htfs 200-p hall effect possible interference

Hi,
I have finished building a DC data logger for monitoring amps from 0-300a and also voltage from 0-30v which will be used to monitor how a vehicle is used to help with Charge balance calculations for the batteries and equipment that I've installed.
I have used an Audruino mega 2560 with an adafruit ADS1115 to achieve 16 bit resolution, plus x2 HTFS 200-P lem hall effect sensors.
My problem is, that now we have done some mapping in the code, at first glance when testing from 0-300 amps, it appeared to follow a proven data loggers accuracy (75mV shunt). I went up in 10 amp intervals and it followed it within a 2 amp range.
I then noticed that if I wait a while (20-30 seconds) on a set amperage, it began to run away with itself and increase in amperage. The runaway seems to start around the 75 amp mark and gets worse the higher I go. So 7 amps out at 75 amp and 22 amps out at 300 amps. This is not a constant, and it will settle back down to the correct amperage after a while, then start to climb again in cycles.
I added a 4.7nF cap between 0v and output on the sensor and a 47nF cap between 0v and 5v on the sensor, but it hasn't helped.
I believe this to be an interference problem but am in danger of running round in circles so just wondered if anybody had tried a similar project and solved the problem before?

Here's a link to the data sheet for the sensor... http://www.europowercomponents.com/media/uploads/HTFS%20200-P.pdf

@100 A through the sensor, what is the output to the ADS1115? Does it vary over time? What is the digital output from ADS1115? Does it vary over time? Can you post your program + the Serial.print output?

How did you connect an internal reference voltage based A/D to a sensor with ratiometric offset.
Did you also sample the offset voltage?
Can you post schematic and code.
Leo..