I've been doing some digging and i think it might be set to an insulation resistance value of 1MOhm on TLs, HFs is much lower at 0.2MOhms and will only generate a warning rather than a fault because of the separation.
The Mini Central's were changed via firmware a couple of years ago because of this problem but everyone i've spoken to at SMA say it isn't the issue.
From SMA
'The inverter type Sunny Mini Central SMC 9000TL/10000TL/11000TL is equipped with a measurement technology that goes well beyond the test conditions demanded in the standard and is capable of identifying insulation errors that can be difficult to detect. In order to meet the limiting value of 700 k Ω required by DIN VDE 0126-1-1, the devices were delivered with a limiting value of 900 k Ω .'
and
'As of December 1, 2010, SMA Solar Technology AG delivers the Sunny Mini Central 9000TL/10000TL/11000TL with a firmware where the limiting values for the insulation resistance were adapted to the new
general conditions. SMA Solar Technology offers this firmware as a free update on request for all previously
delivered inverters. The new thresholds ensure that the PV plants run safely and only actual insulation failures
(module break, marten bites) are reported as a disturbance.'