Photos of Phones and Phonographs (occasionally)

PAX Mega-Update, and a video.

To celebrate that I’ve got enough tie lines working to phone from one side of the living room to the other, via the asterisk server at the Dean Forest Railway.  I made a video of me dialling 30-0-50-91-491.

The red phone (GPO 706) is on my AT&E 50/7A PAX.

Dialing 30 takes you out through an outgoing tie line and onto the junction selector, 0 takes you back into the 50/7A on an incoming tie line, 50 takes you out through another tieline and into my asterisk server. 91 are the access digits which take you in to the asterisk server at the Dean Forest Railway, 491 then takes you back into my asterisk server, which then dials “39” on your behalf which takes you out *another* tie line into my 10/2A PAX and then it finally dials a 5 to ring the black phone (also a 706)

Two 1950s strowger PAXes, a home-brew junction selector, two asterisk servers and a load of internet later – and I’ve phoned from one side of my front room to the other.

This is what happens when you throw your TV out and find something else to do with your evenings.

PAX Mega-Update!!!
As promised, here’s a list of everything I’ve done to my phone exchanges since I last wrote anything about them about 8 months ago.

My asterisk server blew up its HD controller, so I’ve completely rebuilt it, this time using a Neoware CA10 thin client (my power consumption has dropped from 60W to about 14W – win!) and a cheap and cheerful Chinaroby TDM400P which unlike most TDM400 clones can power the FXS ports without an external molex connector – which is handy because the CA10 doesn’t have any molex connectors inside!


In the process of doing this, I updated my build notes for the easiest way to go from bare metal -> debian -> asterisk/dahdi installed and working from packages in the repos (so you automatically get security updates!), with some basic OS hardening in the mix as well.  It doesn’t take much doing, is so much easier than building from source, and is more secure in the long term.  At some point I’ll finish the rant I’ve got brewing about why doing it any other way is madness!

I’ve also done a lot of work on the dialplan, built a prototype of a “Dial-a-Disc” macro although I’m not happy with it as it’s only got the long form of the announcement.  I’m working on a better replica.

I’ve done loads on the big exchange!

  • Back in March I had problems with FS4 not clearing down properly so I busied it out.  Until I found that entry in the exchane diary I’d forgotten about that fault.  I must fix it!
  • I ripped out the screw terminal DP I had in the bottom of the pax, and replaced it with a krone box.  OK, I know it’s not “heritage” but it is much more practical.


  • Terminated all 50 line circuits on the krone strips
  • Fitted a small panel of 6xRJ45 sockets, these are also terminated on the krone blocks, and are jumpered so that each RJ45 carries 4 extensions. 21-24, 25-28, 41-44, 45-48, 61-64, 65-68.  This way I can run cat-5 around the house and use adaptors to break it out to phone sockets.  The whole PAX is then self contained and not wired into my house.  Which will make moving it easier in the future.
  • The Asterisk junctions also terminated on the krone block, and are jumpered in to 51 (from the asterisk FXS) and 50 (to the asterisk FXO)
  • I wired my homebrew junction selector back in.  Again this is terminated on the krone strips, and is jumpered to 30 (to the junction selector) with outlet 0 jumpered to 31.
  • Exchange phone is now jumpered to 68 rather than 35 as it used to be.  This is brought out on a BT socket in the bottom of the PAX.
  • I’ve been right through all the odd strapping arrangements I found on my PAX back in 2011, I’ve worked out what all of them do apart from two lines – and have reversed all the strapping back to stock.
  • The two lines which are strapped strangely seem to go off to the FA tags, which are associated with the external fire alarm equipment (which I don’t have) and the FA relays in the selectors (which I also don’t have) – I’ve sunk a lot of time into failing to work out what’s going on there.  I’ll sink a lot more time before I give up and revert the strapping!
  • Cleaned and adjusted the RV and BV contacts, so the tones are a bit nicer now.

I’ve also been giving the small exchange some TLC

  • Sorted out the earthing, so if the PSU goes faulty it’s not going to kill the cat.
  • Modified Extension 0 for tie line working.  This involved an awful lot of chatter on the THG mailing list about how I should go about it (all I wanted to know was the dimensions of the standoffs for mounting the rectifiers!)  For info, they’re 6BA threaded stud to 2BA tapped hole.  The standoff is about 1″ total length, and each threaded portion is 0.25″ long.  They look like this (bare standoffs on the left, the ones fitted to my 50/7A – including the rectifiers – on the right):


  • I’ve hooked extension 0 up to the 50/7A using the tie line on 39.
  • I wired out all the extensions down to a set of RJ45 sockets in a similar manner to the 50/7A
  • I’ve fitted the line relays for the extra position 11 on the LF mult, but haven’t wired them as it’s not a comfortable PAX to work on.  I need to see if I can get the side off it or something to give me space to see where I’m prodding a soldering iron.

My really dinky 5 extension electronic PABX.  I’ve had a good rummage in this and have determined that extension 5 is completely dead, 2,3,4 are OK. Extn 1 gets tones, can dial and will ring if dialled – but there is no speech path.  This is either a really easy fix, or a really hard fix.  It all depends which component has gone bang and if I can get a replacement.  Once my workbench is clear of projects I’ll look at it properly.

PAX Update – FS3 woes


Yesterdays update was about BOV, so it’s only fitting that todays update is about the other large project at home, my phone exchange.

Since my last update back in September, I’ve made a fair bit of progress and uncovered one incredibly perplexing fault.

Progress first…

  • I built a frame to hold a single U-Point (very kindly provided by Martin Loach, thanks Martin!) so I could mount my outgoing junction selector.  This has been cabled up and is accessible from outlet 30.
  • The electronic PAX I had wired in to the outgoing junction selector went “pop” so now the only thing hanging off it is a route back in to the 50/7A
  • I added a sipgate trunk to my Asterisk box, so I now have a landline number which causes my exchange to rattle into life.
  • The outgoing junction selector wasn’t homing smoothly, so I readjusted the interrupters and it’s much happier now.

Somewhere in there (but not documented in my local exchange diary, because it was work done at the railway) I also got my PAX talking to the UAX at the railway again.

Perplexing Fault:

FS3 has a fault.  The outgoing tie line to my asterisk (50) works fine, but if I try and go out the junction selector (30) FS3 just skips over the bank position, lands on outlet 11 and returns busy tone to the caller.  None of the other 3 final selectors I’ve got have the same problem, and if I swap selectors around the fault follows FS3 (so I know it’s something in the selector rather than the shelf position, the multiple, the line circuit for 30, the outgoing junciton set or anything like that)

Closer inspection when it’s misbehaving reveals that the H relay doesn’t pull in when the selector lands on 30 (but it does when it lands on 50).  Manually operating the H relay causes normal operation, and I can dial through the tie line without any problems – so something in the path for the 900R winding of the H relay isn’t happy.

My first thought was that perhaps the wipers were out of alignment and that somehow I was getting an earth on the PN wiper (which would happen if 30 was busy, and skipping on to outlet 11 would be desirable behaviour in that case).  Not an unreasonable theory given that I had to set up the wipers on FS3 back in August 2012 – and when setting up wipers I generally use level 5 as my reference level (and we’ve established that the tie line on level 5 works)

However, if I busy out the selector, including the vertical/rotary magnets, and manually position the selector on outlet 30 – all the wipers have the conditions I would expect on them.  Indeed comparing them with another selector in the same state revealed no obvious differences.

As usual when fault finding, I gave all the relay contacts etc a good clean, and checked they were making/breaking contact (that didn’t make any difference either)

Consulting the diagram notes has shown that it’s something in the path from (NEG, 1300K, PN bank and wiper, 900H, C7, 500G, TL5, E2, NR4, N2, B1, POS) but it’s getting a bit late at night to be poking about in that sort of detail – so it’ll have to wait for another day.

Connecting my PAX collection


My collection of phone exchanges hasn’t really grown at all in over a year now, but they’re still all very much isolated islands.  You can’t phone from one exchange into any of the others (with the exception of the 50 line exchange and the asterisk server, which are connected by a tie line which you access by dialling 50) – I’d like to link them all up, in a manner similar to that shown above.

Extensions on the asterisk server are 2 digits (because it makes the dialplan consistent) To get from PAX extension 26 to VOIP extension 81, you dial 5081.  The 50 connects you to asterisk, and 81 connects you to the VOIP phone.  The asterisk server has been set up so that you also dial 50 to access the PAX, so from 81 you can dial 5026 to call in the other direction.

This gives a nice consistent 4 digit dialing scheme!  Great!

My 10 line PAXs throw a bit of a spanner into that though as they have a single digit numbering scheme.  I can have extra outgoing tie lines from the 50 line PAX (for example on 30) but if I were to connect them directly, they’d be 3 digits and that would *never* do!


So at the weekend, I designed and built this relay set.  It hangs off the 30 tie line, and stores the next digit you dial.  Depending on what digit you dial, it can connect you to one of up to ten 10 line exchanges (I don’t have 10, but it’s best to plan ahead!)  So now, Dialing 30 connects you to this relay set, dialing 0 connects you to a 10 line PAX, and then you can ring the phone at extension 4 – giving you 3004!   Yay!  4 digit phone numbers!

OK, so going the other way isn’t quite as neat (Dial 0 to get out of the 10 line PAX, then you’re landed straight into the 50 line, so dialling 026 will ring extension 26, or 05081 will dial VOIP 81) but it’s better than nothing… and my ATE 10 line PAX doesn’t have any outgoing tie lines at all yet so I’m not that worried!


Anyway, I’m quite pleased with this project as it’s the first non-trivial relay set I’ve designed, built, and debugged without assistance!  I say designed, most of the circuit elements are pretty standard.  The junction half of the circuit is based on the auto-auto junctions we use at the railway.  It’s a pretty basic stone transmission bridge, with 2 stage dropback and the A2 contact repeating dial pulses out to the line.

The uniselector which reacts to the dialed digit is controlled by a circuit based loosely on the stepping circuit from a PABX4 group selector but with some modifications to work with a uniselector rather than the 2000 type selector used in the PABX4.

For example, uniselectors don’t have “off normal” contacts (mechanically operated contacts which close when the selector steps off its home position) so instead I’m using the homing arc on the uniselector I’m using for the same purpose.  It’s also used to drive the uniselector home at the end of the call.

Oh, and thanks to the asterisk server at the railway having extension 491 configured to dial back through my exchange to ring extension 26, and me hooking outlet 9 on my new junction selector back in to the 50 line pax…

I can now call from extension 25 to 26 by dialling: 50309500491 – which goes through my 50 line PAX to the asterisk (50), back into my pax and back out again to my new junction set (30), back in to the PAX again (9) and then back in to the asterisk box (50) off to the railway (0) and then back through my asterisk server and back into the PAX to ring extension 26 (491) – *phew*

Gloriously pointless, but it yields surprisingly good call quality and not a single pulse was lost!


Priority Calling Modifications

My description of priority calling in yesterdays post wasn’t quite accurate.  I’ve dug out the Operational Bulletin and checked.  It says:

3.4 Priority Service
An official having this service may speak instantly to any extension on the exchange, whether the extension is already engaged on a call or not.

The caller dials the required number and the Link Selector wipers extend the call to the required line.

If this line is free the bell rings and the called extension answers.  If the called line is engaged no Busy Tone is heard and the caller is instantly switched into the existing conversation.

The call at this stage is not private; when privacy is required both engaged parties must replace their handsets, ringing current is then automatically connected to the required party’s bell and the call proceeds in the normal manner.

The circuit diagram notes say this:

7. Priority Facility
The priority facility enables an extension to speak to an engaged extension when the line that is dialed is already engaged.

An extension line with priority has tags K and PC strapped on the tagblock.  On completion of dialing the two digits, relay G operates if the line is engaged.

Relay G operated:
G5 completes a circuit to operate relay E (NEG. 800E, C2, G5, (TL6) (NPB) S1, T1 & 2, U3, FS bank tags PC to K, K2, L1, FS (P) bank, U27, B6, T9 and 10, POS)

Relay E operated:
E1 completes the circuit to operate relay D (NEG. 200D, 500YB, G3, NR1, E1, B1, POS)
E4 disconnects Busy Tone from the calling line.

Relay D operated:
D2 & D3 complete a speech circuit across the called extension lines and the caller can speak and hear.

If required the priority extension may request the extensions to clear the call in order that he may speak in secret to the required party.  In this case on the line circuit of the required line being restored, NEG is replaced on the PN wire from 1300K, in the line circuit and relay H is now operated from (NEG, 1300K, PN bank, 900H, C7, E2, NR4, N2, B1, POS)

Relay H operated:
H6 releases relay G followed by relays D and E.
With relay E normal, ringing current is connected to the called line at E3, and the call proceeds in the normal manner.

So that’s how it’s supposed to operate.  How does the modification I’ve got change this operation?


 This diagram should ideally be read in conjunction with the original diagram – as the diagram above doesn’t show the whole circuit, just the PC modifications (with a few relevant bits of circuitry just so you can orientate yourself on the main diagram!)  As far as I can tell, the new sequence of operations is:

Relay PC operates:
PC is operated as follows: (NEG. 2000E, (TL6) (NPB) S1, T1 & 2, U3, FS bank tags PC to K, K2, L1, FS (P) bank, U27, B6, T9 and 10, POS)

PC2 completes a circuit to operate relay E (NEG. 800E, C2, G5, PC2, (TL6) (NPB) S1, T1 & 2, U3, FS bank tags PC to K, K2, L1, FS (P) bank, U27, B6, T9 and 10, POS)

Relay G operated:
G5 completes a circuit to operate relay E

Relay E operated:
E1 completes the circuit to operate relay D (NEG. 200D, 500YB, G3, NR1, E1, B1, POS)
E4 disconnects Busy Tone from the calling line.

Relay D operated:
D2 & D3 complete a speech circuit across the called extension lines and the caller can speak and hear.

If required the priority extension may request the extensions to clear the call in order that he may speak in secret to the required party.  In this case on the line circuit of the required line being restored, NEG is replaced on the PN wire from 1300K, in the line circuit and relay H is now operated from (NEG, 1300K, PN bank, 900H, C7, E2, NR4, N2, B1, POS)

Relay H operated:
H6 releases relay G followed by relays D and E.
With relay E normal, Int Ring Tone is connected via PC1 to the called line at E3, and the call proceeds in the normal manner.

I have no idea why you would do this.

Int Ring Tone isn’t designed to ring a bell (it’s a tone!) and as far as I can tell, the only conditions under which PC1 is operated are when the called party is off hook – so I’m at a complete loss as to what this mod was attempting to achieve. Having documented it thoroughly, and spent a number of hours headscratching trying to work out what it’s for… I think I’m going to declare it useless and reverse the modification.

PAX mega update!


It’s been over 6 months since I last updated on this project, which is a shame because I’ve done rather a lot since then!

Apologies if you’re one of my non-telephone readers, this probably isn’t going to be a very interesting post, and it’s rather long.  I’ll make up for it by trying to go for a photowalk tomorrow or something.

The biggest bits of progress in Feb were that I started fitting the TL (Tie Line) relays to the selectors, and built an asterisk VOIP server.  The two projects are linked in that the TL relays are needed to allow me to call through my PAX and into my asterisk server.  I did actually blog about the TL relays at the time, but I don’t appear to have mentioned asterisk here before.

While I did most of the modifications to my first selector in Feb, it didn’t work.

March was a productive month.  If you look at the selector which is 3rd from the left in the picture up top, you’ll see it’s got an extra relay.  This appears to be a non-standard local modification from when my PAX was in use at the Ffestiniog Railway.  I spent some time in March mapping out what it does.

It appears to be related to the Priority Calling feature these exchanges have.  If I’ve read it correctly, it works something like this:

Alice is in conversation with Bob, but Mr Miggins (Bobs manager) wants to talk to him urgently.  Mr Miggins has Priority Calling set up on his extension (because he’s a bigwig!) and dials Bobs number.  Bob and Alice get interrupted by a bust of ring tone to signal that they should end their call immediately and hang up.  When they hang up, Bobs phone rings and Mr Miggins can tell him off for flirting with Alice on company time.

In its original form, the circuitry connects Bob and Alice to ringing current (designed to ring a bell), but with this modification it connects them to ring tone instead (the sound you hear when a phone is ringing)  I’m not entirely clear why this is useful, or why it needs a relay to achieve this!

Also this month, I bought a BT Voyager 220V adsl router from a car boot sale for £1.  I reconfigured it to work as an ATA and give me a phone on the DFR telephone network.  I did start writing up how to do that, but it turns out not to be all that easy to describe.  I’ll come back to it at some point as ATAs usually cost around £30.

I cleared a load of faults on the selector I’d modified (FS2), learning a lot in the process

FS2 developed a release fault, which means that at the end of the call, the selector didn’t restore to normal but instead got stuck.  There are some clever widgets called “heat coils” which stop things from catching fire when this happens.  Here’s a picture of them on the PAX:


and here’s a closeup of one of them:


They have a pin (on the right of the picture above) which is held in the body of the heatcoil with a drop of solder.  The idea is that if the release magnet gets held on the current flowing through the heatcoil causes it to heat up, melting the solder and pulling the pin out.  The mounting then springs apart, earthing a contact which in turn lights the alarm lamp so you’re alerted to the fault.

Heatcoils appear to be next to impossible to get hold of, so it’s a good job they can be repared by carefully pushing the pin back in with a hot soldering iron.

I didn’t get anything done to the PAX in May, probably because I was rushing around like a mad thing trying to do up BOV and then I switched everything off and went on holiday for 2 weeks.

Actually, I tell a lie.  Looking at the svn commit messages, May was the month I got my asterisk server talking down a SIP trunk to the server at the railway.  I can now call from my office at home, through my PAX, into my asterisk box, through the internet to the asterisk box at the railway, out into the UAX13, then out through a junction and a few miles up the road to a PABX4 and listen to the test number.

Which is nice.

I came back from holiday, built a charging unit for the PAX batteries – then switched my asterisk server back on, only to discover that it wouldn’t boot.

I decided to leave phones alone for a bit and work on BOV instead.

I traced the asterisk fault to a dead compactflash card, so I resorted to old fasioned spinning hard disk instead (so much for low power consumption!)

This was a pleasing experience, because 95% of the config for the server was in svn so it was dead easy to reinstall everything and get back up and running.  I did identify a few things that weren’t in svn and documented them in the form of a shell script – and checked that in to svn.  Sure it’s not quite puppet like I use at work… but it’s better than nothing.

I then turned my attention back to FS2 which was still causing problems.

My paper exchange diary shows lots of slow, careful, methodical troubleshooting (cleaning contacts, adjusting spring sets, that sort of thing) – eventually resulting in me realising that white heatcoils are no good, only green will do – and declaring FS2 successfully modified – a mere 5 months after I started. “must do better” I thought.

I started adding the TL relay to FS3.  Managed to get all the wiring done in one evening!  It took me almost 6 hours, and by that time it was too late to be testing it incase the rattling/swearing woke up the neighbours.

Needless to say, it didn’t work first time – and had a really odd fault.  Dialing 50 caused the line circuits for 63 to pull in and you end up back at dialtone – which allowed you to dial another number!  That’s not right at all!  I checked all the new wiring and it was fine.

*scratches head*

Purely by chance, during one test I noticed that one set of wipers was horribly misaligned and was shorting out all sorts of stuff it shouldn’t be!  I realigned the wipers and it worked!  Yay!  I now had 3 working TL selectors!

High on my success, I tidied up the 50 pair tie cable on the back of the PAX, fitted a battery isolation switch, tidied up all the mains cabling for the asterisk/charging circuit etc then set about adding a tie line relay to FS4.

That went much better.  Having learnt from all my previous mistakes, it only took about 4 hours to wire it all out and it worked perfectly first time.

2012/08/12 – a year and a day after bringing the PAX home from Ffestiniog it had 4 working TL selectors.  I did do a bit of a happy dance in the front room, and the cat gave me a dissaproving look.

But wait! There’s more!

I added a speaking clock to my asterisk server, with the voice of Pat Simmons.  Pat was the second voice of the GPO speaking clock in the UK from 1963-1985 (the first being Ethel Cain, from 1936-1963) – I’ve started to build something similar with Ethel’s voice, but suitable recordings don’t appear to be easy to come by and I’m missing a lot of numbers – and crucially the “o’clock” sample.

I’ve also ordered a “Wave Shield” for my arduino with a plan to prototype a hardware speaking clock.  We’ll see if I get any further with that than I have previous arduino projects.  They tend to fall apart when it gets to writing the software – I lose interest because it becomes too much like my day job ;)

Todays *major* excitement was that I went to a car boot sale, and bought these for £3


They may look like a pair of long nose pliers to most people – but they’re BT /GPO “81’s” (or “Pliers, Wiring, No2” if I remember correctly, they’re usually just called 81’s) and they’re the *right* pliers for working on telephone equipment.

I’ve been looking for a pair for about 2.5 years and I’m absolutely stoked that I’ve now got some!



I’ve not been posting much this week as I’ve spent most of my evenings going cross eyed trying to modify a chunk of my phone exchange so that it can talk to other phone exchanges.  The easiest way to make sure I’ve done all the modifications seems to be to highlight the changes required on a copy of the diagram, then cross them out as I add/remove that chunk of wiring.


I’m quite glad that I’ve already got a working modified piece to use as a reference.  You can see the nest of wiring I’m having to work my way through!  The already modified one is at the top of the picture, the one that is a work in progress is at the bottom.  I think I’ve done most of it now, I just need to reroute and reconnect some of the white wires and I’m done.


Then if it works, I’ve only got to do it twice more on the other selectors!  Fingers crossed…



Not an exciting photo this evening, but it does represent a good chunk of what I’ve been up to tonight.  I’m trying to get my head around how to modify the selectors on my PAX to provide tie line working.  It seems I need to track down some rather specific spring sets, I’ve got the appropriate relay coils, bases and armatures but I don’t have the right spring sets (or even anything close!)

So if you’ve got a stock of spare GPO 3000 type relays knocking around, I’d be grateful if you could have a look and see if you’ve got anything with the following contacts:

Contact Number Springs Type
TL1 1,2,3 Changeover Contact
TL2 4,5,6 Changeover Contact
TL3 7,8,9 Changeover Contact
TL4 21,22 Make Contact
TL5 23,24,25 Changeover Contact
TL6 26,27 Break
TL7 28,29 Break

So in total, I need 4*changeover, 1*make and 2*break contacts. The order of the contacts on the springset probably isn’t critical as I can re-map contacts as needed. In an ideal world, I’d be looking for enough contacts to make up 3 relays like this – but I’d settle for anything I can get!

Update: John Bathgate at the Dean Forrest Railway has found me the appropriate spring sets and buffer blocks.  Thanks John!

PAX Update


My last progress report on the PAX restoration was over a month ago.  I’ve made some significant progress since then – I just haven’t written about it.  Well, I say that.  I’m keeping a paper diary, I just haven’t managed to update here (mumble mumble busy at work etc etc)


So In the last month I have:

  • Repaired a couple fo damaged heat coils.  I’m pretty sure these aren’t a repairable part, and while my repair works electrically, I’m not sure it would stand up to the lighting strike they’re supposed to protect it from.  Mind you, I’m not planning on having any external plant hanging off this PAX so that’s probably not an issue.
  • I’ve fitted a wooden platform to the base of the PAX, to act as a shelf to hold the batteries, charging equipment and a more accessible connection box for connecting external equipment
  • Installed the batteries, with a removable connector to make it easy to isolate the battery supply if I need to
  • Fully tested all 4 selectors, I’ve identified two of them as having faults on the called subs side of the circuit.  One doesn’t pass ring current on to the called sub, and the other has a ring trip fault.  Neither passes speech.  This is hopefully just more relays to adjust.
  • Desoldered, cleaned up and re-wired a 60 pair “Box Connection” kindly donated by John Bathgate from the DFR
  • Desoldered the old 50 pair cable from the tagstrip at the back of the PAX
  • Mapped out the strapping on the tagstrip.  Which took ages, as there are 500 connections just for the line circuits (and another 50 or so for as-yet-unidentified-stuff)
  • Fanned out, laced, and wired in most of a new 50 pair cable from the box connection to the tagstrip on the pax (I left out the lines with unusual strapping until I understand how they’ve been strapped) – that’s what the photo at the top of the post is of.

I’ve still got a lot of work to go, but I’ve got 30 lines available and ready for phones so far.  I’ve identified 10 line circuits (spread across levels 3 and 5) which are all strapped for various forms of tie line functionality.  This is a bit odd, as from looking at the line circuits only 5 tie lines are fitted.

I really need to get my head around how the tie lines work, but my current working assumption is that it’s currently wired for 5 incoming and 5 outgoing junctions.  There are another 10 lines which have unusual strapping, that I haven’t yet identified (they don’t seem to match anything on the diagram)


That is one page of my strapping map.  There’s another page of it as well.  I’m hoping I can use that, combined with the wiring diagram and a series of train journeys to work out what on earth those last 10 line circuits do!

Oh yes.  And I’m building a PC that I can use to hook my PAX up to the internet.  More news on that when I’ve finished collecting the parts together.

Automatic Telephone and Electric Company (ATM/ATE) – Adverts

Following on from my post the other week about determining what year my PAX was made, I’ve been digging through old telephony books looking for adverts relating to ATM/ATE.  While I haven’t (yet) found any relating to my PAX, I have found some adverts from the same company.  So here they are:

From the October 1935 edition of the POEE Journal

From A Handbook of Telecommunication by Bertram S. Cohen published 1946

From Telephony by Atkinson, 1948 edition

From  Telephony by Atkinson, 1950 edition

I’ve enjoyed working my way through the 30 or so adverts I’ve scanned so far (you can browse the entire collection here) but now that I’ve started looking, I’m going to have to carry on looking!  There must be so much of this material hidden away in books, magazines and journals – overlooked by most people who are more interested in the published articles and the content of the books than the advertising which offset the publishing costs.

So, when was my PAX built?

ATE Logo, 1936-1958(ish)ATE Logo, 1958(ish) onwards

Since I first got it, I’ve been trying to work out how to tell when my PAX was built. Initially, I thought it was much earlier than it is.  My first post here describes it as “1930s” which turns out to be wrong.  The initial design of it may be late 1930s, but my example is much later.  As far as I can tell, there were at least 3 generations manufactured:

  • A cabinet with a “black crystalline finish” – this is the version described in an article from 1938.  I’m pretty sure there was a 25 line example of this at Ffestiniog when I collected mine.
  • A cabinet with a “grey hammered finish” – wired throughout with silk covered wiring (this is the one I’ve got)
  • A cabinet with a “grey hammered finish” – wired throughout with PVC covered wiring.  This would be later than mine, and there was certainly an example of this at Ffestiniog.

The company which made it, has had several names, and at least three logos.

  • A circle containing the letters ATM, with the company name “Automatic Telephone Manufacturing Company”
  • A circle containing the letters ATM, with the company name “Automatic Telephone & Electric Company” (as found on my 50 line PAX, shown above top)
  • A circle containing the letters ATE, with the company name “Automatic Telephone & Electric Company” (as found on my 10 line PAX, shown above bottom)

Here’s the timeline my internet research dug up.  Most of the links below are from the same site, but that’s only because there isn’t a huge amount of information online about this company:

  • 1911 – Automatic Telephone Manufacturing Company established (From Uk Telephone History and History of ATM)
  • 1912 – Automatic Telephone Manufacturing Company established (From A history of Plessey, Information & Publicity Services, PTOSL, Beeston.  This conflicts with the 1911 date claimed elsewhere, but that may be due to a difference in the definition of “established”)
  • 1936 – Company changes name to “Automatic Telephone & Electric Company” or AT&E, keeps the ATM roundel in the logo despite the name change (From History of ATM)
  • 1938 – AT&E magazine published, describing the first model of my PAX.  The unit is described as having a “Black crystaline finish” – my unit is a gray hammered paint finish so is obviously later  (From Private Automatic Exchange Equipments, A new series of designs)
  • 1946 – AT&E advert, showing the same logo (ATM roundel, “Automatic Telephone & Electric Company” text)  as the front of my PAX (From
  • 1950(?) – Another advert, same logo in use (From
  • 1951 – An extract from an AT&E brochure, also describing the “Black crystaline finish”
  • 1957 – My 10 line PAX has a stamp inside indicating it’s built to “Diagram S202662/Issue 8” – my copy of “Diagram S202662/Issue 8” is dated 1957.  However, this just means it was built *after* the Issue 8 design drawing was released.
  • 1961 – AT&E merges with ETL and Plessey (From A history of Plessey and History of ATM – the latter notes that AT&E continued to use the name for a while)

So, based on the paint finish, mine can’t be any earlier than 1951, and based on the company name it can’t be later than “a while” after 1961.  Hmm, that’s not satisfyingly accurate enough.

All those dates and still no hard evidence of when my PAX was built.  It was at this point, I remembered something from a discussion about relay markings.  The 3000 type relays used in my PAXes have markings on the end which tell you useful stuff about any special characteristics the relays have.  Amongst the markings it tells you which company built the relay, and (crucially) when!

This extract from Telephony by J. Atkinson (Volume 1, Page 113) explains the markings:


I felt like a bit of a muppet.  That’s a much easier way to date my PAX and far more accurate than playing “guess when the logo was used”!  So, from looking at the equipment permanently installed in my PAX (rather than the selector cans, which can be removed), I found relays with the following dates:

  • 1956,1957 (First 40 line circuits)
  • 1960 (Last 10 line circuits.  From the looks of the mounting strip, they’re a later addition)
  • 1951, 1957 (Ringing, Tones, Misc relays)

Ignoring the 1960s additional line circuits, and assuming that the 1951 relays were a batch being used up (or spares replaced from old stock) I think it’s reasonable to conclude that my PAX was built in 1957.  So not quite “1930s” as I originally stated, but still, it was built 54 years ago!

If I take a look at the selectors (which are removable, so unreliable for dating purposes as they could be swapped around) I have dates of 1955, 1962, 1965 and 1969.  By fluke, I had arranged the selectors on the shelf in date order!  I also note that the very first selector I got working was the oldest example.

While in a new-found-dating-frenzy, I thought I’d date my 10 line PAX.  This is a little easier to date as it has had less work done to it over the years.  The relays are all dated 1958 or 1959.

With this piece of information, I can deduce that the logo with the ATM roundel was in use from the merger (1936) to some time between 1957 (My 50/7A) and 1959 (My 10/2A) and that the change from silk to PVC wiring also likely occurred at around that time (as my 10/2A is PVC wired)

wordpress theme based on zenphoto