Searching \ for '[EE] Simple circuit design favor?' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: www.piclist.com/techref/index.htm?key=simple+circuit+design
Search entire site for: 'Simple circuit design favor?'.

Exact match. Not showing close matches.
PICList Thread
'[EE] Simple circuit design favor?'
2007\02\08@110653 by William Couture

face picon face
I'm hoping someone on the list can do a 5 minute (probably a bad estimate)
design for me.

This will be a "proof of concept" to show someone here at work that I'm
not crazy (or someone can show me that I am!).

We're talking about a re-design on several products (consolidate into
single PC board with different BOMs, better technology, etc).  One
variation of a current product has a 3-digit mechanical "digital switch"
(12 wires), a slide switch (2 positions, one input bit), and a pot (1 analog
in), all on a "front panel" that is at right angles to the main board.

The number of connections makes it hard to manufacture.  What I'm
suggesting is that it can be done with 4 or 5 connections if we put a
small processor on a PC board and talk to it via SPI or I2C.

We're moving to Atmel processors, so something like the ATTiny26
seems like it would fit the bill (as a side note:  Does Atmel have a
parametric parts search like Microchip?).

Would it really be as simple as switch wires to digital pins (possibly with
pull-up/pull-down resistors) for the "digital switch", a pull-up / pull-down
for the slide switch, and the pot slider to an analog pin, 2 or 3 lines to
the main processor or SPI / I2C, power, ground, plus some decoupling
caps?  Or am I missing something big?  We don't need a lot of speed
nor accuracy for  communications, so the internal oscillator should be
more than enough for the clock.

Thanks,
  Bill

--
Psst...  Hey, you... Buddy...  Want a kitten?  straycatblues.petfinder.org

2007\02\08@112538 by Harold Hallikainen

face
flavicon
face
What you propose sounds quite doable. One thing to consider about the
"digital switch," which, I think is a thumbwheel switch. We use those in a
few products. There are right angle DIP sockets that allow you to mount
this to the board with no hand wiring. We use a 3 digit thumbwheel mounted
this way.

If you do need to go to another board, any serial protocol should work
fine. I've done a couple designs where I have a 7 segment display and
several switches remote from the main board. I'm doing a one wire
communications system between the main board and the remote board using a
UART at each end. Tie the UART RX pins together and add a pull-up. Put a
diode between each of the TX pins and the associated RX pin with the
cathode towards the TX pin. This makes an open drain circuit where either
UART can pull the data line down to the space condition. Do a simple
packet protocol that includes a destination address since both UART
receivers see all data. I have the main board send a packet about 50 times
a second. The remote board decodes the packet, updates the display, and
responds with the current switch settings. I recently added a TAOS light
to voltage converter to the remote board so now it responds with switch
settings and the sensed light level (these are watching the light level
coming down an optical fiber, then adjusting the light source to regulate
that light level). So, you can do quite a bit with one data line, +5V, and
ground.

Harold


--
FCC Rules Updated Daily at http://www.hallikainen.com - Advertising
opportunities available!

2007\02\08@114007 by David VanHorn

picon face
>
>
> We're moving to Atmel processors, so something like the ATTiny26
> seems like it would fit the bill (as a side note:  Does Atmel have a
> parametric parts search like Microchip?).


No.. :(

Tiny-26 is going away, but there are plenty of other parts that will get you
there.
The mega-48 is pretty much equivalent in cost, depending on quantity.

Would it really be as simple as switch wires to digital pins (possibly with
> pull-up/pull-down resistors) for the "digital switch", a pull-up /
> pull-down
> for the slide switch, and the pot slider to an analog pin, 2 or 3 lines to
> the main processor or SPI / I2C, power, ground, plus some decoupling
> caps?


Well.. Depends on your design philosophy.  You could do it that way, but
since it's a front panel, I'd be adding ESD protection on anything the user
touches. Mostly diodes, caps, and resistors.
But the general concept is sound.  On-chip RC clock is plenty fast, and you
can use either the on-board SPI, or bit-bang it yourself, or whatever makes
sense.

2007\02\08@114826 by Stephen R Phillips

picon face

--- William Couture <spam_OUTbcoutureTakeThisOuTspamgmail.com> wrote:

{Quote hidden}

One project I worked on had the User Interface (UI) on a serially
connected device from a PIC. LED's etc.  Essentially a terminal.  Here
are the list of things that are NOT ok to do that with.  Anything that
is for emergency stop, anything that requires a very high reliability
IE starting something dangerous. Otherwise it's fine.  We used RS485
and a 24V DC power (an early adoption of USB! ;) ).

The UI acts like a traditional dumb terminal in this case.  Your case
it depends on what you are wanting to do.  I2C and SPI are
unidirectional without complicating things overly much. So as long as
you need not turn lights on etc, you are good to go.  I on the other
hand had an LCD backlight and several indicator lights as well as a
sounder to turn on.    As well as spewing text to an LCD.  The key
strokes were events and the humidity was poled.

Yes it was the operating panel for a HVAC control system.

In any case you need to be sure that which must be hardwired is
hardwired, merely for safety reasons.  Anything that is merely an event
and has a low safety factor associated with it, using a uC for that is
fine.  Just be sure it works I always say.


Stephen R. Phillips was here
Please be advised what was said may be absolutely wrong, and hereby this disclaimer follows.  I reserve the right to be wrong and admit it in front of the entire world.



____________________________________________________________________________________
The fish are biting.
Get more visitors on your site using Yahoo! Search Marketing.
searchmarketing.yahoo.com/arp/sponsoredsearch_v2.php

2007\02\08@115952 by William Chops Westfield

face picon face

On Feb 8, 2007, at 8:06 AM, William Couture wrote:

> What I'm suggesting is that it can be done with 4 or 5
> connections if we put a small processor on a PC board
> and talk to it via SPI or I2C.
>
Well, sure, the hardware folk could do something like that, but
it would make a lot of work for the SW people...  :-)

You might use a typical serial mouse as your proof of concept.
Whenever there was a change of some mouse-position or switch,
the mouse would fire off a couple bytes containing all of the
relevant data: "position" (or was it position delta) and the
current state of all switches...  (PS2 and USB mice are similar,
as far a I know...)

BillW

2007\02\08@131418 by William Couture

face picon face
On 2/8/07, David VanHorn <.....dvanhornKILLspamspam@spam@microbrix.com> wrote:

> > We're moving to Atmel processors, so something like the ATTiny26
> > seems like it would fit the bill (as a side note:  Does Atmel have a
> > parametric parts search like Microchip?).
>
>
> No.. :(

Bummer.  Do they even have a "line card" that lists processors / pins /
peripherials / memory as a matrix?

> Tiny-26 is going away, but there are plenty of other parts that will get you
> there. The mega-48 is pretty much equivalent in cost, depending on quantity.

There is the ATTiny261 and it's relatives, so a "drop-in" chip is available.

As for the Mega48, yah it's pretty close in price, but has more pins than
we need and is (probably) in a bigger package the SOIC-20, which makes
things harder to fit.

{Quote hidden}

Problem is, I can see our EE saying that anything that connects to the
processor needs opto-isolators (he's already said that in relation to the
machine I/O), and that, as a result, this will be too big to fit.

Sigh...

Bill

--
Psst...  Hey, you... Buddy...  Want a kitten?  straycatblues.petfinder.org

2007\02\08@135458 by David VanHorn

picon face
>
> Bummer.  Do they even have a "line card" that lists processors / pins /
> peripherials / memory as a matrix?


No, they really need to do that..
This is about as close as it gets:
http://www.atmel.com/products/avr/overview.asp

{Quote hidden}

We're using it in MLF package, at 4mm square.
Not bad to prototype with.


> Problem is, I can see our EE saying that anything that connects to the
> processor needs opto-isolators (he's already said that in relation to the
> machine I/O), and that, as a result, this will be too big to fit.


We did similar things with Verifone terminals, display, keyboard, and card
reader on one board, talking to the main board.  Didn't really need much
protection, they even stood up well on the carpets at the shows in Las
Vegas.  It says something when the vendor across from you, with the
stainless steel ATM, is out spraying down the carpet, and your plastic box
products are withstanding the best "plastic sole shuffle" attempts to kill
it with no special precautions.

There's not a lot that you cant fix with attention to the plastics/keypad
design, and good design/layout.

2007\02\08@143411 by William Couture

face picon face
On 2/8/07, David VanHorn <dvanhornspamKILLspammicrobrix.com> wrote:
> >
> > Bummer.  Do they even have a "line card" that lists processors / pins /
> > peripherials / memory as a matrix?
>
>
> No, they really need to do that..
> This is about as close as it gets:
> http://www.atmel.com/products/avr/overview.asp

Actually, I found my own answer:

Yes, they do have a parametric table.  You can even sort on various
columns.  Not as nice as Microchips, though...

http://www.atmel.com/dyn/products/param_table.asp?family_id=607&OrderBy=part_no&Direction=ASC

You can also download it as a spreadsheet to Excel.

Bill

--
Psst...  Hey, you... Buddy...  Want a kitten?  straycatblues.petfinder.org

2007\02\08@145552 by Harold Hallikainen

face
flavicon
face

> On 2/8/07, David VanHorn <.....dvanhornKILLspamspam.....microbrix.com> wrote:
>
>> > We're moving to Atmel processors, so something like the ATTiny26
>> > seems like it would fit the bill (as a side note:  Does Atmel have a
>> > parametric parts search like Microchip?).
>>
>>
>> No.. :(


I'm not familiar with Atmel processors. Why are you moving to them?

Harold

--
FCC Rules Updated Daily at http://www.hallikainen.com - Advertising
opportunities available!

2007\02\08@152841 by William Couture

face picon face
On 2/8/07, Harold Hallikainen <EraseMEharoldspam_OUTspamTakeThisOuThallikainen.org> wrote:
>
> > On 2/8/07, David VanHorn <dvanhornspamspam_OUTmicrobrix.com> wrote:
> >
> >> > We're moving to Atmel processors
>
> I'm not familiar with Atmel processors. Why are you moving to them?

Price, performance, features, the EE likes them

Downside:  They tend to obsolete processors regularly.

Upside:  Upwards compatible processors when they do go obsolete.

Bill

--
Psst...  Hey, you... Buddy...  Want a kitten?  straycatblues.petfinder.org

2007\02\08@161607 by Stephen R Phillips

picon face

--- William Couture <@spam@bcoutureKILLspamspamgmail.com> wrote:

> Problem is, I can see our EE saying that anything that connects to
> the processor needs opto-isolators (he's already said that in
relation
> to the machine I/O), and that, as a result, this will be too big to
> fit.
>
You can fit the opto isolation for I2C/SPI on the main board that it
connects too. Give it an isolated power supply. This isolates the board
but doesn't put the space cramps on the user interface.  Everything is
a trade off as they say.  If the EE wants isolation sure it's not
really that big of a problem that way.  Either way it can be isolated
or not the board would remain the same just the main system has to
adapt some.


Stephen R. Phillips was here
Please be advised what was said may be absolutely wrong, and hereby this disclaimer follows.  I reserve the right to be wrong and admit it in front of the entire world.



____________________________________________________________________________________
Don't pick lemons.
See all the new 2007 cars at Yahoo! Autos.
http://autos.yahoo.com/new_cars.html

2007\02\08@161649 by Harold Hallikainen

face
flavicon
face

> On 2/8/07, Harold Hallikainen <KILLspamharoldKILLspamspamhallikainen.org> wrote:
>>
>> > On 2/8/07, David VanHorn <RemoveMEdvanhornTakeThisOuTspammicrobrix.com> wrote:
>> >
>> >> > We're moving to Atmel processors
>>
>> I'm not familiar with Atmel processors. Why are you moving to them?
>
> Price, performance, features, the EE likes them
>
> Downside:  They tend to obsolete processors regularly.
>
> Upside:  Upwards compatible processors when they do go obsolete.
>
> Bill

Thanks! I'm generally able to do what I want with a PIC, though I could
sometimes use more MIPS (the project in front of me is running 40MIPS) and
more UARTs.

Harold

--
FCC Rules Updated Daily at http://www.hallikainen.com - Advertising
opportunities available!

2007\02\08@181755 by Bob Axtell

face picon face
William Couture wrote:
{Quote hidden}

the problem NOW with front panels is that they must be capable of
handling 5KV ESD spikes. This is not a trivial
problem, since most toggle switches do not meet that. It means that
those neat exposed switches like you see
on the NASA panels of the various Apollo movies are no longer allowed
(won't meet UL/CE standards).

So all front panels I design now use lots of PB switches, either
membrane or thru the chassis but with a membrane
cover to simply cover the actual PB tab. The membrane cover is available
with a drain wire to solve most of the  ESD
problems.  To enter an analog value, punch it in with up/down PBs and a
small LCD display. use a PIC or ATMEL
to pass the data from the panel; I recommend serial transfer but not
necessarily RS232-TTL, it it needs a fairly precise
clock.

If you MUST use exposed switches,  technology has come to your rescue,
in that fast, cheap ESD suppressors are now
for sale from several people that crush those spikes without adding a
lot of capacitance to the line.

--Bob

2007\02\08@184612 by David VanHorn

picon face
>
>
> the problem NOW with front panels is that they must be capable of
> handling 5KV ESD spikes. This is not a trivial
> problem, since most toggle switches do not meet that. It means that
> those neat exposed switches like you see
> on the NASA panels of the various Apollo movies are no longer allowed
> (won't meet UL/CE standards).


Imagine the fun we had with mag heads for card readers.
Dissipating the charge relatively slowly, as the user approaches, is better
than just letting him draw an arc to ground.

When the mag head rubs paper, it's easy to build up a few kV right on the
head, which can have interesting effects on your amplifier.

For front panels, we used elastomer keypads and ABS enclosure/keytops
designed to maximize the path length for discharge, and drain wires to take
any leakage to ground through 1k resistors.

More... (looser matching)
- Last day of these posts
- In 2007 , 2008 only
- Today
- New search...