I've recently found the need to control a Pioneer Laserdisc player with LIRC, so I've hacked up an FT232RL board from an old Arduino NG with the AVR removed (picture). This is effectively the same hardware that I was using when Albert and I submitted the original version of the ftdi driver back in 2008. The FT232RL has the date code 0551-A.
With lirc 0.10.1, both reception and transmission run at the wrong rate - the received timings reported by mode2 are 25% of what they should be for the Pioneer's NEC protocol.
In the original version of the ftdi driver (which works with the FT232RL), we followed the datasheet which says that the bitbang mode sample rate is 16x the baud rate you set. When transmission support was added in January 2010, this was changed to 32x (with a comment saying "32 seems to be correct", although I don't know which chip that was tested with). Then in #275, it was changed to 64x based on testing with the FT230X.
If I change the multipliers back to 16x for reception and 8x for transmission, both transmission and reception work correctly again on the FT232R. (Or, at least, as well as they usually do given how buggy the FTDI chips are!)
It sounds to me like this is a bug in libftdi's ftdi_set_baudrate - the baud rate fudge that it does for bitbang mode is correct for the FT232 but not for other chips. I'm not sure what the best way to fix this from LIRC's perspective is - we could add driver options for the multipliers?
I see the same problem. any news?