This has been a long standing issue with the product.
I have just seen an Aspen communication from 2010 specifying an accuracy of +/-4C.
That isn’t useful for OAT measurement, or for TAS calculation, so what is it used for in the system?
Mind you, Garmin’s +/-6C GTX330 OAT probe spec (in 2005 when my GTX330 was fitted) is even worse. Somebody developed a fix for that one.
The Aspen ACU2 supports an external probe of your own liking and there is a calibration mechanism.
I installed a secondary probe but ended up putting it exactly in the exhaust stream under the belly. Engineering is about trial & error 
As Achimha indicates, an external probe can be used, both external and internal can be calibrated however, so not a real issue.
