DIYbanter - View Single Post

Don Y[_3_]

On 9/3/2015 1:42 PM, wrote:
On Thu, 03 Sep 2015 12:25:55 -0700, Don Y
wrote:

On 9/3/2015 12:13 PM, wrote:
On Thu, 03 Sep 2015 09:32:59 -0700, Don Y
wrote:

Everything is a CPU these days. It is cheaper to write software and
use an off the shelf CPU chip than to design a purpose built chip.

That depends on the functionality you intend in the product. Note
that CPUs go out of production just like "dedicated chips"...

You can still buy 8080 chips but these people are using a standard PIC
of some sort.

How about SC/MP's? 2650's? 8x300's? A bazillion *specific* 8051
derivatives? etc.

In the 70's and 80's, the MPU market was all about "second sources"...
you wanted to have (at least one) backup vendors for the parts that
you'd design into a product. Nowadays, I don't think there are any
parts that are made by two different vendors that are "pin compatible".

Fixing mistakes is a lot easier too.

grin When was the last time you GOT an update to your microwave
oven software? The GPS *software* (not MAPS) in your car? Any of
the dozens of ECU's in your vehicle? The controller in your furnace?
You washing machine/dryer/dishwasher?

It is still easier to fix the ones on the line.

Yes. But not much help for the folks who already purchased the
"previous bug-set".

There are flash changes for cars and you certainly see a lot of
microcode upgrades on a smart TV.

Only *smart* (network connected) TV's. And, only if you have it
connected to the 'net. At the same time, you are at the mercy
of the manufacturer to *not* leave you with a LESS DESIREABLE
product than the one you purchased. Or, *not* install additional
spyware, etc.

Imagine coming to work and finding your computer has been "upgraded"
overnight, without your forewarning. What does that do to productivity?

It's one thing if the upgrades fix broken behaviors. But, more often than
not, they *change* behaviors -- often in BIG ways!

Fixing is a misnomer. *Changing* is a better description. Manufacturers
make *changes* (going forward) which may (or may not) "fix" problems.
But, folks *with* those problems end up living with them. I.e., the
CPU doesn't buy the consumer anything!

It makes the product cheaper to make. How much of that filters down to
the customer is debatable.

If you factor in the cost of added (user) complexity and dubious
functionality, I wonder if there *is* a net improvement!

In the 70's, we embraced MPU's as a means of replacing dedicated logic
to achieve comparable/improved performance at reduced cost. But, this
quickly got out of hand. "Feeping Creaturism" took over and folks
started cramming *too* much functionality into things that weren't
intended to have that level of complexity. E.g., our microwave has
buttons that we've NEVER PRESSED! In 15+ years!! Likewise, the
"probe" that allows the oven to monitor the interior temperature
of whatever it's cooking... never been used, I doubt I could
even tell you where it's *stored*! But, the probe, the connector,
the electronics and the software were all added to the cost of
the microwave.

New cars have support for XM built in. What if I never want an XM
subscription? How do I get "credit" for the extra, unused, potential
for bugs/failure/complexity increases that the "feature" has cost me?

What is the cost of providing those buttons (tangible hardware cost)
and the software behind them? I.e., we've bought features that
we'll never use -- and didn't really have a choice in the matter!

You are preaching to the choir here. I like hard wired circuits vs
processors but nobody listens to me.

It depends on the functionality that you want to provide. E.g., if
you wanted your pool/solar controller to assume a higher degree of
utilization on *holidays* than on WORKdays, you'd be hard-pressed to
do this with discrete logic.

E.g., we have "dual cooling", here -- ACbrrr and Evaporative/swamp Cooler.
You simply can't find a thermostat that knows how and when to use each!
Instead, the *human* (occupant) becomes an integral part of the
control loop. *Then*, you can use a simpler implementation (e.g.,
a simple thermostat).

OTOH, if you want to take the human *out* of the loop, then you need
a sh*tload more "smarts" in the control system. What's the current
humidity/dew point? How effective will the swamp cooler be *at*
that DP? How is that DP likely to *change* in the coming HOURS?
(if it's likely to RAIN later today, you surely don't want to push
even *more* moisture into the house with the cooler -- that the ACbrrr
will eventually have to remove ALONGSIDE the increased humidity)

*When* do you want the house to attain a particular comfort level?
If the occupants are away (vacation)... shrug. If they're at *work*,
then you probably want the house to be comfortable *when* they get
home -- not some time *after* they're home! So, start getting things
ready BEFORE they get home. But, how far in advance is appropriate?
Delay too long and they'll come home to an uncomfortable house. Start
too early and you've kept an *empty* house comfortable! :-/

My spa controller is 4xxx CMOS and my pool/solar controller is very
old school with a 24 hour timer motor, 3 cams with microswitches and
very simple switch and relay logic.to control 5 valve servos

My first (audio) cassette deck had TTL control logic. It's replacement
has several processors doing all sorts of different things! Yet, the
user interface hasn't appreciably changed (play, stop, pause, rewind, etc.)