Nuts and Volts - November 2008

firmware, which then lead me to the usual story of the code growing too big for the Atmega8’s ROM space. To fix this one, I have changed to the AVR Atmega168 chip with its 16K ROM and identical pin-out.

A light show timer problem that got through the cracks (but didn’t create any email) involved running the program past midnight. Let’s say you start at dusk — 5: 30 pm locally — and you shut down at 10: 30 pm, for a five hour show. The data is easy to enter through the push switch and rotary encoder, and could be changed again once your light show is installed out on the lawn. What if you run it past midnight? Technically this is the “next day” and one o’clock in the morning is actually numerically smaller than 5: 30 in the evening, causing a firmware math error.

The clock display and math are in a 24 hour format, so that am and pm are never confused. Funny story ... once I accidently set an alarm clock to wake me up at dinner time instead of breakfast! The two BCD encoded data strings (hh:mm) are saved to the EEPROM in the RTC chip: one for start and one for stop. These are converted to minutes and compared to the RTC’s clock time (also converted to minutes from its native BCD hh:mm format). When the RTC sum is greater than the start sum, the show runs, and when the RTC sum is greater than the stop sum,

the show is idle. If the stop time goes past midnight,

I set a flag, telling the firmware to continue to the next day once the show is running. It’s now possible to have any length show from just one minute duration up to

11 hours, 59 minutes starting at any minute through the 24 hour day (both examples being rather silly).

My prototype controller PCB used the standard 10 pin header to connect an Atmel AVRISP for programming the AVR chip. Shortly after this project, I damaged my AVRISP and discovered the replacement (an AVRISP mark II) only uses the six pin header. So, my next AVR project used the six pin style header and I made a ribbon cable adapter to service the older 10 pin style projects. The AVRISP mkII also has a USB (instead of serial com) connector on the PC side, and has revised circuitry in the pod.

I found that the new AVRISP would reset the target AVR at random and I traced this to noise from my project’s 5V rail getting back into the pod. I added a 10K resistor and a 10 µF capacitor as decoupling, and that did the trick. Unfortunately, those values didn’t make it to the published controller BOM, which has the correct six pin header. I got quite a few emails wondering why there was an undocumented C4 and R25 on those new PCBs!