Emmanuel Blot

If you have anyway to capture the actual USB transfers between the host and the FTDI device, it guess it would help to understand if a faulty command is sent...

HW USB analysers are quite expensive (~ $1000 for USB2 HS). SW USB analysers are cheaper, some are free, but they are usually OS-specific. There are many for Windows, a...

> Is there an API reference for the MPSSE engine I can use to decode the data coming in/going out? The reference from FTDI chip would be [Application Note 108](https://www.ftdichip.com/Support/Documents/AppNotes/AN_108_Command_Processor_for_MPSSE_and_MCU_Host_Bus_Emulation_Modes.pdf)...

Note that v0.44.0 fixes a couple of MPSSE parsing issues, you might want to upgrade if you use the tracer to help debugging.

Well... TL;DR: **do not** :-) Long version: Never *ever* use a destructor in Python except if you have a very, very specific and verified need to do so. [`__del__`](https://docs.python.org/3/reference/datamodel.html?highlight=__del__#object.__del__) is...

`atexit.register` is the easiest way. However, I'm not sure why you need to call `close`/`terminate`. I'm missing the point: there's a bug if you do not call it?

Ok, let me try to sum up the issue: #### Issue The I2C invalid clock frequency occurs at random when: - two I2C master are created on the same FTDI...

> 1. No, also manifests when only 2nd I2C interface is instantiated > 2. Yes, see [#159 (comment)](https://github.com/eblot/pyftdi/issues/159#issuecomment-589907089) Ok, so the issue can be simplified/summed as: Closing PyFTDI on the...

> Given that it only happens with the 2232H, and that it seems to be related to the number of commands sent (See [#159 (comment)](https://github.com/eblot/pyftdi/issues/159#issuecomment-566696153)), is it possible that an...

I've added a new API, `Ftdi.reset()` in 98a808d18bd0143577a5362a0e2b81b970080def, which optionally performs a full USB-level reset. It is not automatically called, as it resets all interfaces at once, but it might...