AVR-Chat

On Wed, Dec 23, 2009 at 09:22:00AM -0600, Tim McDonough wrote:
> David Kelly wrote:
> > On Mon, Dec 21, 2009 at 12:06:41PM -0000, leon Heller wrote:
> >> Designers select a chip which gets the job done as cheaply and
> >> easily as possible. In some situations it might be an AVR, in
> >> another it might be PIC, or something else.
> > 
> > Only *good* designers do that. Others have chip/vendor religion.
> > When one is not willing to use any tool other than a hammer, all
> > problems look like nails.
> 
> When I used to work in a larger shop we tended to to stick with a
> particular supplier/series of parts for financial reasons beyond piece
> part price. With many dozens of products based on the same family of
> parts there is a lot of code available to shorten prototype and
> development time. Quoting lower development costs and still delivering
> products that meet the customer's requirements wins business.
> 
> That's not to say you should NEVER look at another processor. But in
> some shops mostly sticking with a certain family is a sound choice.

Yes, of course there is always the cost of adding yet another line item
to one's inventory. But there are many who will bust butt to shoehorn
something like ethernet on a CPU which doesn't fit "because we have
always used that CPU" when just about any other CPU would be a better
choice.

Once upon a time we had need of a test fixture for an item on a low
volume production line. Another engineer had already tasked our
excellent technician to build and wire the fixture from sketches
including an RS-232 interface to be connected to a PC. It was completed
in less than a morning. When I heard about it the Visual Basic
"programmer" assigned the task was complaining about how many weeks his
task was going to take, and take away from other things he was (not)
doing.

I designed (and coded) the product under test. Was disgusted at the VB
guy's answer, not not surprised. Thought about options during the
meeting. Then less than 2 hours later I came back to the tech with a
programmed 16C505 (I forgot, but know it was a '505, and a precious
expensive $10 ceramic reprogrammable one at that) and asked that he
rewire the wire wrapped MAX232 for the PIC, and add a push button for
"Start Of Test".

It worked 90% first time. I had bit-banged the asynchronous TX and we
didn't use a crystal on the PIC, and IIRC there are no timers or IRQ in
the '505. My baud rate was off just a bit. But on thinking about it we
decided we *liked* the tester sending a scrambled messages 1 out of 10
times as it demonstrated the unit under test's oscillator was spot-on
accurate rather than just "good enough" accurate. We didn't change a
thing. That was 15 years ago and the tester is still occasionally used
at the repair depot.

The entire PIC firmware image was less than 192 bytes (or words, its a
PIC and I forgot, probably 14 bit words?).


-- 
David Kelly N4HHE, dkelly@HiWAAY.net
========================================================================
Whom computers would destroy, they must first drive mad.