Problem with K8067 Temperature Sensor with extended probe

I have 2 k8067 temperature sensors connected to a k8055 USB board. One has the temperature sensor directly soldered to the board while the other has the sensor soldered to a shielded cable about 950mm long (less than the specified 1m as noted in the assembly manual). Presently both are in the same location along with a thermometer for calibration and they both read about 20°C. Over time though the readings will spike for one sample (samples are taken about twice per minute). The board with the sensor mounted directly on it has spikes both up and down of about 2 or 3°C while the one with the shielded wire connection has spikes up and down of 10 to 15°C. Is this an effect of the shielded cable extension and can it be improved by shortening the cable?

Normally, the board should not show any spikes, however, due to surrounding equipment, interference, power supply induced problems etc…, this could be the case, especially with long sensor wires.
It is always a good idea to include software filtering if these spikes cannot be filtered out 100%

Thanks for your quick reply. I will check for surrounding equipment, interference, power supply induced problems but it seems they would affect both sensors equally as they are in the same location right now and share the same power supply. The only difference is the extra shielded cable. For the permanent installation I could shorten the cable from 950mm to say 150mm. Do you think that might have a good effect? I guess it cannot hurt to try it and see what happens. I have implemented software filtering already looking for no more than a 1.5°C difference in sensor input over thirty seconds (frequency of data collection from k8055). With the temperatures I am trying to measure, a hot water tank outside skin and an engine room air space, the real temperature should never rise or fall that fast so the filtering should not affect the data negatively. What do you think?

Basically we prefer to keep the sensor on the board, to avoid all troubles.
Make sure sensor does not get damp or wet.

Thanks for your reply. I agree with keeping the sensor on the board. I am going to try shortening the cable first so that the board can stay in its enclosure. If I put the sensor on the board then the whole board will have to be located between the insulation and the outside of the steel hot water tank. As long as the sensor faces the steel and the soldered side of the board faces the insulation there should not be any shorting issues.