Here you can read a collection of articles from the magazine of the Telecomms Heritage Group about PAXs that members own or recall.


AT&E 50-line

Ericsson 22-line

Ericsson 50-line

TMC Temco


The following [written in Spring 1989] is an account of how a collector of many interests came into possession of an Automatic Telephone & Electric Co. Ltd (AT&E) 50-line Private Automatic Exchange (PAX).


Approximately a year ago I had heard word that there was an old AT&E 50-line PAX in a building 'somewhere in Leicester' and that it was going to be 'thrown out'... At that time I already owned a GEC 25 line uniselector PAX and had a B.P.O. PABX No 6 in pieces lying around the house. I was in the process of converting an outhouse and outside W.C. into a room to house these items... and so when the news of this AT&E PAX arrived I immediately did a bit of mental-juggling and decided that there was room for 'one more' if the GEC PAX remained in the house. The temptation was too much!


I made some enquiries and went and had a look. The PAX was in the basement of a building in the centre of Leicester, which had been standing empty since 1971 but was now being gutted and re-furbished to be the new 'Rates Repayment Office' (formally the old 'Leicester Permanent Building Society' offices in Pocklintons Walk if you know Leicester at all).


I shone a torch into the gloom and my luck was in... despite the covers being off and building work going on all around - sitting undamaged in a dark corner was a 1951 AT&E 50-line PAX... after a quick inspection, and gasps of wonderment, I taped the covers on and left it until I could arrange a van and assistance to move it.


Three weeks later a friend-with-a-van and myself arrived to collect it as arranged with the foreman... unfortunately the builders had picked that particular day to erect some scaffolding on the stairs to the basement! We discovered another route down but it was a bit 'around the houses' to say the least - and worse still involved a climb up a metal fire escape.


I took out the 5 two-motion selectors and removed all the covers from the cabinet. There should have been a mains power supply unit within the PAX but in this case it seems that the power was fed from a thick cable coming from the ceiling... I crossed my fingers, said a small prayer, and hacksawed through it. No sparks! Then the same fate for the cables coming out of the dis-case on the wall, which was then removed and, because it was still attached to the PAX by a large umbilical cable, stowed in the bottom of the PAX. All ready to go...


The PAX was fortunately light enough for two persons to lift which is as I'd hoped. There was no lighting down there to see by - it was freezing cold and the builders were drilling and hammering - the noise was awful... We managed to get it up some concrete steps one-at-a-time but then came the difficult bits across mud and heaps of bricks etc... I slipped and it sat slowly onto my knees... (at the end of 'the move' it was my knees that ached the most) the builders said they were too busy to help us lift it up the fire escape (in fact the foreman, who had been very friendly the other times I'd visited, suddenly became snarling and abusive). Somehow we managed to get it up to the ground floor and went to get the van... meanwhile the foreman - bless him - had fetched his own van and had blocked up the entrance we were planning to use! We had to carry the PAX another eternity over MORE rubble etc...


The rest was easy compared, and that night in the pub I felt that I had earned the pints I was quaffing... meanwhile the PAX stood in the outhouse and there it remained for many weeks as Christmas passed and until I was able to find the time to finish the wiring and painting in its new home.


The first job was to generally clean the PAX - it had a good layer of dust all over the insides which was vacuumed and brushed off with a sash brush - the selectors which didn't have covers/cans were given the same treatment. I then washed the covers which are a black crackle-finish paint - a lot of dirt came off, but the finish remained dull and lifeless so I gave it a light coat of Three-In-One oil which was applied with a cloth and worked wonders! [Editor's note: Johnson's baby oil works very well too.]


In order that I can get at the back of the cabinet I have lifted it onto a board which has heavy-duty castors screwed on the bottom and I can simply wheel the PAX away from the wall if necessary.


The dis-case is a fairly large wooden case with hinged cover - two strips of terminals are mounted inside in 6 blocks of 20 pairs. I cut off all the old distribution cable-heads and removed all the jumper wires and cleaned up the tags etc, leaving only the cable with the 50 pairs coming from the PAX terminated. The wood was polished using beeswax polish and it cleaned up a treat. I saved all the old jumper wire, which is red & blue fabric covered shellac'd copper wire and very authentic! (I have used some of it since to give it that 1951 vintage 'feeling' ) - the dis-case is now screwed onto a wall once more. The jumper wires are terminated using screws and the distribution cables are soldered onto tags. In case you were wondering what exactly a 'jumper' is - it is a pair of insulated wires used to connect the equipment terminals of the exchange to the terminals of the cables which are wired around the building and subsequently to the extension telephones. Thus when a new extension is connected a jumper is run between its cable pair and the equipment pair of the required number. This gives 'flexibility' in that any cable pair can be wired to any extension number by simply moving a jumper.


The selectors are standard 2000-type mechanisms with two wiper assemblies on the shaft. Only two wipers are required on this PAX because it has only 50 lines, so there is no 'wiper switching' circuitry involved (A typical 2000-type selector has three wiper assemblies, each with two wipers - there are ten outlets on each level of the selector bank and the wipers can be switched to either the upper contacts of the outlet required or the lower thus giving the selector 200 outlets with three 'wires' - two for speech and one for control and holding of the call). A further 'novel' approach is that the selector banks only need five levels (5 x 10 = 50 outlets for the 50 extensions) and so instead of having the usual levels 1 to 9 & 0 these banks only have contacts on levels 2 to 6. On the other levels the selector wipers simply step through space (any number dialled that leaves the wipers hanging in mid-air returns 'Busy Tone' to the calling extension). Thus the numbering range of the AT&E 50 line PAX is 21-20, 31-30, 41-40, 51-50 & 61-60.


The selector mechanisms themselves were stiff and needed cleaning and lubricating - a task which I've had to do many times when I was working on Strowger equipment in the distant past for 'The Post Office' - but never with quite this much 'job satisfaction'! I also polished them with a soft cloth and cleaned all the relay contacts. There wasn't much wrong with them after that.


Other mechanisms in the cabinet that needed attention were the Linefinder Uniselectors and the Allotter Uniselector (the Linefinders are associated with a Selector and when an extension wishes to make a call a Linefinder will search for that extensions calling equipment 'on behalf' of the Selector... this means that the 50 extensions can 'share' 5 selectors - the 5 linefinders doing the 'dishing up'. Likewise, in order that each selector 'takes its turn' and all get even wear-and-tear, the Allotter Uniselector dictates which Linefinder will be the first to be used on the next call - stepping on each time a call is begun). All 6 of the Uniselectors were removed, dismantled, cleaned and oiled, banks cleaned and mechanisms replaced. I'm sure one of them sighed afterwards.


For some reason-unknown all of the old 0.5 amp type 44 fuses had 'blown' during transit and needed replacing. One of the relay had suffered a bent contact which was adjusted with no apparent side-effects and all relays were checked for proper armature operation and contact wear - the only relay with excessive wear was, predictably, the vibrator ringer - however it works and is best left alone until it doesn't! (a vibrator Ringer is a special type of relay that 'self interrupts' and thus oscillates - a separate set of heavy-duty contacts are used to provide a crude form of alternating current to feed out to extensions when their bells are to be rung - it is these contacts that wear out - that is why I used the word 'predictably' earlier). Quite a few of the relays were 'creaky' at first but a few taps and the odd armature cleaning cleared those problems. All was ready for 'power on'...


Because there was no mains eliminator supply within the PAX cabinet I needed a 50V supply - the PABX No 6 is sited near to this PAX and it was simply a matter of running a hefty copper tie-cable from one to the other. The PABX No 6's supply can handle the two units with no problems - especially with the miniscule traffic loads that will be presented in their present home.


At first I was a little confused as to what was going on... there were two power feeds plus the earth return. I wondered if the PAX needed both a negative and a positive supply perhaps, but a telephone call to someone 'in the know' revealed all. It seems that in the early days, when smoothing of power supplies was costly due to the capacitor sizes required, two separate 50V supplies (both negative) would be employed. One supplied a higher power but raw (ie the smoothing was minimal) -50V DC to the switching relays and selector magnets - the other supplying a smoothed -50v DC purely for the speech transmission bridges. Thus the speech would be hum-free during conversation - the relays holding the call connected with their humming current having no effect on the transmission quality of the call whatsoever.


Nowadays the size and price of capacitors has reduced considerably compared to their capacity - thus all that was needed is a single -50V feed and the two separate inputs teeing together.


I took the precaution of measuring the resistance across the power feed and there was no movement of the scale... only when a loop was applied to one of the extensions lines, or a selector was stepped manually off-normal did the resistance lower. All was well. Power ON. No smell of burning or fuses blowing. I pressed and released an A relay in one of the selectors - it stepped then released around the bank and plopped out of the other side... a historic moment indeed! 17 years of silence broken by the primordial 'raspberry' of a 2000 type selector rotary interrupters at work. Some minor adjustments were necessary on some of the uniselectors but considering the amount of time that the PAX had been un-used it was in remarkably good shape.


All of that is now history... I have managed to copy a set of diagrams and diagram notes - and during the summer I was lucky to hear of another AT&E 50 line PAX that was being scrapped at an electricity substation in Kent. I travelled down with another like-minded accomplice and whilst he was cannibalising two other PAX's I was able to recover almost every useful spare from the 'donor' PAX. I must say though that it was a great pity to have to cut through all those wires and to finally leave the empty cabinet 'ravaged' in a skip... I would have liked to have transported the whole thing back to Leicester but it would have been expensive hiring a van and paying for petrol etc. The PAX that I cannibalised was a more recent model than the one at home - it was a sad sight to see when I opened the door... there, neatly tucked away in the bottom of the cabinet were the lubricating kit, cleaning cloths and bank-cleaning tape, a selector adjustments instruction book and a test telephone No 180 - all where the engineer had left them before it was scrapped. They are now neatly tucked away in my PAX! I recovered spares for my own PAX and was pleased to find that the two-motion selectors were of a different circuit design to the ones I have at home. The difference is that they have an extra relay fitted - a 'TL' relay which is used for inter PAX tie-lines (hence 'TL').


I have put these new selectors in my PAX and have rigged up 'tie-lines' between it and the PABX No 6. The tie line relay in a selector will operate when a designated number is dialled. That number has special 'straps' inserted which cause the relay to operate if the wipers step onto the designated line's outlet. When the TL relay operates it prevents ringing current and Ring Tone being applied as would happen on a normal call to a free extension - it instead allows loop disconnect pulses from further dialled digits to be applied to another PAX via a tie-line and also provides a transmission bridge. Thus I dial the digits '39' and instead of ring tone I receive a further dial-tone from the PABX No 6 and can dial one of the extensions on that. The new selectors also have covers fitted which keep out the dust and look quite impressive after polishing.


As I mentioned earlier there is no load on the exchange power supply except when a call is in progress - unlike the PABX No 6 which has all sorts of resistors and relays hanging across the supply... these obviously draw a standing current and are thus warm, some even hot, and in the sealed cabinet of the PABX the air remains warm. As it was winter when I first put the PAX out in its new home I was a little concerned about it getting damp through condensation (I suppose 17 years in a cold basement didn't really do it any harm - but 'better to be safe than sorry'!) and so I mounted a light fitting on a wooden base and connected a 240V 100W light bulb accross the 50V supply. It glows dimly in the depths of the cabinet and gives a small amount of warmth whilst all around is cold.


The house is now 'cabled' and there are phones off the two exchanges all over the place. I also own a GEC 25 line uniselector PAX which I have had for many years - it has a glass front and looks far too good to be out with the other two - it's in the front room where it's warm! There are just two telephones connected to that for the odd demonstration - I don't trust its power supply for constant use. When I first got that all the selectors were gummed up solid with oildag - but that's another story!


Finally some miscellaneous facts and figures about these PAXs...


The only reference, other than the AT&E documentation, that I have come across is in 'Complete Electrical Engineering' Vol II (E.Molloy). There is a chapter titled 'Private Automatic Telephone Systems' by C.W.Mant which gives a brief explanation and has a few illustrations... it also states that, besides the 50 line PAX, AT&E also produced a 10 line version using uniselectors, a 25 line version with a maximum of four two-motion selectors and a 100 line version which had ten two-motion selectors (max) for connecting calls and ten two-motion selectors used as linefinders (Two of these racks could be used to provide a 200 line unit).



Tones and Ringing

Ringing Current is, as I mentioned earlier, obtained from a Vibrator Ringer arrangement. Vibrator relay 'RV' self oscillates at 25Hz and provides intermittent connection of direct current to a transformer primary winding induces alternating current into the secondary winding - a choke and capacitor are used to make the waveform more sinusoidal. Ringing Currents use is described below.

  DIAL TONE is actually a feed from the ringing current fed through a simple frequency doubler and is thus a continuous 50Hz tone - a sort of rasping mains hum!


  BUSY TONE is created by a self interrupted relay 'BV' which oscillates at around 300Hz (it sounds a bit like a wasp in a jam-jar). The 300Hz tone is then interrupted every 2/3 of a second ( my guess!) giving a continual on-off-on-off sequence.


  These interruptions are controlled by a series of slow operate and release relays which also interrupt both DIAL TONE to give INT(errupted) RING TONE, and Ringing Current to give INT(errupted) RING which is used to ring the extension telephones bells when required.

Note: words in capital letters above are those actually used to describe the various relays and supply feeds in the AT&E documentation.


Release Alarm

A two-motion selector is quite a complex piece of engineering - the 2000 type model is very reliable but can at times require adjustment. One particularly vunerable area is that of 'releasing' after a call has finished. The selector utilises powerful electro-magnets to operate and, if they are energised for long periods, can be damaged or even become a fire hazard (a rare occurance indeed - usually a wisp of smoke and awful burning smell). The usual candidate for this kind of ordeal is the Release Magnet, which can remain energised due to a mechanical failure of the selector to release.


A circuit element known as a 'Release Alarm' is used in most Strowger exchanges and this PAX is no exception. All Release Magnets feed their energizing current through a special 'Release Alarm' relay RA. Thus as well as the release magnet operating so does the 'RA' relay. This operates further circuitry, including a 'thermal' relay which takes approximately 30 seconds to operate. If, during this time, the selector fails to release (the Release Magnet still energized due to the fault) then the release alarm reduces the current supply to all selectors in the PAX. Only a small 'trickle' of current is allowed to flow... this is used to monitor the release magnet in case it does release.


This reduction of current means that any established calls on other selectors will not release the selector (although the subscribers circuit will release normally) after the call has finished... to overcome this difficulty it is arranged that as soon as a new call is attempted by a subscriber then full power is once again re-applied to the selectors release magnets. if our original failing selector is still 'stuck' then the whole cycle is repeated... but quite often a 'stuck' selector can release perfectly happily once the power has been disconnected and re-connected. The only way that a problem of this kind would be noticed is if a diligent engineer gave the switch a routine test or it finally fell to pieces and remained stuck ( even then it would still need an engineer to inspect the equipment to spot it). A bell can be connected to ring when such an alarm occurs to notify an attendant.


Fuse Alarm

If a fuse should blow due to a fault condition how is anyone to know? A simple mechanism is employed using a special type of fuse consisting of two springs held together by the fuse wire. Should the fuse 'blow' then the two springs fly apart... one connects the original supply to an alarm conducter that it springs onto, the other sticks up proud of the fuse and serves to indicate which fuse has blown. The alarm conducter is wired to a relay which operates - again it could be wired to ring a bell to attract the attendant.



The overall dimensions of a 50 line PAX are:-


  Height .... 4'6"

  Width ..... 3'2"

  Depth ..... 1'4"


  However mine has an extra unit (see below - Key calling) mounted on top which adds an extra foot or so to the height. The whole of the equipment is mounted on a steel rack totally enclosed by sheet iron covers 'suitably finished' (quote from AT&E's documentation).


Special Services

In addition to the normal PAX telephone service any or all of the following special services can be (or could be) supplied.


1. Priority cut-in on an engaged subscriber.

2a.Tie Line working between 2 or more exchanges by dialling one routing digit for each group of tie lines.

2b.Tie Line working between 2 or more exchanges by dialling two routing digits for each group of tie lines.

3. Secretarial Service.

4. Key Calling by means of an external Desk Cabinet.

5. Key Calling Service combined with Conference Facility by means of external Desk Cabinet.

6. Code Call (two methods) (a) Bells, (b) Lamps and Bells.

7. Fire Alarm Working.


  " The PAX is normally supplied without the additional apparatus necessary to provide the special services, which in all cases require to be ordered separately."


  "The design of the equipment is such that any or all of the services can be added at any time."



Thanks to Norman Pearce, Andy Emmerson and Steve Bradbury for technical advice, and to Mark Walters for the help with moving it!


Alan Gildersleve


This small and very compact exchange known as LCC (Line Circuit Control) was used by Telephone Rentals in large numbers from the late 1930s onwards.


It was a 26 volt exchange and was notable for its very few relays. There was a maximum of 4 connector circuits, each with its 'jack-in' relay set, with about six relays in each. [These had round-fronted cans, just like other similar pre-2000 apparatus. Most of the relays (all but the vibrators) were 3000-type. - AE].


The most interesting things about the exchange were the multi-purpose line relay and the very crude but efficient ring and tone system.


The line relays (one per line) were used as line relays, A relays, and F relays rolled into one. In other words they operated on lifting the handset, were impulsed by the dial, and tripped the ringing on incoming calls, when handset was lifted.


There were two Ericsson-type uniselectors per connector, i.e. one linefinder, and one connector. The interdigital positions of the connectors were at 0, and 00 so the numbering scheme was 1 to 9, 01 to 09, and 001 to 004. Later PAXs came out with a capacity of 25 lines and this was achieved by doubling up on contacts 1, 2, and 3 (contact 1 being the home position and dial tone feed). These contacts fed extensions 1, 2, 3, and 005, 006, and 007 by means of a switching relay which operated from the 00 (21st contact) position of the connector switch.


The way the ring trip worked was that the line relay was pulsed in and out by the AC ringing via a small metal rectifier mounted in the small can which housed each line relay. These cans had to be in place otherwise there was cross talk between adjacent lines because the line relay was also used as the speech bridge. When the line was answered the line relay was held in by the speech loop and because it had not released the timing of the ringer circuit was such that the ringing was cut off. If answered during a ring there was a loud burp of ringing in the receiver. If the line relay was sticky this caused premature ring trip.


The ringing was produced by a polechanger vibrator and there was no transformer. All it did was to reverse the polarity of the lines rapidly so as to cause a rapid bell tinkle due to the instrument capacitor charging and discharging. Special arrangements had to be made for extension bells etc.. The ringing cadence was the usual 0.5 seconds on, 0.2 seconds off type used by most internal systems.


Tie lines and Round Call could be fitted. These used a Call Finder to pick up the line that was making the call and threw it off the linefinder circuit by the simple expedient of stepping the line finder off the line and homing it, so as to leave it free for others to use. If Round Call was fitted then extensions 7 and 8 were not available for lines, and if a Tie Line was fitted then extension 9 was used. The Round Call was a 10 code staff locator. One dialled 7, waited for dial tone again (from the Round Call unit) and then dialled a number from 1 to 0. This caused a uniselector to be stepped round slowly by interacting relays, passing over various strapped contacts to send dots and dashes on buzzers and hooters around the factory. On hearing one's code, one dashed to the nearest extension and dialled 8 and carried on a conversation. The strappings were controlled by a uniselector which was stepped by the code digit 1 to 0. The codes were . .. ... .- -. .... ..- -.. -- and .-. The speech bridge was incorporated in the jack-in Round Call unit.


These exchanges were very slim and bolted to the wall, and were on a grey steel frame with a steel cover over the switches. Earlier models had a grey wooden cover with a glass pane, covering the switches.


>>>Alan and I would be pleased to have your feedback on this article: did you find it useful, would you like to read more like it, do you need more detail? Does anyone else have one of these fascinating little exchanges? They were extremely common at one time and were still being manufactured into the 1970s, but they are disappearing rapidly now. I have one of the pre-war S22 types, originally used at Marylebone station. The Festiniog Railway in Wales had some of these to begin with, but I'm not aware of anyone else with one. The later S25 version (with modern uniselectors, 25 lines and modern square-fronted relay cans) is used on the Kent and East Sussex Railway, and I know at least one other THG member has one [Andy Emmerson]

ERICSSON 50 LINE NON-EXTENSIBLE PAX, as used by Telephone Rentals.

Alan Gildersleve

This PAX was a very interesting attempt to produce a small 50-line exchange with a uniform 2-digit numbering scheme and it was produced in large quantities for TR. It was really a 'marker controlled' system.


It was equipped with a maximum of 6 line finders and connectors, all of the Ericsson type 25 outlet uniselector type, but each selector had a wiper unit consisting of 2 groups of 4 wipers at 180 degrees to each other, thus producing a total wipe of 50 outlets, i.e. 25 + 25. The banks were +, -, P & M; the M was the marking bank for the calling and the called lines. There were also 2 registers each consisting of a relay set, a register finder uniselector, a tens uniselector, and a units uniselector.


Upon lifting the handset both registers attempted to operate but the register with the faster operating RS relay cut the other one out and took the call. If the register finder was not standing on a free line finder it stepped until it found one. This line finder then drove to find the calling line by means of a voltage on the appropriate contact of the M bank, whereupon it stopped and dial tone from a vibrator was received.


The required number was then dialled. The first train of pulses stepped the tens uniselector, and during the Inter-Digital Pause the tens selector was replaced by the units selector which then stepped to the required units outlet. The operation of the tens switch connected a marking voltage to one of the banks of the units switch, and this switch then extended the marking voltage to the M bank of the connector switch which then drove to the wanted line. When this had happened the register released and the tens and units switches homed, first the tens and then the units: ZIPP, ZIPP.


If the wanted line was free, ringing was extended to it in the usual manner after the H relay in the connector relay set had operated. If it was busy, then the H did not operate, and busy tone was heard. There was a special circuit on these PAXs whereby if the calling line had a priority facility one could press the priority button and butt in to the conversation and request both parties to replace their handsets, whereupon the bell of the wanted party would ring and the conversation would then be private. This was done over a 3rd and 4th wire (earth and priority).


As the extension line numbers ran from 11 to 59, the tens digits 6, 7, 8, 9 and 0 were available for miscellaneous purposes. 6 was usually reserved for conference facilities. 7 was always Round Call originate (as per the 25 line PAX dealt with last time), 8 was always Round Call answer. 9 and 0 were tie line access digits, although any of these numbers could be strapped to any facility. If any of these facilities were used, then one or more of the lines 56 to 59 could not be used for extensions. The addition of the appropriate strap by-passed the units switch and drove the connector to whichever line number had been chosen, i.e. 56 to 59.


The tie line and round call units were all fairly standard but there was a useful facility of priority access to tie lines if one had a priority telephone. As with an engaged extension, one could butt in and instruct the parties to clear the line and seize it for oneself.


As these PAXs ran on 50 volts nominal, there could be problems when tie-lines went to a 25 line 24 volt PAX. This involved fitting rectifiers in the line to prevent a permanent loop being set up and of course the incoming tie line relays TA had to suit the incoming voltage from the other end.


These PAXs were in use right to the very end of the internal telephone era. They first came as wall mounted units with a glass fronted cover over the uniselectors, and several had a larger glass cover that covered the line relays also. Then they came as an 'island site' unit and the latest ones used standard P.O. type uniselectors which did not run nearly as smoothly as the original Ericsson type with the hexagonal nut on the front.


These 'island site' units were in a battleship grey steel cabinet with a lift-out door front and back.


There were many versions of these PAXs right up to about issue 14, I believe. The schematic number of the later issues was 63612 followed by an oblique stroke followed by the diagram section number. Under this was printed ISSUE (whatever it was).


Originally they used smoothed and unsmoothed voltages, and on several occasions an engineer wired the smoothed voltage to a round call unit instead of using the unsmoothed. This caused all the bells to tinkle in time to the stepping of the round call code switch due to the voltage fluctuations! I still have one of these round call code units.




Alan Gildersleve describes a novel - and obscure - product which was TMC of Dulwich's only venture into the automatic exchange business. Called Temco, it had a single motion switch with up to 75 vertical levels. Each extension telephone was connected to the exchange by two wires and had its own switch (36 volt working); you could have as many conversations as people wanted. In effect the switch replaced the manually-operated rotary selector normally found on intercom telephones, and the circuitry was quite similar, with a 'home line'. The action of a DC trembler bell in the called extension gave a ringing tone back to the calling extension.


Each telephone had a knurled knob, which the user set to one of 75 positions. This sent the desired number of impulses in one train. Exchanges with as few as 25 lines were made, mainly for Telephone Rentals Ltd ,who shared the same boss (a Mr Jackson) as TMC in those days. Foyles Bookshop in London ( who still have a Tele Rentals PAX) had one of these Temco PAXs for many years, and three instruments and part of a switch were saved for the Science Museum.



Back to Technical index