thanks! Jim J.