Re: CM11 Status Request (C# Language)

[Charles Sullivan <cwsulliv@xxxxxxxxxxxx>]

On Thu, 23 Nov 2006 11:10:32 +0000, Dave Houston wrote:

> Neil Cherry <njc@xxxxxxxxxxx> wrote:
>
>>On Thu, 23 Nov 2006 06:00:25 GMT, Charles Sullivan wrote:
>>> I've seen some unusual behavior of USB->Serial adapters compared with
>>> actual serial ports under Linux.  I've assumed these are bugs in the
>>> Linux drivers but Windows drivers (for some adapters at least) may exhibit
>>> similar behavior.
>>>
>>> E.g., a blocking read of N bytes may return after the first byte read.  So
>>> Dave's polling ought to work, while expecting to get all N bytes with one
>>> read may fail.
>>
>>I heard of such things and my understanding is that these USB/serial
>>dongles are usually dongles meant for PDA's for some reason they have
>>a lot of trouble acting as normal serial ports. I'm unsure as to why
>>this is.
>
> A lot of the cheaper ones don't do anything with the handshake lines. But
> the CM11A doesn't use any handshake lines so even the cheap ones should work
> as far as the hardware goes. As for driver flaws, that's a possibility,
> although the complaints I've seen from Windows users can almost always be
> traced to the missing handshake lines. They are slower than standard serial
> ports because they must translate from/from serial-USB at each end of the
> link but this isn't a major factor at 4800bps.

All the USB->Serial adapters I've tested (I have five) implement at least
some modem control lines.  The cheapest I have is a $10 model from
www.byterunner.com; the most expensive a $40 Keyspan USA-19HS.  The chips
represented among these (as identified by the Linux drivers) are an older
Prolific PL2303, a newer Prolific PL2303 (in the $10 model), an FTDI
FT232BM, and whatever chip Keyspan uses.

I've used all five in Linux to actuate a CM17A Firecracker with the RTS/DTR
lines, and I've started using the RI line to detect pending incoming PLC
signals from the CM11A.  The only problems I've encountered appear to be
Linux-related: When used with the latest Linux 2.6.18 kernel the RI line
on the older PL2303 chip models gets stuck in the active state and the
Keyspan doesn't work at all.  But both work fine with earlier kernels.
(Bug reports have been filed.)

I can't say what may or may not be implemented with Windows drivers, and
granted, there may be timing issues with the control lines when one of
these adapters is used to connect to a high-speed modem, but that's
not something I've tested.

Regards,
Charles Sullivan