| ??? 10/22/07 21:44 Read: times |
#146021 - I am having fun Responding to: ???'s previous message |
Erik Malund said:
the disappointment will come.
as so many before, you confuse resolution and precision. If you implying that I said high resolution equates to precision or accuracy then you are wrong Erik. I don't recall saying that, I merely stated that I have 20 bits to play with, as YOU say, that doesn't mean I have 20 bits of accuracy and I agree with you fully. However, once the ADC is working(and it is) then we can add differential measurement technique for example, the commonly known as auto zero, which is just a switch preceeding the x10 x1 amplifier. So the MCU shorts the amplifiers input to zero and takes a measurement(Vz) and stores it in memory then the MCU measures Vin (by toggling the auto zero switch) and stores it. The MCU then performs Vin-Vz and gives the result. This is the same technique as used in analogue reset/zero amplifiers so it's nothing new, but it cuts the amplifier/ADC drift to a minimum.The problem then is that the source resistance of Vin might be high so this auto/zero switching causes a CR charge wave to come out of the amplifier feeding the ADC, I have to make sure that Vin has stabilised sufficiently for the ADC read. To overcome the latter problem, a precharge amplifier is switched in momentary to overcome long cr time which is brought about by high source resistances as previously explained. Erik Malund said:
That's DEAD EASY, getting better than, say 10 bits of precision is really tough. Agreed. Erik Malund said:
sure, but realize that e.g. a Fluke voltmeter is no better than about 10 bits in precision )corresponding to 1mV in a 2V range). Well, it's as good as what the spec says, for the Fluke 8840 this works out to be 0.005%+x counts per reading for 1 year after calibration. Cheers Jason |
