Philips 963SA mods: AD1955 question

Dear all,

I just finished some mods on my Philips 963SA. I've added a home brewn linear power supply to the analog section which is now running at 12V i.s.o. 8V. I did some capacitor upgrades to Panasonic FC (since the original ones were rated 10V only), replaced the opamps for the front channels (I/V and output) by AD8066, added a relais for muting and finally removed the output AC-coupling capacitors completely leaving the remaining DC offset to the DC servo of my Amplifier. And here comes my question:

By using low offset FET op-amps I expected the output offset to be very low. Actually in CD mode it is less than 5 mV. However when I put in a SACD the offset increases on both channels by 95 mV. This offset remains also in stop mode and even after removing the SACD and comes back down to the 5mV when I put in a redbook CD again. There seems to be a current offset (about 0.1mA) from the AD1955 DAC in SACD mode, however I could not find any hints in the datasheets. Has anybody else encoutered the same effect? Is it normal? Where does it come from?

Finally I managed to trim the output offset down by 50 mV by slightly varying the pull up resistors in the I/V section. So I now have -50 mV in CD modes and +50 mV in SACD mode. And that works for my amp. But still I'd like to know what causes this effect.

A second question: AD recommends for best THD performance a bias voltage in the I/V section of 2.8 V. To save parts Philips used the DAC's 2.4V reference voltage instead. Anyone tried to use a bias of 2.8V? Does this lead to an audible improvement?

All in all I´quite happy with the mods. Especially the heights are much improved.

Best regards and happy listening