2011-02-26
555 Adding Machine
This is my entry for the extreme category of the 555 Contest. It is a decimal adding machine built using 102 LM555s. A fusion of my long-time thoughts about building ring-counting nixie or neon bulb clocks and a childhood fascination with decadic dialling mechanisms.
Architecturally the machine is a 9 decade accumulator adding machine with an indexing counter driving a multiplexer to simplify data input. It can do addition or subtraction using 9's complement arithmetic and has optional carry-around. In a historical context it is an improved electronic implementation of the 1642 Pascaline.
Video
Build
Building the machine was a heck of a lot of work. It took about two weeks, a few hours per night.
Some PostScript was written first to produce the front panel, this was glued to a scrap of thick cardboard and punched for all the LED holes. The files are attached at the end of this article.
Strips of double-sided tape were laid down on the backside to hold the 555s and the chips placed. Then came the power supply rails and LEDs.
Current limiting resistors and pull ups went in next. Each stage had its transfer capacitors added, then the clock caps and global asynchronous reset diodes each with its own rail. Some BJTs for the clock, reset and carry chains, then some rail decoupling caps completed the decade rings.
The indexer and multiplexer was constructed in a similar manner, and the de-bouncers added last.
Circuit
Better version soon!
Stats
The quiescent device pulls 450 mA from its 9 volt supply. Yep, 4 watts! You can feel the warmth coming from the back of it. I could not find ICM7555s in quantity, so was forced to use LM555CNs, making it a rather power hungry device. The LEDs pull only about 70 mA total, the rest is just burnt up in the 555s.
- 102 LM555CN
- 119 1N4148
- 99 LEDs (90 Red and 9 Blue)
- 40 BJTs
- 30 2N3904
- 10 2N3906
- 353 resistors
- 217 110 k
- 121 1 k
- 15 10 k
- 232 capacitors
- 202 10 nF axial monolithic
- 19 10 uF electrolytic
- 10 100 nF MKT + 2 greencap
- 1 330 nF MKT
- 13 metres of wire
- > 2000 soldered joints
Notes
The device was built with experimentation and modular testing in mind. Each decade ring is independently resetable and most multi-drop signals have diode isolators. Similarly carry-chain defeat is supported should I want to break up the accumulator into several parts. As such it uses a few more parts than are absolutely necessary. You could easily omit some of the diodes and resistors, but it would have been much harder to test each component in isolation without the current topology.
The only notable issue I had building it was a frustrating reset bug that turned out to be a rail sagging problem. Additional decoupling at the edge of the board cured it. The 555s really crowbar the supply while changing state and at reset all of them draw a rather large pulse of current.
It can count at up to 100 Hz reliably. I often tested parts of it using mains noise on my body or a signal generator to clock it.
The component values are reasonable, but I had a great quantity of 1 K, 110 K and 10 nF devices so they were used the most.
Why do this? If nothing else it is a warm-up for my relay computer... I recently came into possession of 550 4PDT 48 volt relays. When I get around to building that it will be a far larger undertaking!
11 comments.
Attachments
| title | type | size |
|---|---|---|
| Index Front Panel Artwork. | application/postscript | 2.070 kbytes |
| Accumulator Front Panel Artwork. | application/postscript | 4.774 kbytes |
Updates
- 2011-02-27: Random Dialler for 555 Adding Machine
- A quick random dialling hack for the adding machine to demonstrate it at the Wyong Field Day.