1
\$\begingroup\$

So I am contemplating building a relay computer. And I have not seen any information on why one would select one voltage over another.

It seems that it would be ideal to use 5V relays as that is slightly easier to interface to microcontrollers (which I would personally want to use for debugging during design/building). As well as being slightly cheaper.

Harry Porter's relay computer runs on 24V and the TIM8 uses mostly 12V relays (some where 6V unbeknownst to the designer at the time). Also "Fist Full of Relays" is 12V.

The main problem I can see as a possiblity would be fanout... I could also see how in that case it might be advantageous to use multiple voltages. For instance a single 5V relay could drive a bunch of 24V relays at 24v.

Another possible problem is sourcing enough 5V power since 5v rails tend to be lower amperage.

\$\endgroup\$
5
  • \$\begingroup\$ Making economy to wired used \$\endgroup\$
    – GR Tech
    Commented Feb 6, 2015 at 4:01
  • \$\begingroup\$ Not sure of historical reason but the advantage of the higher voltage would be lower current. I'm not sure I follow on the fanout, the contacts of most relays could drive the coils of many other relays. \$\endgroup\$
    – PeterJ
    Commented Feb 6, 2015 at 4:03
  • \$\begingroup\$ Hmm... not sure current is a problem though. For instance songle.com/en/pdf/20084141618381000.pdf as you can see on the Songle SRE relays which I imagine are pretty typical humdrum relays. The nominal current does go up but the power dissipation remains about the same 24*5 = 360 mW and 5*71=355mW. So assuming I used 200 relays I'd dissipate roughly 14 watts in the relays which doesn't sound too bad. \$\endgroup\$
    – cb88
    Commented Feb 6, 2015 at 4:12
  • \$\begingroup\$ My guess would be historically relays had greater tolerance ranges so any relay that activated between say 12-24 volts would work fine. The lower the voltage you go, the tighter the tolerances need to be. Relays are manufactured to very tight tolerances these days so I doubt that would be an issue for you with 5V relays. \$\endgroup\$
    – horta
    Commented Feb 6, 2015 at 4:17
  • \$\begingroup\$ @horta that seems reasonable Zusie for instance was built with vintage phone circuit switching relays from what I remember... although Harry Porter sourced his directly brand new. Perhaps people just like how 24V relays sound :D \$\endgroup\$
    – cb88
    Commented Feb 6, 2015 at 4:19

3 Answers 3

7
\$\begingroup\$

Historically, the largest use of relays has been associated with the telephone industry. Before electronic telephone exchanges were introduced in the 1960's, virtually all switching was done using a combination of conventional relays and specialized configurations like crossbar and step-by-step switches. Relays provided a means to connect two parties directly across a telephone switch with a direct circuit.

In the telephone industry, these relays used 48VDC, which was the standard voltage throughout the exchange. Initially these were open-frame construction, like this one:

enter image description here

which had the advantage the contacts could be cleaned and adjusted as needed, but they were also subject to dust. So for industrial use, relays were equipped with dust covers, like this one:

enter image description here

and became available in 6, 12, 24, 48 (AC or DC), and 120/240 AC coil voltages. Note there are no conventional relays currently available at Digi-Key with either 3.3V or 5V coil voltages. I am guessing in particular 3.3V is not sufficient to be able to develop enough of a magnetic field to operate this large a relay.

The relay shown above has a 4PDT configuration (or four "Form C" contacts). This contact arrangement is the same type used in Harry Porter's relay computer (all 415 relays in his computer are the same type). There is also a 6PDT version available.

Another type of relay is the reed relay, first used in the 1930's. Once again, it initially was used primarily by the telephone industry. For example, General Telephone's first electronic switch, #1 EAX introduced in the early 1970's used reed relays to connect the subscribers, controlled by a computer and other logic built using some of the first TTL integrated circuits.

enter image description here

There are also small general purpose relays that are sealed, but are not reed relays. They may be either rectangular, or more of a cube in some cases like this one:

enter image description here

All relays available in the 3.3V and 5V coil voltages at Digi-Key appear to be either reed relays or sealed general purpose relays. However the most complex contact configuration you can get is DPDT. So it would not be possible to directly replicate Harry Porter's machine using these smaller relays (in many cases, you would have to use two relays in parallel).

Besides coil voltages of 3.3V and 5V, these are also available in several other voltages below 5V and also in 12V and 24V versions.

There are three advantages reed relays and the small sealed GP relays have over conventional ones: they are smaller, they cost much less, and they are faster -- for example a reed relay can often switch in a few hundred microseconds, and a conventional relay may be ten times slower ( a few milliseconds).

You could build a relay computer using reed relays or the small sealed GP relays, and it would certainly be smaller than Harry Porter's, but you wouldn't have the satisfactions of being able to watch the relays operate, and more importantly, hear them operate like you can in his video. (If you do decided to use conventional relays with a dust cover like the photo above, you may want to check out the surplus market, since those relays cost about $12 apiece new.)

I'm a little confused about the issue regarding fanout -- all of these relays, both conventional or reed, have contacts that can typically handle either 1A or 2A, so driving twenty other relays would not be an issue. I'm sure that's plenty.

\$\endgroup\$
3
  • \$\begingroup\$ I like your answer except for one thing.... I'm pretty sure the smaller relays on Digikey are not all reed relays... for instance the Songle 5v relays are not reed relays they also make a 3v version youtube.com/watch?v=kJEO8FHCXag. Also after a short search digikey has 638 relays that are 5VDC or 5V DC/AC that are not reed relays. Just because you can't see into the package doesn't mean you should assume it is a reed relay. \$\endgroup\$
    – cb88
    Commented Feb 6, 2015 at 17:58
  • \$\begingroup\$ You're right, I see now that a lot of 3.3v and 5v relays that looked like reed relays (because they are sealed) are indeed listed as general purpose. I wasn't aware if the Songle relays since Digi-Key doesn't carry them. I'll update my answer. \$\endgroup\$
    – tcrosley
    Commented Feb 6, 2015 at 18:13
  • \$\begingroup\$ Yeah songle seem to be el-cheapo. But they ought to be fine for my use unless they die too fast... hopefully they hold up well enough. I think I'm going to go with 5V relays .... worst case I should at least be able to build an ALU out of them. \$\endgroup\$
    – cb88
    Commented Feb 6, 2015 at 18:15
0
\$\begingroup\$

Perhaps 24 volt relays are used due to Tradition. Even before transistors, 24 volts was used in industrial control systems, so 24 volt relays were readily available.

I suspect 5 volt relays were not readily available until sometime after 5 volt digital logic became widely used. But even then, you can easily drive 24 volt relays from 5 volt logic.

\$\endgroup\$
1
  • \$\begingroup\$ Yes a transistor can drive a 24 relay just fine. Acutally, I drive some PLC IO from an optoisolator at 24V at work. I suppose it would just annoy me a bit if they operated at different voltages. \$\endgroup\$
    – cb88
    Commented Feb 6, 2015 at 4:22
0
\$\begingroup\$

Some reasons driving the design for a relatively high drive voltage in relays:

-To increase the magnetic flux slew rate (dynamic behaviour)

-To compensate the ambiend temperature increase

-To minimize the accidental operation or release state of the relay (wider voltage gap, risk of drive voltage fluctuation or excessive ripple, no return spring...).

-To increase the sensitivity of relay (increasing the resistance of relay coil)

-Decrease the self-heating temperature by lowering the current (necessary when many relays operated continuously in a limited space)

To my perspective, it is half of some variables trade-off and half of modern materials used.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.