The reason why X-B.I.Ts don't work with some 1.1 boxes is simply because X-B.I.T is not strong enough to pull D0 line low.
Today we got our hands on a 1.1 box that refused to work with X-B.I.T. We tried a couple of X-B.I.T chips to rule out the possibility of a faulty chip, but the results were same with all the chips.
We used an oscilloscope to investigate what really goes on with the D0 line. When X-B.I.T is in stealth mode and box is turned on, the D0 line goes up for about 300 ms and after that the contents of flash is read and thus D0 goes up and down all the time. The HIGH level is about 5 V and the LOW level is about 0 V.
When X-B.I.T is not in stealth mode the D0 line goes initially up to 5V for about 20 ms. After that the X-B.I.T obviously kicks in and we get about 2.2 Volts for about 280 milliseconds after which the data is transmitted but this time the HIGH level is only 2.2 Volts.
So after these tests it is crystal clear that XBIT tries to pull the D0 to ground, but it is not strong enough. While it is supposed to get D0 to LOW value it only manages to get it to 2.2 Volts which is still considered to be HIGH by most chips.
This is not due to a poor contact with D0 since the measurement was done on the XBIT board. If there was poor contact, we would get lower voltages in which case the problem would be even worse.
The problem is clearly with the X-B.I.T chip. With competing chips we got D0 values as low as 0.1 - 0.2 Volts, which can safely be assumed to be interpreted as LOW in all circumstances.
Edit: 280 milliseconds, not 280 seconds