ABC of electronics terms
(To find chips for distinct functions, see the functional pages.)
A chip that isn't produced anymore
Original Equipment Manufacturer
See also CEM.
One wire bus protocol by Dallas.
This section became to big, so see local.
Early TTL chips used to have an internal transistor at their output that could either
provide 0V or it would have a high impedance. To use it for normal digital logic
purposes an external transistor was needed to pull up the output voltage to 5V (nomally).
To calculate which value the resistor has to have consider that when the output is
pulled low by the chip, the resistor will be a short from the 0V output to 5V.
Therefore the resistor value needs to be as high as possible. In the other case the
resistor value must be low enough to provide 5V through it to the input of the
circuitry that is connected to the output of the chip. You'll need to know what
the internal resistance of this circuitry is or what amount current it draws.
You can also determine the value of the resistor experimentally. Start with a very
high value like 1M or 100K and divide the value by 10 until the resistor manages
to pull up the voltage at the output of the chip to well enough over 2.7 V (for
traditional TTL). Beware: In some applications you will also need to consider the
rise time of the signal. Instead of pull-up, pull-down is also used sometimes,
but it's much less common.
The open-collector system has still very valid uses, because you can connect a
lot of open-collector outputs together and they'll form a natural inverted OR* port.
If one or more of the ports is low the resulting signal will be low. Or put otherwise,
only when all the outputs are 'high' (=high impedance) the resulting signal will
be high. This OR* port mechanism has two very useful applications: For connecting
a lot of interrupt signals together, because often new devices are connected to
a computer system over time and the second usage is when connecting a lot of
systems over a bus where often also devices are added at a later stage and the
devices have to be able to talk in turn bidirectionally over a signal line.
The line can also be used to determine which device is allowed to take over the
bus next. At a certain stage of the protocol they will all send their own ID
in binary over the line starting with the most important bit and as soon as a
device notices that at a moment that he was sending a 1 but the line was 0, and
thus pulled down by another device it will stop competing. This way at the end
only one device is left (when all devices have unique numbers) and it may take the
bus. Once it's ready the other devices will compete all over again and the device
with the next lowest number will win etc.
Philips' I2C bus uses an open collector bus.
By the way, many modern micro* controllers have output pins that can also be
set to work as inputs. When they are used as inputs they are high-impedance
making them effectively tri-state ports. It's now very easy to implement an
open collector bus, since you just switch between 0V output and input to get
the two desired states. And during 'high' you can measure if the line was pulled
down by another device... Don't forget to add a resistor as a pull-up to the system! ;-)
Ah and if you want to be able to detect a shortage on the line, you'll also need to
be able to check if high is really high. With most micro* controllers you'll need to
use an extra input because you can't read the port as input when you're using it
as output. :-(
It's very easy to make mistakes with these kinds of ports when you're using
bit or-ing and and-ing (or the controller does this itself when you're using
bit set and reset instructions) because when reading back the port you'll usually
don't get the previous state of the output ports but the inputs...
It would have been much better when the chip designers had used different bytes
for the data of the ports for the input and the output functions of the port.
Besides 'open source software' there are also project groups working on 'open source hardware'.
Often abbreviated to OpAmp*.
See next item.
From: Frank Fremerey
To: Jaap van Ganswijk
Subject: Electronic Optic Chip breakthrough report
I finally found my note on the above subject. It reads:
"Raphael Tsu, an electrical engineering professor and Qi Zhang, a
postdoctoral fellow, say they've successfully linked electronic and
optical technology on the same silicon chip"
Both live and work at the University of North Carolina at Charlotte
(UNCC) and say, their discovery could make chips 100.000 times faster.
They further say, that a lot of unsuccessful experiments tried this
before, but they have now made it. The money comes from the Naval
Research and Army Research Office.
The thing was reported by GMSV on 19970916 (www.sjmercury.com) and further
reading is to be found in SCIENCE-magazine and I will cover the findings in my article.
See Operating System
Making a MPU or MCU run faster than intended by the manufacturer.
Faster processors from Intel to replace older slower ones.
||The upcoming AMD Athlon SocketA Thunderbird processor with performance
||enhancing cache memory, cooled to a chilly -40°C will give you the high
||performance processing power you could only imagine until now.