Automotive temp MCU

In message <4016DCE0.524F2D66@yahoo.com>, Ralph Malph <noone@yahoo.com> 
writes
>Syd Rumpo wrote:
>>
>> In message <4016B281.7BB2760A@yahoo.com>, Ralph Malph <noone@yahoo.com>
>> writes
>>
>> [snip]
>>
>> >I am building a board that will run at up to 85C, but it will shut
>> >itself off at 85C and survive up to 125C.  The part that controls the
>> >on/off (which also includes a low power mode with multiple wake up
>> >options) is simplest if I use a small MCU.
>>
>> An 85C PIC will easily run at well over 125C - just don't use the
>> internal RC oscillator or watchdog.  Are 125C rated parts any different?
>> Don't think so.  Try it.
>
>Yes, they are different, they are guaranteed to meet spec at 125C if
>nothing else.

Just words.  Words which may be important, depending on liability, but 
that's not for me to judge.  So you get one of these parts which is 
guaranteed to work at 125C and it fails... Money back?

Fact is, the 85C parts will work, but it's then *your* responsibility.

-- 
Syd

Syd Rumpo wrote:
> 
> In message <4016DCE0.524F2D66@yahoo.com>, Ralph Malph <noone@yahoo.com>
> writes
> >Syd Rumpo wrote:
> >>
> >> In message <4016B281.7BB2760A@yahoo.com>, Ralph Malph <noone@yahoo.com>
> >> writes
> >>
> >> [snip]
> >>
> >> >I am building a board that will run at up to 85C, but it will shut
> >> >itself off at 85C and survive up to 125C.  The part that controls the
> >> >on/off (which also includes a low power mode with multiple wake up
> >> >options) is simplest if I use a small MCU.
> >>
> >> An 85C PIC will easily run at well over 125C - just don't use the
> >> internal RC oscillator or watchdog.  Are 125C rated parts any different?
> >> Don't think so.  Try it.
> >
> >Yes, they are different, they are guaranteed to meet spec at 125C if
> >nothing else.
> 
> Just words.  Words which may be important, depending on liability, but
> that's not for me to judge.  So you get one of these parts which is
> guaranteed to work at 125C and it fails... Money back?
> 
> Fact is, the 85C parts will work, but it's then *your* responsibility.

*Now* you understand...

Its also marketing.  Selling a board for 125C operation and using chips
clearly marked for 85C that control power to the rest of the board can
be hard to explain.  I'd rather let my customers have confidence in the
chips that I pick rather than asking them to have confidence in me (an
unknown to many of them).

Ralph Malph wrote:

> Alan wrote:
> 
>>>Did I miss any good candiates?  I am shy of the Asian companies because
>>>they can be very hard to buy from, much less get support.

> Thanks for the reply, but the temperature is what this question is all
> about.  I can find lots of chips that meet all the other requirements
> including the tiny QFN package and the low price.  
> 
> I can also find parts that are automotive temp, but they fail in one of
> the other requirements.  I found one that is only slightly out of my
> price range and is about twice as large as a 28QFN.  So I can use it if
> I have to.  I am just trying to cover all the ground to make sure I have
> not missed anything.  
> 
> Too bad Atmel does not have an automotive temp AVRs.  I would love to
> use one of them.

Take a look at the Zilog eZ8 family ?

If you like the AVRs, imagine adding a register Frame pointer, and
some efficent direct memory opcodes, and you are close to a Z8

eZ8's come in extended temp (-40'C to 105'C), which is a gain on your
"sleep above 85'C" - they start at SSOP20, and go up to 80 pins.

  System wise, I'd design in as much thermal inertia as possible,
get power-paranoid to reduce the self-heating, and record/log
the actual temperatures, as well as do a memory checksum and train (sic) 
the customers they need to replace your modules after so many LogT * 
Hours product.

-jg

Ralph Malph wrote:
>Thanks for the info.  The P89LPC932/3 seems to be about exactly what I
>am looking for other than the temperature.  Motorola seems to be big in
>automotive temps, but it is a real PITA to use their web site.  I
>finally found a couple of candidates there.  I seem to recall that
>Motorola does not make it easy to get a development system, you have to
>shell out a few bucks.  

Motorola are very good for automotive temperature parts.  Visit
http://www.mot-sps.com/ to bypass the non-semi stuff.

I don't think you'll find much for under $3 though.  A reprogrammable
HC05 of HC11 will cost you significantly more than that.

I think a PIC may be your best bet.  They work reliably at 125 ambient
and fit your price requirement.

Andy

On Wed, 28 Jan 2004 09:32:46 +0000, Andy Sinclair <me@privacy.net>
wrote:

[...]
>
>Motorola are very good for automotive temperature parts.  Visit
>http://www.mot-sps.com/ to bypass the non-semi stuff.
>
>I don't think you'll find much for under $3 though.  A reprogrammable
>HC05 of HC11 will cost you significantly more than that.

I think you may need to look at a more recent price list (or maybe
just a more recent part).  Newark's single-piece price for a
68HC908QT4 in -40 to +125 temperature grade is US$2.99.  That drops to
$1.97 when you buy 25. I can't give you Digikey's price because they
carry only commercial-grade (-40 to +85) parts.

>
>I think a PIC may be your best bet.  They work reliably at 125 ambient
>and fit your price requirement.

The QT4 gives you a much nicer core and more Flash (4k vs. 1k)and RAM
(128 bytes vs. 64 bytes) than a PIC12F675, but the I/O is not quite a
flexible.  Digikey's price for the PIC in the extended temperature
range is $2.30 for 1, $1.48 for 25.  I can't give you Newark's price
because they don't carry the right temperature grade.  And so it
goes...

Regards,

                               -=Dave
-- 
Change is inevitable, progress is not.

Andy Sinclair wrote:
> 
> Ralph Malph wrote:
> >Thanks for the info.  The P89LPC932/3 seems to be about exactly what I
> >am looking for other than the temperature.  Motorola seems to be big in
> >automotive temps, but it is a real PITA to use their web site.  I
> >finally found a couple of candidates there.  I seem to recall that
> >Motorola does not make it easy to get a development system, you have to
> >shell out a few bucks.
> 
> Motorola are very good for automotive temperature parts.  Visit
> http://www.mot-sps.com/ to bypass the non-semi stuff.
> 
> I don't think you'll find much for under $3 though.  A reprogrammable
> HC05 of HC11 will cost you significantly more than that.
> 
> I think a PIC may be your best bet.  They work reliably at 125 ambient
> and fit your price requirement.

Thanks for the info.  I have looked at the Motorola stuff.  As you say,
they have good parts, but they are not in the price ball park.  I may
revisit them just one more time.  I seem to remember looking at some of
their larger 8 bit parts before and found that their ADCs could not work
with an external reference and don't have an internal one.  It uses the
Vdd level as the ADC reference.  I tried to discuss this with them and
the problems it can create since most Vdd sources are not better than
about +-3% once you add in the tolerance of the set resistors.  But they
seemed to think this was not an issue.  So what is the point of having a
10 or 12 bit converter if it is only accurate to 5 bits?  Heck, I am
trying to measure the Vdd with this - it would always read the same
level!!!  I think the COP parts had the same problem.

Dave Hansen wrote:
> 
> On Wed, 28 Jan 2004 09:32:46 +0000, Andy Sinclair <me@privacy.net>
> wrote:
> 
> [...]
> >
> >Motorola are very good for automotive temperature parts.  Visit
> >http://www.mot-sps.com/ to bypass the non-semi stuff.
> >
> >I don't think you'll find much for under $3 though.  A reprogrammable
> >HC05 of HC11 will cost you significantly more than that.
> 
> I think you may need to look at a more recent price list (or maybe
> just a more recent part).  Newark's single-piece price for a
> 68HC908QT4 in -40 to +125 temperature grade is US$2.99.  That drops to
> $1.97 when you buy 25. I can't give you Digikey's price because they
> carry only commercial-grade (-40 to +85) parts.
> 
> >
> >I think a PIC may be your best bet.  They work reliably at 125 ambient
> >and fit your price requirement.
> 
> The QT4 gives you a much nicer core and more Flash (4k vs. 1k)and RAM
> (128 bytes vs. 64 bytes) than a PIC12F675, but the I/O is not quite a
> flexible.  Digikey's price for the PIC in the extended temperature
> range is $2.30 for 1, $1.48 for 25.  I can't give you Newark's price
> because they don't carry the right temperature grade.  And so it
> goes...

Thanks for the info Dave, but this chip only comes in 8 pin packages.  I
need at least 20 pins and am more comfortable with 28 (20+ IOs).  The
Motorola site lists the 1k price as about a buck, but then a lot of 8
pin MCUs are a buck.  

MC68HC908JL3  - LQFP48 - <$3  - no dedicated serial IO
MC68HC908JL3E - LQFP48 - <$3.57  - no dedicated serial IO
MC68HC908GZ8  - LQFP32 - $4  - SPI
MC68HC908EY8  - LQFP32 - $(can't find)  - SPI

So there seems to be some Motorola options, but they are no smaller than
the PIC as Motorola does not have the small packages at this time and
all the more cost effective parts (less Flash and RAM) also have too few
IOs.  They don't seem to use any TSSOPs above 16 pins and they don't
have any QFNs listed at all that I can find in the Q1 '04 selection
guide.  So without a smaller package, they would have to have a very low
price to be a better choice than the PIC.

Ralph Malph <noone@yahoo.com> writes:
> So there seems to be some Motorola options, but they are no smaller than
> the PIC as Motorola does not have the small packages at this time and
> all the more cost effective parts (less Flash and RAM) also have too few
> IOs.  They don't seem to use any TSSOPs above 16 pins and they don't
> have any QFNs listed at all that I can find in the Q1 '04 selection
> guide.  So without a smaller package, they would have to have a very low
> price to be a better choice than the PIC.

I'm baffled by all the postings I've seen on this thread.
In one of them it was said that the application was instru-
mentation for a locomotive, but there's an obsession with
finding the physically smallest MCU possible; given the size
of most railroad locomotives, it's hard to imagine size
being particularly critical.

It was not stated as to whether test instrumentation or
permanent, operational instrumentation was meant, but if
the former, it's hard to imagine a slight difference in
price of an MCU being much of a factor;  if the latter,
given the multi-million dollar price of locomotives, a
dollar or two difference in the price of an MCU is negligible.

Another posting said something about wanting to monitor Vdd.
This would seem to be a matter of trying to measure a voltage
with the voltage being measured, regardless of how one goes
about doing it.  It's also hard to imagine a need to monitor
Vdd (except for operating/fail) in any event.

On Thu, 29 Jan 2004 08:57:39 PST, the renowned mojaveg@IWVISP.com
(Everett M. Greene) wrote:

>Ralph Malph <noone@yahoo.com> writes:
>> So there seems to be some Motorola options, but they are no smaller than
>> the PIC as Motorola does not have the small packages at this time and
>> all the more cost effective parts (less Flash and RAM) also have too few
>> IOs.  They don't seem to use any TSSOPs above 16 pins and they don't
>> have any QFNs listed at all that I can find in the Q1 '04 selection
>> guide.  So without a smaller package, they would have to have a very low
>> price to be a better choice than the PIC.
>
>I'm baffled by all the postings I've seen on this thread.
>In one of them it was said that the application was instru-
>mentation for a locomotive, but there's an obsession with
>finding the physically smallest MCU possible; given the size
>of most railroad locomotives, it's hard to imagine size
>being particularly critical.
>
>It was not stated as to whether test instrumentation or
>permanent, operational instrumentation was meant, but if
>the former, it's hard to imagine a slight difference in
>price of an MCU being much of a factor;  if the latter,
>given the multi-million dollar price of locomotives, a
>dollar or two difference in the price of an MCU is negligible.
>
>Another posting said something about wanting to monitor Vdd.
>This would seem to be a matter of trying to measure a voltage
>with the voltage being measured, regardless of how one goes
>about doing it.  It's also hard to imagine a need to monitor
>Vdd (except for operating/fail) in any event.

Sounds like a batter-powered instrument of some kind, maybe
specialized data logger. 

Best regards, 
Spehro Pefhany
-- 
"it's the network..."                          "The Journey is the reward"
speff@interlog.com             Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog  Info for designers:  http://www.speff.com

"Everett M. Greene" wrote:
> 
> Ralph Malph <noone@yahoo.com> writes:
> > So there seems to be some Motorola options, but they are no smaller than
> > the PIC as Motorola does not have the small packages at this time and
> > all the more cost effective parts (less Flash and RAM) also have too few
> > IOs.  They don't seem to use any TSSOPs above 16 pins and they don't
> > have any QFNs listed at all that I can find in the Q1 '04 selection
> > guide.  So without a smaller package, they would have to have a very low
> > price to be a better choice than the PIC.
> 
> I'm baffled by all the postings I've seen on this thread.
> In one of them it was said that the application was instru-
> mentation for a locomotive, but there's an obsession with
> finding the physically smallest MCU possible; given the size
> of most railroad locomotives, it's hard to imagine size
> being particularly critical.

This board is for multiple applications.  The locomotive app was one
that illustrates the need for high temperatures.  

> It was not stated as to whether test instrumentation or
> permanent, operational instrumentation was meant, but if
> the former, it's hard to imagine a slight difference in
> price of an MCU being much of a factor;  if the latter,
> given the multi-million dollar price of locomotives, a
> dollar or two difference in the price of an MCU is negligible.

Again, this board is to be sold for multiple apps.  A dollar may not
make a big different in the end cost, but if selling 10,000 units over
the next few years, a buck parts savings will provide an additional
$10,000 in profits.  Certainly it is worth a thorough examination of the
available parts.  


> Another posting said something about wanting to monitor Vdd.
> This would seem to be a matter of trying to measure a voltage
> with the voltage being measured, regardless of how one goes
> about doing it.  It's also hard to imagine a need to monitor
> Vdd (except for operating/fail) in any event.

Since when is Vdd used as a reference for measuring *anything*?  That
was my point.  Some of the chips use Vdd as the refernce for the on chip
ADC.  With most Vdds being 2 or 3% accurate, this seriously degrades the
performance of the ADC.  Most boards these days have multiple power
supplies on board.  This board will have five DCDC converters and will
have two separate 5 volt sections.  

I find that most engineers want to rework the problem that is being
solved.  I have minimized the solution to either one chip or two.  Now I
am just trying to find the best choice to suit that decision.  In any
case, this chip must meet automotive temps.  

I have been told that the Cypress PSOC chips will be available in
automotive temp range shortly.  If I can confirm that, this may be a $2,
one chip solution.  Smallest size (QFN), lowest cost and fewest chips; I
see that as worth a few days of searching.