VAX02: AVR experimental boards

VAX01 covered some ground, but it was too complex. Why add integrators on the analog input? Why use a reset controller and a canned crystal oscillator? Why, why why. The answer is easy: I want to be complete. I rather be safe than sorry. So I tend to build in failsafe mechanisms. VAX02 still has these failsafe mechanisms in place, but now they are tailored to the AVR and not to the Z80 anymore....

VAX is short for either Verhoeven's Avr eXperiments or for Versatile Avr eXperimentboard. But VAX is also a third rate brand of vacuumcleaners. And, more importantly, VAX is a trademark of Digital Equipment Corporation (DEC). VAX computers are famous among engineers. VAX computers run complex process industries. Oil refineries. Warehouses. Photopaper mills. The lot. If reliability is an issue, you needed a VAX to run your mill.

Primarily, VAX is short for one of the two abbrevations above. But not in the least, my board's name is a tribute to what was probably the best computer architecture an engineer could imagine. Although the VAX has been phased out by DEC and HP (who aquired DEC at some time) old VAXes never die. Many of them are still stoimping out millions of tons of petrol a year. Why change a process computer that will never quit? Why change VMS for a coloured shell that gives you a blue screen just because the moon is in the first phase?

Below is a table with explanations why things were changed in version 2 of my VAX.

Changed	Because of
RC integrators	In VAX01 I had RC circuits and zener diodes on each ADC input. This was a great idea, if the user was about to do only measurements with port A pins. But since pincount is always an issue, board space too and, last but not least, this would limit the freedom of the end user, I decided to strip RC and zeners from the PortA pins. Now the PortA pins are unprotected. But they come available to the user for all purposes, AND this allows the user to build his own level of protection on the analog inputs.
Oscillator	VAX01 had a canned crystal oscillator. In the good old days of slow microprocessors this was a good design practice. Spend some money on a crystal oscillator that will always start, run smoothly and is not too much temperature dependant. The AVR is less demanding here. A simple crystal and two 22 pF capacitors will do the same as a costly and board space consuming canned oscillator.
Reset	Originally I used an ADM 698 reset controller to keep the processor in a RESET condition when power supply voltages were in doubt. Again: a good design practice from the days of the Z80. Good then, still good now. But it offered a problem. The ADM 698 has a push/pull (totempole) output stage. Hence, a NO-RESET condition is enforced by a transistor. Hence, a modest state AVR programmer will not be able anymore to enforce a RESET condition too easily. And if it does, the 698 will 'see it' as a short circuit. So it was better to change the integrated circuit for a trusty old RC pair. I could have used it in combination with a TS803 resetcontroller (which is an open collector to ground) but decidced against it. Perhaps in VAX03. If ever.
MAX 232	Due to misinterpretation of the MAX232 datasheet I installed a capacitor between pins 2 and 6. This was wrong. Both pins need a capacitor to ground (according to TI). Maxim prefer to have a capacitor to ground on pin 2 and one to +5V on pin 6. In the new state, I use the TI configuration since it makes more sense. Also, I changed the circuit so that the PCB has identifications to use polarized capacitors. I drew in a MAX232 but a MAX202 will do fine as well. As long as you keep the optimal capacitor values in mind. A MAX232 circuit runs best with 10 uF capacitors. A MAX202 will do so with 100 nF caps as well. Yet, it loves to have some headroom so it prefers capacitors around 1 uF each. Keep this in mind. Either chip will work, as long as the capacitors are big enough. I run my experiments with 1 uF (35 Volts) solid tantalum capacitors installed.
RS232	In VAX01 I used two IO pins to interface with the serial connector. This is rather old fashioned. So I hardwired the handshake signals, thereby freeing up two more IO pins.
IRQ's	In VAX01 the IRQ's (INT0 and INT1) were available on one 3-pin header. This was not too logical, so I changed it for the better. Now both IRQ's are available on two 2-pin headers. If you want to change the IRQ for some reason, just reconnect the headers...
I2C	Originally the I2C (=TWI) interface connector had three pins (SDA, SCK and ground but that's a rather stupid layout. I2C will generate an interrupt on lots of interesting events. So now the I2C connector is 4 pin: INT1, GND, SCL, SDA. This will relieve the processor from ever having to poll the I2C bus.
ICSP	I forgot to put in series resistors in the ICSP signal lines. Corrected.
Power	I added a simple yet reliable and failsafe power supply. It consists of a simple DC jack and an automotive grade LDO (Low DropOut) voltage regulator. Just add 6 Volts of DC and you're done. The LM2940 regulator can take -48 volts of reverse connections.
LED	The two unused RS232 handshake signal lines were suddenly free. Yet I didn't want to add one more connector on the board perimeter. Since there was no optical feedback on the board yet, I decided it was time to add it. So I mounted a two pin, two colour LED (with a series resistor) between pins PD4 and PD5. Now you can change the state of the LED in three steps: OFF, RED, GREEN. Nothing beats a simple optical feedback device.