in-amp or transducer drift

[V_King]

Hello,

I am using AD623 instrumentation amplifier for the load cell output signal conditioning.

The signal is constantly drifting by a very small amount, which makes hard to set reference values for the calculations in the software.

Firstly I was assuming that the temperature is the only cause, as it makes the biggest difference in the readings. But even keeping the load cell constant temperature I get a drift.

One friend, who has more experience with electronics suggested that that might be due to the fact that I am using single supply for the AD623 in-amp.

I wonder if swapping to the dual supply and even swaping in amp to AD620 help if not to solve, at least to reduce the drift problem? I used the reference circuit from AD623, datasheet p.17, fig45 for the circuit.

Cheers

[Odysseus]

How much is it drifting by, and how much gain have you configured the circuit for (i.e. what is the value of Rg)?

I ask because it may simply be due to the noise generated by the chip itself.  Table 3 on page 6 and figure 10 and 13 on page 9 contain some relevant information.  For example, if your amplifier is configured for a gain of 1000, then according to the table, the low frequency 0.1-10Hz RTI noise is 1.5uVpp.  Thus the output referred noise will 1000x this, i.e. 1.5mVpp of drift.

[qno]

Be shure about temperature.
I have seen drift due to a connector near a voltage regulator.
One side of the connector was a tiny bit warmer than the other side.
Removing the voltage regulator solved the problem.

A thermocouple effect can have big influence with a gain of 1000.

[V_King]

thanks for answers.

the load cell has an internal pre-amplifier to the output of 0-5V and load cell manufacturer supplied the power supply as well. I am just using the ad623 at the other end as a buffer and to reduce common noise at a unity gain, as the cable is 4m long (shielded) and the load cell is in noisy environment.

the drift is around 20mV/hour, which corresponds to 10kg in the load cell and is 10 times larger than the load cell specs, so I presume there is space for improvement. I made some rough compensation for the drift and it sort of works as the load cell is used max 10-15 min at a tme, but the engineers curiosity keeps me coming back to look for improvements   

[tszaboo]

Do you have a bias current path? Is the transducer floating?