Electronics Forum / Basics / September 2006

Hello Michael,

Seriously, as Linnix wrote, you may not need one. I started with the TI
MSP430, thought about buying a book but after taking a look at one I
realized how fast some of the contents become obsolete. By the time it's
printed there is a slew of new uC variations that are now used by everyone.

Main things to get started: Learn the design environment, play around,
try to make the blinky-LED work, try to change blink rates or pattern,
load again and so on. Then study the family guide and the data sheet for
a long, long time. After that you know all the assembler words the uC of
choice understands and what they mean. Even if you never do assembler
you need to become familiar with the peripherals on board and how they
can affect or impair each other.

If you really want a book recommendation I'd post again at
comp.arch.embedded where among many uC experts there is also Ulf
Samuelsson from Atmel. I am sure he'd know which books are good. Ulf is
very responsive, I wish TI and others would participate as well.

In sci.electronics.design jrwalliker@gmail.com wrote:

What's the remedy for such situation ..?

Yes, a Max232

Here is a battery charger with RS232 and LED read out.   The LEDs are
time multiplexed and maintain a rather constant brightness across
voltage range of 2.4V to 2.8V (two NiMH cells).  It's using the
Atmega169, but should work with 165 (cheaper) as well.  The basic
components are:

165/169
Max232 w/ 5 Cs
Analog power filter (L, C and 2 R)
LM317

SMPS is optional
Analog multiplexer is not needed.

http://linnix.com/proto

In sci.electronics.design mrdarrett@gmail.com wrote:
[snip]

The ATtiny11 has no RAM, only registers. C really needs some RAM for its
stack, or you have to do some funky things to make it work at all. You
can't have much fun with just 32 registers to store all your data (this
paragraph contains more than 8x that much data). The small ones really
are incredibly limited.

Best advice for a hobbyist: Buy big. Considering the cost of shipping,
it's a false economy to buy the really low-end chips. You'll end up
making another order very soon when you find your first project is
larger than you expect. What you though was a saving ends up as being
more profit for FexEx.

I'd suggest an ATmega8 (or two) and a few matching sockets - you can
then swap the chip from project to project (no need for the sockets if
you don't plan to solder stuff together yet, a breadboard is one big
socket!). The ATmega8 has a UART, which means it has hardware to make
talking to a PC's serial port easier[1]. It has a good amount of RAM
and flash, which means you can use things like C's printf function to
make life easier (printf and associated gubbins takes up around 2.5K of
flash). It has multiple timers, plenty of pins, PWM and a bunch of other
stuff you'll soon find you want.

[snipped a question]

You can use a comparator (gives a signal when one input has a higher
voltage than the other) as an ADC, but you have to use it to measure the
time it takes for the voltage you are measuring to charge up a capacitor
to some reference voltage. You have to provide the capacitor, reference
voltage, code to time it accurately, coded to turn that exponentially
varying number into a linear voltage and make sure the current you take
charging the capacitor doesn't affect what you're trying to measure
(this method has a low input impedance - it puts a significant load on
what you're measuring, changing it in the process). It takes two pins
and you can only measure one voltage with your one comparator. This is
not much fun, though it can shave a few cents of the price of your
device, which might be worth it if you want to make a million of them. I
doubt you want to make a million of them. Buy a chip with a proper ADC
built-in; no external components, very little code, more accurate
results, multiple inputs which take just 1 pin each and you can connect
the pins directly to virtually anything without it affecting what you
connect to (the ADC inputs have very high impedance). Don't connect the
inputs to anything outside the AVR's supply range (0-5V probably, but
you have some choice in the matter) though, or you'll fry it.

For the battery, connect - to circuit ground, + to an ADC input on the
AVR. The AVR's ADC can then measure the voltage with half a dozen lines
of code. That wouldn't actually make for a great battery tester; there's
more to testing batteries than just measuring the voltage, but that's
another subject.

Best to use the serial port, as that's by far the easiest. You might
need a USB-serial converter if your PC doesn't have serial ports.
Sending raw bytes is most efficient, but you'd then have to write
software for the PC to decode those bytes. Much easier to just buy a
nice big AVR and use printf to send ASCII text, at least at first. You
can then use something like Hyperterminal on the PC and get text output
on screen with minimum hassle. You can do clever, efficient protocols
and squeeze them into the really cheap AVRs when you've learnt some
more.

[1] You will need a MAX232 or equivalent chip to convert the AVR's 0-5V
serial port (UART) signals to the +/- 15V RS-232 signals a PC serial
port expects - remember to buy the 4 capacitors such chips require too,
if you don't have that kind of thing lying around. You need to think
about the cable from the chip to the PC as well.

If you haven't already, check out AVRFreaks. There is a lot more there
than there may appear to be at first.

http://avrfreaks.net/

Tim