Hello TI,
We are building a product in which water level readings are critical. We are now using the FDC1004EVM with our own flex designed exactly according to the reference design doc from TI. The test results are too noisy to do the further math to figure our the water level.
Out design is as followed:
MEAS1 => CIN1 => Level
MEAS2 => CIN2 => R_L
MEAS3 => CIN3 => R_E
MEAS4 => CIN4 => Shield2
They are configured as follow:
MEAS2 => CIN2 => R_L
MEAS3 => CIN3 => R_E
MEAS4 => CIN4 => Shield2
They are configured as follow:
MEAS1 CHA=CIN1 CHB=CIN4 (differential measuring to Shield2)
MEAS1 CHA=CIN1 CHB=CIN4
MEAS1 CHA=CIN1 CHB=CIN4
MEAS1 CHA=CIN1 CHB=CIN4
MEAS1 CHA=CIN1 CHB=DISABLE
We are expecting to see CIN1 linearly growing with filling water and relatively steady R_L and R_E so that we can calculate the water level. Instead, as can be seen from the following test results, we have x3 problems.
We are expecting to see CIN1 linearly growing with filling water and relatively steady R_L and R_E so that we can calculate the water level. Instead, as can be seen from the following test results, we have x3 problems.
(1) The channels are messed up for some reason:
As shown, CIN2 actually acts like the LEVEL channel and CIN4 looks like the R_L channel.
After power cycling the EVM and restarting the API, the channels are correct to our design. But there are 2 more problems.
(2) R_L is linearly changing with the LEVEL channel;
(3) Both LEVEL channel and R_L lagged from the actual flushing.
after 2 min:
The lagging problem doesn't seem to happen when filling though. And I also noticed that the lagging is longer when using a bigger bucket.
Any suggestions?
Thanks,
Bohan
