David's Blog

The last time I owned a desktop computer was either April, May or June 2007. I sold it in one of those months so I could make the last payment to Team Challenge and go to Malawi for a month for a mixture of project work and trekking. It really was a great time with a great team. For the last two and a bit years then, I’ve been without a desktop machine and have just had a laptop, which until now has catered for my requirements quite well. However over the last few months my requirements have changed such that the laptop wasn’t really cutting it. I would have replaced it with a newer laptop, because they are smaller, lighter, quieter and generally unobtrusive in that way. I then realised just how significant the price difference for a given specification is between laptops and desktops and decided to instead build a desktop. So, last Monday I ordered some parts. They arrived last Tuesday and I put them together between various other things like supervisions. It would be fair to say it has a little bit of memory. I’ll have to make it last a good few years!

Pictures:

[View with PicLens]

Once again (and this is the last one in the series!) this post is a Systems & Control coursework of mine from a couple of years ago – my A2 major project (November 2007 – April 2008). I’m publishing now because publishing my own coursework while still doing the A-Level it was for wouldn’t have been a good idea.

I would recommend having a quick flick through the full documentation – read the whole thing if you want but you probably have a lot of things that are more worthwhile!

iBantumi was the name of the major project I did for the A2 part of my Systems & Control A-Level. Over the summer between AS and A2 Levels, I spent a month in Malawi on a Team Challenge expedition with people from school. While there I discovered the family of Mancala games and bought myself a wooden Bao board. The games are great, but particularly with some variants you can end up spending a lot of time doing repetitive seeding. It was in fact while I was still in Malawi on the expedition that I decided I would make an electronic board to automate some of the more repetitive parts of play for my A2 project, and so I created iBantumi.

iBantumi as the name suggests is an electronic version of a particular Mancala game – Bantumi which I had played years before on my Nokia 5510 – rather than the slightly larger and more complex Bao (like the wooden board I had bought). Initially I planned to recreate Bao but I would have required in excess of a thousand LEDs to do so, which I thought perhaps a little unrealistic for an A2 project given the time constraints and limited equipment (would have needed to go surface mount for that really). Bantumi is a simpler game, so I only required 336 LEDs for iBantumi.

At the heart of the project are 14 Microchip PIC16F882 PICS programmed in assembly to function as serially controlled “pit registers” – essentially allowing me to control the 336 LEDs individually over one wire from the main controlling PIC – a PIC18F252. The LEDs are grouped into pits of 24 LEDs. The main controlling PIC was programmed in C. The beauty of iBantumi is that the player only has to choose which pit to initiate their turn with and the main controlling PIC will then automate the rest of the move.

Reflecting on iBantumi as a project, my time management of it was significantly better than that of the f00n DS4 - I finished most of the project a few weeks before the deadline, leaving me a decent enough amount of time that I should have been able to develop a decent computer player and LCD menu system for it. Unfortunately the PIC18F252s I was using as the main controlling PICs all failed at about this time in a way that meant I could not flash new programs onto them, effectively ceasing development entirely until I was able to buy some more – which happened to be the day before the deadline. I knocked up an AI player in the lunch hour of the deadline day but was unable to do anything about LCD menus, so the LCD remains unused. I got full marks for the project.

I designed the case using PTC ProDESKTOP, which was very useful in allowing me to cram everything into the smallest case possible without interference between components. Some of the photo realistic renders are included in the pictures below.

Warning – the documentation is by no means lightweight, but should be a brilliant guide if you want to build something similar out of the same components.

Full documentation: iBantumi (electronic Mancala game) (48)

Pictures:

[View with PicLens]

Again this post is a Systems & Control coursework of mine from a couple of years ago – my AS major project (December 2006 – March 2007). I’m publishing now because publishing my own coursework while still doing the A-Level it was for wouldn’t have been a good idea.

I would recommend having a quick flick through the full documentation – read the whole thing if you want but you probably have a lot of things that are more worthwhile!

Bizarre names aside (f00n was the name of a Counter Strike: Source clan myself and a few friends ran, and DS4 stands for Dream Sleeper 4 – four being the number of artwork versions the project went through), the project was and still is a very useful electronics product to me. I was inspired to build it after having missed the bus to school one time too many – traditional alarm clocks are such a rude awakening I would often end up switching them off and going back to sleep.

The f00n DS4 has a 50W halogen bulb and a high airflow fan that are slowly brought up from 0 to 100% over a user configured length of time, smoothly waking the user up. The concept actually works and rather well. At the heart of the system is a PICAXE 18X, a Dallas Semiconductor DS1307 Real time clock and a backlit 16×2 LCD. The PICAXE communicates with the DS1307 by I2C and the LCD by a serial line. A coin cell provides backup for the DS1307 when main power is removed so that time is still kept and user settings remembered (the PICAXE stores settings in the general purpose memory of the DS1307). The potentiometer on the front allows the lamp to be controlled manually as a reading lamp.

Looking back on it now, yes there are some things that could be improved (such as the casework quality and making the backlights switch off after a timeout). It was done as an AS coursework and I had strict deadlines to work to – given this, I’m pleased with what I managed. I got full marks for this.

Warning – the documentation is by no means lightweight, but should be a brilliant guide if you want to build something similar out of the same components. I’ll point out now there is no CAD model of the case due to the electronics and software development overrunning on this project – this meant the casework was a bit rushed.

Full documentation: f00n DS4 (Heavy Sleeper's Alarm Clock) Documentation (67)

Pictures:

[View with PicLens]

I think the only useful purpose this blog can serve is as an outlet for designs and projects of mine – when I ran my last blog, Facebook and Twitter either didn’t exist or were unheard of, but this being 2009, they are best for most non-technical things. With this in mind, my first post with real content is as the title suggests, a partial design for an electronic Peltier Element Controller.

I designed this about a week ago. The intended use was as a means to maintain a 72 pint barrel of beer at optimum temperature, but the design also allows it to heat the body it is attached to.

Feature summary:

* Microchip PIC18F452 controller
* NXP SE95D temperature sensors
* Automatic temperature regulation
* Barrel temperature sensor
* Peltier hot plate temperature sensor
* Peltier cold plate temperature sensor
* Heatsink temperature sensor
* Ambient temperature sensor
* Peltier current sensor
* LCD showing above data and the used controlled target temperature
* Automatic control of heatsink fan speed based on temperatures to minimise noise
* Allows Peltier current to be varied allowing development of Proportional-Integral-Derivative Control

As I said, the design is partial. I stopped after creating the schematic because the cost was going to be to high to justify producing it. This means there is no program written for it. Most of the schematic is pretty self explanatory. The temperature sensors interface with the PIC by I2C. The fan is driven by a PWM controlled FET.

The part that needs some explanation is the actual Peltier driver stage. RM1 is the Peltier. The Peltier current is controlled by a PWM driven FET, and is monitored using a network of five 20W 50m? power resistors, for a total resistance of 10m?. The potential at the top of these is amplified by a non-inverting op amp, which in turn is connected to one of the PIC’s ADC inputs. The power resistor values and op amp resistor values should allow Peltier currents of up to 50A to be measured.

In order that the circuit can drive the Peltier both ways (heating and cooling a body), there is a DPDT relay included to allow the current through the Peltier to be reversed. Ideally I would have  used an H-Bridge in place of this DPDT relay and the single FET, however I decided to compromise with a relay to avoid the additional cost of four FETs & heatsinks and in particular the need for a FET driver chip (needed for the charge pumps that they so often include).

The choice of PSU, Peltier, op-amp, FETs and LCD is left to the reader, as also is the PCB layout and design and construction of case and cooling unit (should you choose to separate the controller).

Schematic: Peltier Element Controller (Beer Cooler) (137)

Preview (click to enlarge):