| ??? 09/02/05 16:36 Read: times |
#100467 - I heavily doubt this Responding to: ???'s previous message |
Werner said:
I use higher switched current in most of my projects. But definitely not 100mA but around 8..10mA activ for about 1 msec per 100msecs (heating of PT100 goes with square of current). Then I have RC filtering for High Frequency pulses, plus I do averaging. Thus I get resolutions of better than 0.1°C even with 10bit internal AD converters of Atmel or Infineon or Philips chips. I heavily doubt, that pulsing the PT100 will help in increasing accuracy of PT100. If you take a sample every 100msec, by pulsing the PT100 for a period of 1msec, then your input noise filter can't provide a corner frequency of lower than fc = 1 / (2 x pi x R x C) = 1 / (2 x pi x 200µsec) = 796Hz assuming that the time constant of your single pole RC noise filter must be smaller than 1msec / 5 = 200µsec, which already gives an error of 0.7%. So, in the frequency range between DC and about 800Hz you don't have any filtering at all! Only a bit dampening resulting from the averaging, but which isn't really much. Using a DC current through the PT100, means avoiding the pulsing, on the other hand, would allow you to choose the corner frequency down to 0Hz. A corner frequency of 1.6Hz would yield a dampening of 30dB at 50Hz, 36dB at 100Hz, 46dB at 300Hz and 54dB at 800Hz!! Do you really think, that pulsing yields lower noise and higher precision?? By the way, zener diodes are not suited to protect against ESD! They are not fast enough. Kai |
