The Cyber-Spy.Com Usenet Archive Feeds Directly
From The Open And Publicly Available Newsgroup
This Group And Thousands Of Others Are Available
On Most IS NNTP News Servers On Port 119.
Cyber-Spy.Com Is NOT Responsible For Any Topic,
Opinions Or Content Posted To This Or Any Other
Newsgroup. This Web Archive Of The Newsgroup And
Posts Are For Informational Purposes Only.
From: Chuck Simmons
Organization: You jest.
X-Mailer: Mozilla 4.61 [en] (X11; U; Linux 2.0.33 i586)
Subject: Re: Future power supply concepts for automotive electronics
Date: Sun, 22 Sep 2002 21:21:21 GMT
NNTP-Posting-Date: Sun, 22 Sep 2002 14:21:21 PDT
> Hello everybody,
> Most electronic control units in modern cars consist of an MCU (micro
> controller), some peripheral ICs and a power supply IC. When the car is
> not running the current consumption of the whole electronics must be as
> low as possible (<100uA). To achieve this the MCU is sent to STOP.
> Modern micro controller technologies feature transistors with 0.2um
> channe length. Sending these MCUs to STOP will not reach the current
> consumption target anymore because short channel MOS are too leaky and
> 5V ports (car manufacturers love 5V ports) can no more be built with
> 70Ae gate oxide without current mirrors anymore.
> I see the following consequences:
> 1. simple STOP mode is useless.
> 2. For low consumption the power supply must turn off when the
> alternator is not running.
> This leads to the following issues:
> 1. How will we wake up the system again (there is no more interrupt
> handler that is powered up and can handle 5V without current flow)?
> 2. What ports are anticipated?
> 3. Can we move the interrupt handling somewhere else and what will be
> the communication and data buffering requirements of the moved interrupt
The easiest solution is to retain 0.35 micron process for interrupts and
context. It is a no brainer when migrating to 0.18 micron process by
simply retaining the 0.5 micron oxide which is done anyway to get 5 volt
tolerance on inputs. It is not clear that a migration to 90 nanometers
makes any sense at all in automotive for essential systems. An engine
turning at 10,000 RPM need only be sampled at 1667 times per second (ten
cylinder engine) and engine controls can have at least 5 sample
response. Late 1970's technology is already overkill for engine control.
I should call attention to a gross error that GM made in the 1980s. They
made transmission controls respond too quickly. The result was that when
I flew into Boston and Avis tried to give me a GM car, I asked for a
Ford or Chrysler because of a sore back that could not take the beating
a GM transmission would give it. I was driven to this because of the
disgusting Pontiacs that fiddled their gears constantly on mildly hilly
Of course, I am talking about essential systems. Non-essetial systems
are another can of worms. The engine control system and the transmission
control system need to preserve very little context because most context
reconstructs quickly upon restart. Collision avoidance may require fast
processors but again, there is no need to retain context. On board
navigation is another problem since current context is slow to recover
and past information is occasionally useful. It still seems that storing
context in older technology and powering down the power hogs is the way
to go. Flash for context is OK because the life of a car is usually less
than 50,000 starts and some flash memories are good for at least 100,000
rewrites. That seems to be a fair match.
I suppose it could be slightly exasperating to punch the key in your car
with the strawberry on it (grandma's house) and have the car pipe up and
ask for a clue because it has forgotten the way.
... The times have been,
That, when the brains were out,
the man would die. ... Macbeth
Chuck Simmons firstname.lastname@example.org
Go Back To The Cyber-Spy.Com
Usenet Web Archive Index Of
The sci.electronics.design Newsgroup