Re: [libopenstm32-devel] STM32F Connectivity line, USB Bootloader and HSE crystal Frequency

[Uwe B. <bo...@el...>]

>>>>> "Uwe" == Uwe Bonnes <bo...@el...> writes:

>>>>> "Gareth" == Gareth McMullin <ga...@bl...> writes:
    Gareth> On Tue, Jul 12, 2011 at 6:41 AM, Uwe Bonnes
    Gareth> <bo...@el...> wrote:
    >>> on my STM32F107 board, the USB bootloader enumerates much faster
    >>> with a 25 MHz crystal then with a 8 MHz crystal. With the 8 MHz
    >>> crystal, there are several unseuccessfull attempts like:

    Gareth> I have two boards both enumerate correctly and reliably with
    Gareth> 8MHz crystals.  Check your crystal, but I think it's more likely
    Gareth> a software problem.  If you send me a pcap file (capture with
    Gareth> wireshark) I'll have a look at it.

    Uwe> I am not talking about user software. I talk about the system
    Uwe> software, when booting with Boot1==0 and Boot0 ==1, and I have not
    Uwe> influence on the software.

    Uwe> After enumeration, dfu-util did see the device, I did no further
    Uwe> test for now.

AN3156 clearly describes the effect:
For USB DFU bootloader V2.0, the device first tries the 25 MHz
configuration, then, if it fails, the 14.7456 MHz configuration, and
finally, if it fails, the 8 MHz configuration. If it still fails, this
operation is repeated with a large timeout value (the three configurations
are tested again). If the second trial also fails, a system reset is
generated.

So it should be normal to see two failing enumeration tries before the
succeeding third try when running on a 8 MHz crystal.
-- 
Uwe Bonnes                bo...@el...

Institut fuer Kernphysik  Schlossgartenstrasse 9  64289 Darmstadt
--------- Tel. 06151 162516 -------- Fax. 06151 164321 ----------