I wanted an exchange that would handle 10 dialed digits; 9 internal lines and one outside line.
It should be easy to repair with readily available parts. Avoid microprocessors, PLAs, or special-purpose integrated circuits if possible and practical.
If possible, it would use what I have in my junk box. Besides reducing my pile of junk, it would help with my last goal, low cost.
Low cost is always a goal. For a telephone exchange that has no real purpose other than entertainment of my grandchildren, cost is less important.
I searched the web extensively and found a number of exchanges. Most are now micoprocessor controlled. Tempting, but not what I wanted. On the other extreme was L. D. Gunn's 22-relay exchange ("Miniature, ten-line telephone exchange", Wireless World,August 1980). It used very expensive relays, had no outside line capability, and was less than clearly documented.
I was starting to redesign
Austin Hellier's Link 'P' exchange, which had 8 lines and relays with solid state driving circuits, but only a hint that it could be adapted for an outside line. Then I found Andrew Holmes' exchange. It was the inspiration for the telephone exchange I built.
Andrew Holmes' 8-line, all-solid-state, exchange had no outside line circuitry, but had a suggestion of how to link the PBX to the outside line.
You will see that I borrowed unmercifully from his design. I wanted to use old logic instead of PLAs, but a lot of the circuitry he has was adaptable to what I wanted. His is the best all-solid-state design I found.
Case for Lagging-Edge Technology
If you had built a filter with a National Semiconductor AF151 Dual Universal Filter integrated circuit, or a timer circuit with a LM122, or a counter with a combination BCD-decoder / seven-segment-display-driver, you would know the danger of using custom integrated circuits in a one-off design. Obsolescence.
The more complicated the IC, and newer the product, the most dangerous. Try to find a replacement part for a 10 year old application-specific part. But if you choose lagging edge parts, you can pick one that has a large number of users, which means there is a greater chance that replacements will still be around in 10 years.
The custom part used for a receiver years ago is very good at its purpose, but it is almost impossible to find now. The dual-FET TLP-598G photo isolator used in the all-solid state exchange referenced above is available at major retailers at a high price, but not available at the prices that make a DIY project feasible. I built a stepper-motor driver 12 years ago and used a motor-driver chip in it. That motor driver chip is not available now. I used an erasable prom to sequence the driver, but I haven't used the programmer in probably 10 years, and I'm not sure that if the prom died I could recover the code and get another chip programmed.
Yes, 74LS series logic is old, but it is likely to still be around in 10 years. Will a Stamp processor or a Programmable Logic Array be here in 10 years. No.
Cost of Building the Exchange.
There is no way you can justify building a rotary dial telephone exchange. You just do it for fun. One of my granddaughters showed facination with the my payphone, which spurred me to
make it act like a real payphone. That being done, I added
anti-sidetone circuits to my other old phones, and I had hoped to hook all my old phones to this exchange for her and her sister to play with. Sadly, my granddaughters moved to Tanzania before this project was finished. One will be to old to play with telephones when she returns, and the other one does not show any interest in them. So, it seem I am just doing this for fun and that justifies the cost and time.
Parts - Close to, but under $100.
Time –
design and schematics = 216 hours
layout = 60 hours
assembly = 100 hours
case = 18 hours
testing = 4hours
total = 338 hours
Labor cost - 338 hours times $12 (minimum wage) = $3056