From: Rich Webb on
On Thu, 22 Oct 2009 10:57:24 -0700 (PDT), rickman <gnuarm(a)gmail.com>
wrote:

>On Oct 22, 11:19�am, "Not Really Me"
><sc...(a)validatedQWERTYsoftware.XYZZY.com> wrote:
>> Mel wrote:
>> > I'm having trouble with an EEPROM and an LCD module hooked to an
>> > AT91SAM7S256. �Software is using the at91lib TWI_* routines that came
>> > with
>> > an evaluation kit. �The program seems to be hanging up on incomplete
>> > I/O
>> > with the LCD module, but I don't know why. �Analyzing the TWI (aka
>> > I2C)
>> > protocol would either nail the problem or assure me that there was
>> > nothing
>> > to see there.
>>
>> > Where could I get the use of an analyzer?
>>
>> > Thanks, Mel.
>>
>> Yikes, buy, don't rent. �We have a DigiView from Tech-Tools that we love.
>> $500 athttp://www.tech-tools.com/dv_main.htm.
>>
>> This is phenominal logic analyzer. �Does data interpretation and lots more.
>
>I don't have this unit myself, but one of my customers uses it and has
>more than one.
>
>http://www.pctestinstruments.com/
>
>It seemed to work pretty well to me. $389

That's my normal LA, also. Lives in the laptop bag so it's always handy.

--
Rich Webb Norfolk, VA
From: rickman on
On Oct 22, 4:03 pm, Rich Webb <bbew...(a)mapson.nozirev.ten> wrote:
> On Thu, 22 Oct 2009 10:57:24 -0700 (PDT), rickman <gnu...(a)gmail.com>
> wrote:
>
>
>
> >On Oct 22, 11:19 am, "Not Really Me"
> ><sc...(a)validatedQWERTYsoftware.XYZZY.com> wrote:
> >> Mel wrote:
> >> > I'm having trouble with an EEPROM and an LCD module hooked to an
> >> > AT91SAM7S256.  Software is using the at91lib TWI_* routines that came
> >> > with
> >> > an evaluation kit.  The program seems to be hanging up on incomplete
> >> > I/O
> >> > with the LCD module, but I don't know why.  Analyzing the TWI (aka
> >> > I2C)
> >> > protocol would either nail the problem or assure me that there was
> >> > nothing
> >> > to see there.
>
> >> > Where could I get the use of an analyzer?
>
> >> > Thanks, Mel.
>
> >> Yikes, buy, don't rent.  We have a DigiView from Tech-Tools that we love.
> >> $500 athttp://www.tech-tools.com/dv_main.htm.
>
> >> This is phenominal logic analyzer.  Does data interpretation and lots more.
>
> >I don't have this unit myself, but one of my customers uses it and has
> >more than one.
>
> >http://www.pctestinstruments.com/
>
> >It seemed to work pretty well to me.  $389
>
> That's my normal LA, also. Lives in the laptop bag so it's always handy.

I only wish they had drivers for Linux. I'm not using Linux now, but
I don't want to invest in more things that will make a switch hard.

Rick
From: Neil on
Mel wrote:
> I'm having trouble with an EEPROM and an LCD module hooked to an
> AT91SAM7S256. Software is using the at91lib TWI_* routines that came with
> an evaluation kit. The program seems to be hanging up on incomplete I/O
> with the LCD module, but I don't know why. Analyzing the TWI (aka I2C)
> protocol would either nail the problem or assure me that there was nothing
> to see there.
>
> Where could I get the use of an analyzer?
>
> Thanks, Mel.
>
You could also get n I2C Monitor.
MCC-US.com
Or
www.totalphase.com
From: Paul E Bennett on
Mel wrote:

> I'm having trouble with an EEPROM and an LCD module hooked to an
> AT91SAM7S256. Software is using the at91lib TWI_* routines that came with
> an evaluation kit. The program seems to be hanging up on incomplete I/O
> with the LCD module, but I don't know why. Analyzing the TWI (aka I2C)
> protocol would either nail the problem or assure me that there was nothing
> to see there.
>
> Where could I get the use of an analyzer?
>
> Thanks, Mel.

Firstly, you didn't say which LCD module and EEPROM. Doing so might have got
a wider response already.

Secondly, Logic Analysers are expensive to own and hire so doing the job
without is really a very worthwhile skill that more should learn to
cultivate.

If these are serial devices I would expect to see the traces on a scope
easily enough. Have you got one that can capture a segment of data from pre-
trigger point to post-trigger point (Fluke and Tektronix both do a suitable
type). Such a scope will allow you to scroll back through the problem zone
to see what was happening in detail.

Also learn to instrument the code (toggle a pin with the idle task or
something) so that you can pick up points of failure.

--
********************************************************************
Paul E. Bennett...............<email://Paul_E.Bennett(a)topmail.co.uk>
Forth based HIDECS Consultancy
Mob: +44 (0)7811-639972
Tel: +44 (0)1235-510979
Going Forth Safely ..... EBA. www.electric-boat-association.org.uk..
********************************************************************

From: Mel on
Paul E Bennett wrote:
> Firstly, you didn't say which LCD module and EEPROM. Doing so might have
> got a wider response already.

I hate to damn a manufacturer with problems that are probably my own -- see
below.

> Secondly, Logic Analysers are expensive to own and hire so doing the job
> without is really a very worthwhile skill that more should learn to
> cultivate.

Funny you should say that. After my subconscious got serious about the
problem, and I posted my request, I woke up in the night realizing that I
was clocking the TWI interface too quickly for the LCD. This seems to be
have been true, and bringing the baud rate down to 100KHz has made the board
beautifully stable.

> If these are serial devices I would expect to see the traces on a scope
> easily enough. Have you got one that can capture a segment of data from
> pre- trigger point to post-trigger point (Fluke and Tektronix both do a
> suitable type). Such a scope will allow you to scroll back through the
> problem zone to see what was happening in detail.

The thing I have is a USB scope pod. It would be worth a try to drag the
trigger point way over to the right and see if that left me with the last
available traces. Since the problem involved intermittently locking up the
U.I. loop, and ceasing to send or receive anything from the module, that
ought to leave the smoking gun traces in plain view.

> Also learn to instrument the code (toggle a pin with the idle task or
> something) so that you can pick up points of failure.

Alas, the hardware people didn't leave me any test points. Running that
test would have me playing drums and pressing front-panel buttons with
several hands while holding a scope probe point on an MCU pin with another.
Life is never easy.

Mel.