Moved the project to Github: https://github.com/Dovgaluk/Relay
It's a bit faster than Sourceforge and thus more confortable to use.
The next step is wiring these components altogether.
Created cabinet for registers module:
I also bought cable for connecting modules, switches for ROM block, and 0.47 Ohm resistors to insert them into +V line. This is needed to allow connecting two different power supplies to the same power line.
Last two or three weeks I spent on designing registers module. I decided to make all the registers 8 bit width. I decided to extend 4-bit upper halves of the address registers to increase number of full 8-bit registers that could be used as ALU inputs/outputs.
All registers will be able to be connected to the both data buses. M and LR registers will also include relays for connecting them to the address bus.... read more
I published review of ALU creation process on habrahabr.ru http://habrahabr.ru/post/220865/ (in Russian)
I also added calculation of ALU creation cost into this publication. It was at level of 21670 RUR (about $600 at today's exchange rate).
After last post in the blog I made the following modifications of ALU:
1. All the brass plates were engraved, drilled, and cut.
2. Plates were installed on the main board.
3. Bought and installed plastic backside to the cabinet.
4. Screwed two hinges for hanging cabinet on the wall.
And here it is! Completely assembled ALU:
... read more
I've got some of the plates with text engraved.
Next I will drill the holes and cut the plate apart.
I tested all ALU functions. Now it looks that everything is almost ok (except the heisenbug with subtract block). Assembled ALU is presented on the following photo:
On this photo every relay is switched on and all LEDs (except one - which shows that result is zero) are turned on.
In this mode ALU consumes 4A of power.
Temperature inside the cabinet is about 40 degrees.
Today I finished wiring of the ALU module (excluding wires to output socket).
And I made several attempts to fix or replace faulty relays:
I almost finished wiring of the ALU module. After wiring arithmetic blocks' outputs to the result register I noticed, that some of the result bits do not receive values from the adder.
Here is the adder, which computed 0xff as a result:
Result register shows that two bits in 0xff are zeroes:
... read more
Another one relay computer: http://www.schlaefendorf.de/relaisrechner/dokumentation/index.html
This computer has extendable architecture and consists of multiple modules. It also uses semiconductor diodes for some logic operations, that is why it is not "fair" relay computer. But it's ALU is better than most of the other ones - it can execute 8 operations including SUB/SBC.
Yesterday I finally soldered two shifter modules of the ALU.
Both "modules" consist of wires and enabling circuit.
First module performs RCR/SHR (shift right with/without carry) operations, and second one is for ROR (rotate right).
Suddenly noticed, that first entry in this blog is dated 27-11-2012.
Now I've almost completed ALU module and made rough design of other components of the computer.
i² (i Squared) is a fully programmable 8-bit computer built of of relays: http://isquared.weebly.com/
It uses over 300 6v relays mounted on boards and has taken in excess of 1900 hours to design and construct. The design draws elements from several other relay computers built by Dr. Harry Porter (RC1), Jon Stanley (RC2), MCC (RC3) and Fredrik Andersson (Zusie).
According to the blog, work on this computer is still in progress.
Relay computer "trainer": http://relaysbc.sourceforge.net/
The idea behind this project is to make a low relay count, single board computer similar to the single-board trainers of the early microcomputer era which can be mass produced for a reasonable price.
CPU part of this computer consists of 83 relays. Other parts are based on ICs.
TIM relay computer: http://www.northdownfarm.co.uk/rory/tim/tim-8.htm
It consists of 152 relays and controlled by punch tape and DIP switches.
This computer is the "smallest Turing complete relay computers in the world by relay count (with maybe the exception of the 'DUO Premium')".
Today I finished soldering of the subtractor module.
Subtractor is one of the unique modules that is not presented in other modern relay computers (like Harry Porter's computer).
Bought four 200x200x0.3mm brass plates for making relay nameplates and M3 nuts&bolts for fastening plates on the main board.
The nuts are not made from brass, but they will be located at the hidden side of the plates and will not be visible.
After finding 3 failed relays within the subtractor module I decided to make a testing board for relays.
I checked about 30 relays using this board and all of them seem good. It seems strange because I found ~5 bad relays within set of ~50 already assembled.
I finally assembled the cabinet and inserted the ALU board into it.
The cabinet is made of merbau timber and finished with oil and wax.
The dimensions of the cabinet are the following: height - 74cm, width - 56cm, depth - 14cm.
After installing the board I found that it is not clean enough. Cracks around the holes are also visible.
Bought another 24V power supply and one socket for relay.
I will use them in relay testing module. And power supply block will later be reused in one of the computer modules.
Bought 100 relays SCLD-W-B-L-4PDT-C 24VDC with red LEDs.
Because of many failures of relays I plan to make testing circuit to test them before attaching and soldering.
Bought 4 planks (1.5m each) of merbau to make cabinets for ALU and registers modules.
Finished soldering and testing of ALU adder block. While replacing bad relay found another one with corrupted contacts. I probably need to make quality inspection before attaching relays to the board and soldering them.
I almost finished soldering of ALU adder module (and started using next 10m of wire) and found some strange bugs while testing it.
Some of them were caused by bad soldering (i.e. myself).
And others were really strange.
First one was caused by twisted wires inside the relay. I fixed it by switching inputs of corresponding pins.
... read more