Old school logic circuit

A

agent420

[H]ard|Gawd
Joined
Sep 6, 2001
Messages
1,388
Looking to build a shift register / counting chain using only relays for a potential mod. I'm going for the old school em (electromechanical) clickity thing. Probably some incandescent indicators rather than all high tech leds. Maybe a nixie tube if I feel fancy.

Circuit will be pulse fed (perhaps a cam or something) and should increment for each pulse, resetting at 10 (decade counter). Any contact style relay can be used (spst, dpdt etc.)

I first thought that it would be pretty easy, as relays are as basic a gate building block as you can get, but it's proving to be more involved than I anticipated. Either that or I am just not seeing the answer and over-complicating things.

Googled around, I'm surprised there is so little info on relay gates or circuits considering they were fundamental to several early digital computers.

A link or two or some circuit advice would be appreciatied.
 
Sorry I don't have any links, but look up a beginning digital logic text, and see how the various gates are formed using transistors. -- You can regard a relay as a transistor for most purposes, since they perform a similar switching function, and FWIW, an AND gate is basically just a transistor (or relay.)

Given that, you could just design a basic flip flop using relays, and use standard logic design processes to build yourself a shift register. (Karnaugh maps, next state maps, etc.)

Hope you have a cheap source of relays though.

and don't burn the house down;)
 
heh i just took a couple classes on this crap (shift registers and all) recently. i would actually suggest getting ahold of some college textbooks if you have access to them.
 
Yeah, I started out thinking I would break it all down to equivalent digital blocks. And I'm sure that could be done, but it would require a lot of relays. More than I really want ot use.

I think there are circuit tricks you can do with relays that have multiple contacts, like dpdt or even 4pdt, that would considerably cut the number of parts down. Transistors and the basic gate theory they teach today don't have that ability. I am sure the designers of the early computers had some tricks up their sleeves.

I am thinking there must be a way to get the parts count down to 12-14. Going the equivalent digital block route would probably require 3-4 times that many.
 
How many stages would you want?

You say shift-register / counting chain? Do you mean that you want to just have a series of D flip-flops chained together? This is kinda interesting, I'm going to flip through my digital texts and see if I can find something...
 
I'm sure there are some tricks the old school had, but they haven't been in use since.... then.

Ya know, if you wanted to be cheap and dirty about it, you could always just use a digital logic circuit to do the actual counting, and have relays connected to them to make the fun noises=)

Beyond that, consider it an opportunity to excercice your critical thinking and problem solving skills... good luck eh...
 
fat-tony said:
How many stages would you want?
Click to expand...
I need 10 stages. I think perhaps I did not use the correct terminology earlier... This may not be so much a shift register as more a simple sequencer. I need 10 lamps to illuminate in sequence driven by a input pulse. Compared to an led bar, I would prefer 'dot' vs 'bar' as in only 1 lamp lit at a time (1 - 2 - 3...), but if the 'bar' mode was able to be accomplished with a substantially lower part count, I would go for that.

Aristarchus said:
Ya know, if you wanted to be cheap and dirty about it, you could always just use a digital logic circuit to do the actual counting, and have relays connected to them to make the fun noises=)
Click to expand...
Yeah, I thought of that as well. I almost broke down and started to program an AVR uc, but now I am determined to have the circuit operate authentically... I think it would give the project more 'soul'.

Thanks for the replies. I have been looking into some old telephone circuit texts, the principles are similar and there is slightly more info available on that subject compared to computer logic of that era. Too bad, I bet there's a lot of lost knowledge on this kind of stuff.
 
Actually I have a clock source, it's just the sequencing logic I need to determine.

FWIW, this project will be a system uptime clock. My plan is to build a Regulator style pendulum timekeeper that utilizes an antique style electric circuit that both provides energy to the pendulem as well as the seconds pulses. A permanent magnet is embedded in the bottom of the pendulum. 2 fields coils are located below the pendulum; one is a sensing coil and the other creates a magnetic field that interacts with the pendulum magnet, providing energy to keep it in motion. The actual beat frequency will still be analog and dependant on the pendulum's natural rate. To keep things in line with the 'old school' philosophy, I will use vacuum tubes in the coil sensing and pulse generating circuit if needed, although it appears I will be able to accomplish this using entirely passive components, with mercury switches in the pulse circuit.

I don't have my links available at the moment, but I have found several examples for the pendulum construction. This type of clock movement was especially popular in old master-slave clocks used in factories and schools. The master clock would be in the office and provide pulses to many remote slaves, allowing synchronized timekeeping. In fact, IBM has it's roots in this field.
 
relay_lamp_bar.jpg


Unless I've messed up in my logic, that should make the the lamps light one at a time until the reset ('bar' style). I didn't put the inputs on schema, but you should be able to tell what they are.
 
tom61 said:
Unless I've messed up in my logic, that should make the the lamps light one at a time until the reset ('bar' style). I didn't put the inputs on schema, but you should be able to tell what they are.
Click to expand...
This type of arrangement is the first thing that came to my mind. The problem is you end up with a runaway cascade effect causing all of the relays to trigger at once while the clock signal is high. I need to determine a way to limit the clock to triggering only one stage per high cycle. The clock pulse duration will likely not be easily set to a specific time, and the mechanical properties of the components probably won't allow for tight tolerances. This means the logic will have to accomodate varying pulse timings.

Thanks for the suggestion though. I'm open to any ideas. It's surprising how what appears to be a simple operation that is so easily accomplished with digital ic's is actually quite a puzzle to solve.
 
How in depth do you want this to be? I have logic worked out for D flip-flops and edge detectors for the clocking. It looks like 11 relays per stage though :(.
 
Surf out/research ladder logic. It is the building blocks of PLC circuits. It is rather easy to learn and if you go forward with your "old school" concept will be the way to go.
 
fat-tony said:
How in depth do you want this to be? I have logic worked out for D flip-flops and edge detectors for the clocking. It looks like 11 relays per stage though :(.
Click to expand...
Thatsa losta relays! :eek: Just a tad more than I had in mind. I would be interested to see the circuit however.

PS-RagE said:
Surf out/research ladder logic. It is the building blocks of PLC circuits. It is rather easy to learn and if you go forward with your "old school" concept will be the way to go.
Click to expand...
Funny you should mention that, my research just turned that direction today. I've been checking out some old telephone switching shematics, and many of them are presented in ladder logic format. I've also been looking through this relay logic pdf from a pretty expansive open book page named Socratic Electronics. Lots of good stuff covering all angles of electronics. Including this chapter on ladder logic.
 
You could cheat by using solid state logic to do the thinking, and have the lamps switched by relay. That way it would still click as the lamps turn on or off.
 
i chewed on this for awhile and this is what i came up with:
relay.PNG

Both A and B do the same thing, I think A is easier to understand, and B uses fewer parts.
Each state has two latches, one that clocks on the leading edge and one that clocks on the falling edge.
I've not tested this out at all, but still I hope it helps. Sounds like a very interesting project.
 
normal_relay2.PNG

Thought some more about it... if you set up like this it should light one light at a time.

walkthrough as i see it:
beginning state: clock is low, the bottom line coming in from the left is high (i.e. the previous node's light is on). all latches of the two nodes shown are off.

1. clock goes low. the first latch of the left node turns on.
2. clock goes high. the left node light turns on. the second latch of the left node turns on. the top line going to the left loses power.
3. clock goes low. the first latch of the right node turns on.
4. clock goes high. the right node's light turns on. the second latch of the right node turns on, causing both latches and the light of the left node to turn off.

so, if you make a circular loop of 10 of these and tap in a carry line and something to make the first light turn on when you first start up, that might do it. You can use 4pst's to cut down on the number of clock relays.

oh, if your clock is high for the same ammount of time that it's low, you can probably get by with less relays by double clocking -- so the lights toggle at every transition between high and low.

I'll bet there's some way to do this with relays + caps that really cuts down on how many relays you need.

I wish I had a bunch of relays lying around here so I could test these... it should make a lovely clatter if it works. (or if it doesn't, too.)
 
JesterBLUE said:
i chewed on this for awhile and this is what i came up with
Click to expand...
Those look like very good candidates. I'll have to run them through my simulators to watch them in action (using Circuitmaker 2000 and Proteus Lite). The operation of these circuits can sometimes be difficult to envision and don't always run like I would have thought.

JesterBLUE said:
I'll bet there's some way to do this with relays + caps that really cuts down on how many relays you need.
Click to expand...
Thinking the same. I've got an old Gottlieb EM (electromechanical) pinball machine that has no solid state parts, just tons of relays and solenoids. It's a wonder to watch (and hear!) in action and is the spark behind the idea of this project. There are several functions that make use of delayed latching through a simple RC design.

As I mentioned, I am also looking into some Telephone logic circuits, specifically counting chains. This circuit is pretty clever.

Thanks again for your suggestions. I'm hoping to find a good deal on bulk relays on eBay. I also scavange old parted-out coinop machines.
 
You must log in or register to reply here.
Back
Top