Hi FloW, Can you tell me how many wire RTD you are using? Your setup is best used for a 4-wire RTD. In your calculations you need to make sure you know the level of precision. This in particular would be the reference resistor value. When making a ratiometric measurement, the excitation source drops out of the calculation. The returned code is the ratio of the RTD to the reference resistor (adjusted by gain applied). So it is critical that a high precision, low drift resistor is used for the reference resistor. You would then need to calibrate your system. I would recommend always issuing the SELFOCAL (self-offset calibration) shortly after power up to remove the ADC offset. You would then need to calibrate the RTD to the system by using a 2 point calibration of known values. From the calibration you would need to determine the slope which would be factored into the readings as a part of the resistance calculation. However, before that you need to stabilize your readings. I would suggest that instead of the RTD you use a fixed resistor value. There appears to be a periodic pattern to the readings you sent. As far as the input filtering values, you have very large RC values for the filter (6k Ohms and 51uF). This filter will take a long time to settle. I would suggest removing the common-mode caps and changing the differential input cap from 51uF to 10nF to start and see if you get some stability in the readings. You can adjust the filter values later once you have a stable reading. If your CS is going high before the SCLK has completed, then you need to monitor the buffers to make sure the transmit cycle has completed. A buffer or FIFO in an SPI peripheral will operate independently from your code loop and can return before an action on the peripheral has completed. One way of monitoring is to watch the receive buffer for a flag showing new data are available. Read the RX buffer before issuing the next dummy byte to transmit. Repeat the process until all 3 bytes have been read, then set CS back high. Best regards, Bob B
↧