"The phone is a remarkably complex, simple device,
and very rarely ever needs repairs, once you fix them." - Dan/Panther

Main Menu

Miniature 10-line telephone exchange using 22 relays

Started by Owain, January 29, 2012, 10:21:34 AM

Previous topic - Next topic


Good luck! Otherwise you can maybe explore some of the many other options discussed on the forum.



I had a quick PM from GTC, who hasn't made the PBX yet but still hopes to get around to it.


I received a similar reply to my PM...

I am also in the early stage of building this little exchange since I have most of the relays which will do the job - I only have to source a few on Ebay. I will be using all 48V relays and a PBX 48V power supply with integral isolated 80V,25Hz ringing supply.

On studying the schematic of the 10 line exchange I was rather perturbed by the rather crude design of the called line battery feed relay which also serves as the answer ring detect relay. Insofar as the speech battery feed goes I have no problem but the ring detect function just can't work - at least not in any of my bench experiments and certainly not with 25Hz. There is no way that a standard off-the-shelf relay serving as relay D can be made to be AC immune to the ringing current simply by shunting one half cycle of ring current through a diode-resistor shunt arrangement - it still buzzes no matter what value of resistor I try.

My other concern is that the way the circuit is arranged... if a called telephone answers during a 2 second ring burst it will produce a loud 50/60/25Hz buzz in the receiver which could blast the eardrum depending on how quickly it is placed to the ear. The only time relay D can operate to an answer loop is during the quiet period between ring bursts when the -VE supply line is connected behind the D relay.

I am going to use the standard DC biased ring feed arrangement similar to the old Strowger final selector where the ringing supply has +VE behind it and returns via the called telephone to 200 Ohm -VE. In the case of the 10 line exchange all of the telephones use a common +VE on one side of the line which demands that the ringing supply has -VE behind it and this is shown in my sketch of the ring feed/trip arrangement below. In the final selector a double winding heavily slugged relay (usually F) has ringing fed to line via one winding while the other winding is held short circuited by an 'x' (early operate) normally closed contact which in conjunction with the slug, makes it immune from operating to the AC ringing current. However, when the called telephone answers the DC loop DC current flows through the ring feed winding which causes it to operate sufficiently to remove the short circuit from its other winding which then locks solidly to external circuitry via both of its windings. Relay F thus acts as the ring trip relay and it will operate reliably during bursts of ringing or during the breaks between thereby preventing any possibility of painful bursts of ringing in the eardrum.

The only problem with this is that short of pulling a rather large F relay out of an old selector I have not been able to find a small PCB mount relay with the right characteristics to replicate the function of the F relay used in the selector. Not to be deterred, I found I had a good number of Electrol TLS1A18A10 line sense relays which have two 18 Ohm windings and a single NO reed contact inside. Admittedly, not a slugged relay but I figured I could get it to do the trick (I actually experimented with a line sense relay made by Erg but the only difference is that it has two 25 Ohm windings). By connecting the windings in series opposing and shunting one winding with a diode I was able to make the relay immune to the AC ringing current through it but still allow it to operate to an answer loop which causes the windings to become unbalanced with respect to each other and thus to operate to the DC loop current. Sure, the line sense relay will still chatter at the ringing frequency but at the first make of its contact a standard PCB relay will operate and lock thus cutting the ringing current and causing the D relay of the  exchange to operate and perform its normal battery feed function.

This is the circuit I came up with.


Sounds exciting.

I must confess that following the circuit diagram to understand it is a bit beyond me!

I don't think I've found this page before, but the author discusses the problems of ring trip at 20 Hz on home exchanges (scroll down to Ring Trip section)



Thanks for the link to Dale Thompson's little exchange. I must admit I hadn't come across the web page before despite many Google searches - I guess my search terms didn't include Home PBX or the like. Perhaps he should have titled it to include the terms 'telephone" or 'exchange' which is what I would have searched for.

I must say that he has done a superb job on the design but is certainly looks a lot more complex than Andrew Holme's 8 line exchange. Unfortunately, the page format on my old monitor makes it necessary to scroll the page horizontally in order to read the text and then scroll back to refer to the circuits.

He does cover the Ring Trip problem very well but doing the task with semiconductor devices makes it very complicated when  just 2 relays and a diode make it easy to do. The sketch I attached to my last post is only what I did on the bench to prove that it worked and doesn't show how it hooks into the 10 line exchange circuit, but this detail is easy to work out. I will draw out the relevant section of the circuit to show how it can be done and post back later.

I will takes copies and print out the text so I can study it more in depth when time permits.


Quote from: rherber1 on June 26, 2018, 10:26:08 AMdoing the task with semiconductor devices makes it very complicated when  just 2 relays and a diode make it easy to do

Generations of telephone people would agree with that.


Further to my previous post on a modified ring trip arrangement for this 10 line exchange I have come up with a sketch showing how it would be incorporated into the original circuit. During my study of the circuit diagram in conjunction with the description (this is rather vague and difficult to follow btw) I noted a couple of inconsistencies which require circuit changes. Firstly, to prevent an undesirable momentary operation of relay FC during clear down following hang up, and secondly, the sending of ring tone to the calling line before dialling is completed.

Taking the 2nd point first - when a call is initiated relays A, B and C (among others) operate. At the first break of the dial impulsing contact relay A releases and FC operates via B2 and the diode. Contact FC3 connects ringing via RI1 normal - FC1 operated - D coil - via the 1u8 bridging capacitor and 100nF + 3K3 (ring tone bleed) directly to the top of the A relay and out to the calling line via contact CA2 operated. This means that ring tone is heard by the caller even before dialling is completed. To correct this anomaly it is necessary that the 1u8 bridge capacitor is connected at the other side of the C2 contact to that shown in the circuit. Thus, only when relay C releases at the end of dialling does ringing get connected to the called line, and at the same time ring tone is fed via the 1u8 capacitor and tone bleed to the calling line. Only after the called line answers and relay D operates does the 1u8 capacitor get connected directly between the calling and called lines when speech can take place.

The other point I noticed was that at call completion, because relay B is slow to release due to the capacitor across it, a re-operate path is established for relay FC via the now released contact A1 - via contact B2 (slow release) - Diode at B2 - relay FC and the -VE rail, thus allowing FC to operate for the time it takes for B to fully release. While this momentary operation of FC doesn't appear to do anything untoward it is unnecessary and not good practice. The solution is to introduce a break contact into the initial operate path for FC via the diode at contact B2 which is only closed again after relay B has finally released.

I have incorporated both of these solutions into my modification. I have included only the relevant section of the circuit to show how the modification is applied.

Note that relay LD is a line sense reed relay (Erg PM55/3 or Electrol TLS1A18A10) - I experimented with the former which has 25 ohm coils while the Electrol has 18 ohm coils.

The windings of LD are connected in series opposing so that it will not operate to the AC ringing current (25Hz) going to line. Only when the called telephone answers does DC current flow via the windings of relay LD from +VE via the answering telephone loop - C2 released - RT2 released - diode on top coil of LD - lower coil of LD - FC3 oper - RI1 (either normal or operated) and back to -VE. The DC path for LD is completed to the -VE rail behind the ringing supply when RI1 is released or direct when RI1 is operated thereby allowing ring trip to occur as soon as the called line answers (does not have to be during the silent period as in original cct). Due to the shunting effect of the diode on the top winding of LD the current in the coils is now unbalanced and LD will operate. At the first make of LD1 contact relay RT (new) will operate and lock via RT1 to +VE at B1. RT2 switches the called line over to relay D - which operates, while at the same time opening the ringing supply to the called line. Contact D1 establishes the speech bridge via the 1u8 capacitor (I show 2uF).

After a call is completed relay B will remain operated as long as the called line has not yet hung up. When the called telephone hangs up relay D releases and breaks the holding path for relay B at D2 (now normal) and relay B releases slowly due to the shunt capacitor. Only after B1 is open will relay RT release and via RT3 re-establish the initial operate path for relay FC at the next call establishment.

It also occurred to me that the ringing interrupter flip-flop - due to the same timing components being used on both sides, would seem to indicate that rather than the cadence being 2 seconds on and 4 seconds off (as stated in the description) it would be more like 2 seconds on and 2 seconds off.


That's really interasting the other day I was designing a circuit using 4 relays for a small phone exchange I was going to use in my new service called NPSTN for New Public Switched Telephone Network its similar to C*NET.

If anyone wants a number on the network just PM me :)

But thats for posting this I will see if maybe I can use this circuit! :)


I should have spotted it earlier but it occurred to me after posting my circuit that contact FC3 is now redundant since its function (disconnect ringing supply at answer) is now performed by contact RT2. Therefore FC3 should be taken out of the circuit.


The first part of the Maplin Digi-Tel electronic exchange from the 1980s is in this online version of the Project Book 04. It uses the 45100 crosspoint IC.


Quote from: Owain on July 01, 2018, 08:19:45 AM
The first part of the Maplin Digi-Tel electronic exchange from the 1980s is in this online version of the Project Book 04. It uses the 45100 crosspoint IC.

Thanks for the link.

It is quite a sophisticated project and it wouldn't be cheap to build - even assuming all the parts were available. For instance, I can't find one hit on "45100 crosspoint switch" when searching for a datasheet with Google.

I have spent some considerable time scouring through the magazines and journals on this website and it is quite extraordinary what you can find. It is a great resource.


Quote from: rherber1 on July 01, 2018, 11:44:52 PM
It is quite a sophisticated project and it wouldn't be cheap to build - even assuming all the parts were available.

It wasn't (and I didn't!). Even their "low cost" telephones were about £30-40.  I wouldn't expect that many people actually did build it, but it's amazing the complexity of some of their projects using the technology of the time. Maplin Matinee electronic organ, MOSFET amplifiers. They launched a build-your-own modem kit and had a dial-up ordering/bbs system.

For non-UK readers who don't know, Maplin finally went into administration this year and closed the last of their shops a few weeks ago.



Update on progress:

I have successfully built the dial pulse counter section along with the called line decoder contact tree. I used Panasonic/Aromat DS4E-M-DC48V relays for the SA,SB,SC,SD relays and Meisei M4-H-48 (Now Maluska Elektronik) for the 2 C/O type relays since I had a hundred or more on hand.

I am now in the process of designing the calling line selector circuit. The original circuit uses 4 relays with a diode encoder to produce a 1,2,4,8 BCD encoder circuit around CA, CB, CC and CD relays. When I looked at this arrangement it occurred to me that if 2 phones happened to be lifted simultaneously (unlikely, but possible) ie, Tel 1 and Tel 2, the calling line selector would operate relays CA and CB together and this would decode as Tel 3 originating the call and operation would be halted until both phones hung up. While the selector circuit saves on relays it doesn't provide an exclusive 1 of 10 selection. I decided to scrap it and use 10 individual line relays with a chain of contacts wired so as to allow only one relay to latch if more than one line is looped simultaneously. This means the smaller inexpensive M4 relays can be used instead of the expensive 4 contact types (CA, CC, CD).

Also, the original circuit did not guard against the calling line dialling its own number. The call would progress as normal and a burst of ringing would go to line but would be tripped immediately due to the calling line loop. I think the caller would still receive a short loud burst of ringing in the ear. By arranging for one contact on each line relay to be inserted in the called line contact tree formed by the SA - SD relays then when a calling line is identified by its line relay operating, it disconnects its own line from the called line tree and at the same time connects the line through to the A relay.

Further reports to come...