Using your Telecom Gold Phone on the NBN

Using your Telecom Gold Phone on the NBN.

As described in this article, converting to the NBN for my phone service entailed a number of modifications. Most significant of these was the installation of a Dialgizmo to restore decadic dialling capability to the new service. All the vintage phones could then make outgoing calls. That is, except for the Gold Phone. This problem was overcome, and the following article describes the development of a plug in interface to achieve normal operation once again.

Introduction to the Gold Phone.

The Gold Phone design and development was described in the Telecommunication Journal of Australia, volume 34, No. 1, 1984. This can be downloaded here.
Service and installation manual can be downloaded here.
Operating instructions can be downloaded here.

The Gold Phone was introduced to replace the Red Phone, which had been around since the 1960's. The Red Phone was suitable for local calls only, and with this limitation, the Gold Phone was developed to allow STD and ISD calls, since these require a variable charge rate.
Being of a 1980's design, push button dialling was provided. For most Gold Phones, this was a decadic dialler since DTMF capable exchanges were not common at the time. However, a DTMF dialler was available, but this appears to be quite rare.
The phone is polarity sensitive. If the polarity is incorrect, the line is disconnected after about three seconds. Coin collection is controlled by 50Hz meter pulses, and for this an earth connection is required.
Because of the variable call charging and electronic coin identification, the phone is designed around a microprocessor. This, and the coin collection relay, is powered by a couple of super capacitors charged from the line, at a current low enough not to loop the line. It's typically 1-2mA.

There is a key switch on the side of the phone with three positions; C is normal operation where the caller is required to insert coins on Chargeable calls. D does not require the insertion of coins, but the Display shows what the call charge was. This is used in situations where the caller is billed by the owner of the Gold Phone after the call; e.g. added to a restaurant bill. L is the Lessee position, and the display remains blank at all times. No coins are required to be inserted to make chargeable calls, and is used by the owner when the only phone on the premises is the Gold Phone.

Using a Gold Phone at Home.
Contrary to what has been said on various forums, it is possible to use a Gold Phone on an ordinary exchange line, but of course without the call charging capability. It will allow outgoing calls and receive incoming calls normally. The key switch can be in any of the three positions, but I keep it in the C position - the display comes up with $0.00 when the phone is off-hook. It is not necessary to earth the phone, but correct line polarity is important. In fact, attempts to earth the phone resulted in hum on the line, evidently due to some leakage causing the line to become unbalanced. While the earth connection is partly for surge suppression, the fact that I got away without this for 23 years would indicate it wasn't that important. Yet, the Gold Phone was exposed to the same surges that damaged my answering machine numerous times. I have two Gold Phones; one fitted with a 3m cord and 603 plug. This is the one in use. The second one has a 30cm cord with a terminal block, fitted with a surge arrestor.

The Gold Phone and the NBN.
Two things were immediately evident with the phone connected to the modem via the Dialgizmo. It would not dial out, and after three seconds, the line would disconnect. The line disconnection problem alerted me to incorrect polarity, and indeed, I found that the Dialgizmo had been wired so that it reversed the polarity. This was easily restored by swapping the red and green wires around in the Dialgizmo. However, the line still disconnected, although after a longer period. This seemed odd because the modem puts out the usual 50V DC line voltage, and the Dialgizmo only has a 3V drop across it. When the Gold Phone was connected to a 30V supply limited to 50mA, it remained connected indefinitely.
The problem now was that while the Gold Phone could loop the line and the speech circuit worked, it was still unable to dial out.

Why the Gold Phone won't work with the Dialgizmo.
After some measurements of the dialling waveforms were taken, it became evident why the Gold Phone cannot dial out using the Dialgizmo. The problem is essentially how the Gold Phone loops the line, and dials. It is different to a 'normal' phone as the waveforms show:

For the test, a 30V power supply was used, with resistance to limit the current to around 30mA. While the normal telephone supply is -48V, the difference is irrelevant for the purpose of testing.
Starting with a 'normal' phone (a series 800 with mechanical dial), we can see that when the phone is off hook, the line voltage falls from 30 to 5. When the dial is rotated to the finger stop, the voltage falls to 0, and as the dial returns, the voltage rises to the full 30V when the contacts open, then back down to 0 when they close again. At the conclusion of dialling, the voltage rises back to 5V and the speech circuit is operative again.

When the Gold Phone is tested, there is a marked difference in operation. When the phone is taken off hook, the line voltage falls to 15V. Then, when dialling commences, the voltage rises slightly higher to 17V while the dial contacts are closed, and then up to the full 30V when they open.

This is not a problem for a loop disconnect exchange, since it senses dialling by the absence of line current. However, the Dialgizmo operates by measuring line voltage, rather than line current. Since the Dialgizmo was designed for mechanical dials (and the electronic equivalent thereof), it is expecting to see something close to 0V when the dialling commences, and each time the dial contacts close.
Hence, the Gold Phone dialler is invisible to the Dialgizmo.

It should be noted that the Gold Phone cannot work, even momentarily, with less than about 12V line voltage. It will disconnect from the line, and not reconnect until the line voltage is increased, and the phone is reset by replacing the handset.
Ideas of simply shifting the DC voltage by connecting a 15V supply of opposite polarity in series with the phone won't work therefore. Instead, the phone has to simulate a 0 to line voltage during dialling, while at the same time being continuously fed with at least 15V.

Designing an Interface.
To make the Gold Phone happy, and keep the line looped, it was clear that a local DC supply was required for the phone, and not the 'line' current. And, to dial, this would have to be done by a relay to provide definite loop disconnect pulses, from 0 to full line voltage. The end result was a completely successful interface. To explain how it works is best done by describing its development.

Interface allows Gold Phone to work on NBN phone lines.

Dialling Interface - proof of concept.

A relay coil was connected in series with the phone and the 30V supply. 30V was chosen for convenience, since it's what my regulated bench supply provides. It's a suitable voltage as most phones are designed to work down to at least 24V, which is what some PMBX's and PABX's provide. The normally open contacts were connected across the line.
The relay responded as expected to the Gold Phone line current, looping the line with the handset off hook, and actuating the contacts in time with the dial. The Dialgizmo was completely happy. So, dialling with a relay was proven. However, it's obvious there's no way for the speech circuit to be coupled between the line and the phone, so that although a number can be dialled, no conversation is possible.

The next step was to use the second coil on the relay as a transformer, so as to couple the speech circuits. (Some relays are made with two coils on the one armature). Now, we had dial tone clearly audible in the phone. Unfortunately, it was impossible to dial. I realised what was happening. Once the contacts made, the second coil was being magnetised with the line current, and so the relay would not release, either by dialling or putting the handset back on hook.
It occurred to me that if the polarity of one of the coils was reversed, the magnetism of one coil would cancel out the other, and allow the relay to release. This looked promising, because the on-hook and off-hook operation now worked. And numbers could be dialled - but to a point! If any numbers greater than 6 were dialled, the relay would stop operating. In fact it looked like it was getting weaker the longer the pulsing occurred. Evidently, the cancelling of the magnetic field was not quite enough. It's possible that loading one of the coils with a resistor might have fixed the problem, but transmission would be reduced.
In any case, the concept had been proven, and should work if a separate transformer and relay were used.

Line Isolation Unit.
The kind of transformer suitable for this application would be a 600R to 600R isolating type, commonly used for telephone equipment. Looking through my cupboard of phone parts I came across a couple of line isolation units. These were commercially made and used for isolating things like music on hold systems, or modems, from the phone line. In the olden days, the PMG, and later Telecom, were very strict about what could be connected to their lines. Line isolation units prevented the risk of 240V finding its way back into the phone line, as well as preventing the line voltage finding its way into the external equipment.
One of the LIU's looked very interesting since it also contained three relays, as well as the right kind of transformer. Time to trace out the circuit:

Rough sketch when I first traced the circuit.

Looking at the circuit, I could see that most of what I needed was here and it would only take a few modifications to convert this into the new interface. Let's first have a look at the circuit. RLA3 switches the phone line to either the LIU or to the normal phone. The incoming line is the usual pins 2 and 6 of the plug, and the outgoing line to the existing telephone feeds pins 1 and 5. Unless RLA3 is actuated (the LIU is not in use), the incoming line is switched out to the telephone which works normally. When RLA3 is actuated, the line is switched to the LIU and the telephone is inoperative.
The line connects to the transformer primary via RLA1. With the 1.5uF and 1K across its contacts, it was obviously used as a dialling relay; the RC network protecting the contacts as well as providing a suitable dialling waveform. RLA2 short circuits the transformer to prevent pulses being transmitted across the transformer, into the following equipment during dialling. In effect, RLA1 and RLA2 simulate an ordinary telephone dial in this regard.
Across the transformer secondary are a pair of back to back 3.9V zener diodes, intended to limit the audio voltage fed into the phone line from the external equipment. There is a 2uF 440VAC capacitor which couples the transformer to the following equipment. It provides isolation in case the equipment under fault conditions allows 240V to be fed into the LIU. The 100uF is puzzling and appears superfluous. The secondary is already isolated from DC by the 2uF. As it stands, the 100uF will never receive any polarising voltage anyway. It may have been included more as a theoretical thing, to make both legs of the transformer balanced, but in that case should be the same as the other capacitor. The isolated signal is available from the violet and blue wires.
There are a couple of RC networks across the transformer windings which would appear to be for frequency correction.
Each of the three relay coils is fed from 12V from the orange wire, and switched by the green, yellow, and brown wires.

Initial Test.
By connecting the 30V supply through the Gold Phone and through RLA1's coil, it would be possible to loop and disconnect the line, and dial out, as previously described. But, also by connecting the output of the LIU to the relay coil, there would be the necessary speech circuit coupling. The unit was wired up thus:

RLA1 (and RLA2) are reed relays with a coil resistance of 650R. This is an ideal coil resistance for shunt feeding the speech signal into the LIU. The line current through the Gold Phone was 22mA which is quite suitable. The circuit was completely successful, with good audio transmission and reliable dialling. The back EMF shunt diode across the relay coil had no noticeable effect. RLA2 was not used, since the transformer windings have a DC resistance of only 33R and would not impair the dialling. Loud clicks would be eliminated anyway by the dialling circuit already in the Gold Phone.
With the basic design proven, there were some things to tidy up before making a final version.

Line needs to be fully disconnected when Gold Phone on-hook. The dial waveform shaping components are otherwise across the line.
Ring facility so line switched directly to Gold Phone when there is an incoming ring voltage.
Power supply.

While RLA1 serves the dual purpose of looping the line as well as dialling, the 1K and 1.5uF are across the line when the Gold Phone is on-hook. The loading of these components could reduce the ring voltage as well as load the speech circuit unnecessarily. Seeing as RLA2 was not otherwise required, a cut of one PCB track and a piece of wire reconfigured it so its contacts were in series with the line. How to actuate its coil? Simply by putting it in series with the coil of RLA1. Normally, it would also follow the dial pulses by doing that. However, by placing capacitance across its coil, it will hold during the pulses, but release normally when the phone is on-hook. As it turned out, the superfluous 100uF was a suitable value for this purpose. Because of this capacitor, speech voltage is only developed across RLA1.

The interface under development. Many of my designs start like this.

Ring Circuit.
Since the ring voltage can't be transmitted through the LIU, it would be necessary to switch the phone over to the line when there was ring voltage. The hitherto unused RLA3 would be ideal for this since it is a DPDT type. It could be actuated by a simple opto-coupler ring detector.
I found a suitable opto-coupler on the first PCB I pulled out of my collection of dead power supplies. It was a Liteon LTV-816A. Because the coil of RLA3 is meant for 12V, a 560R 1W resistor was used to allow it to work off 30V. The opto-coupler transistor was capable of switching the relay directly, but to do so would require more LED current than I'd like, which would load down the ring current. Instead, another transistor was added, effectively making a Darlington pair. This provided reliable relay switching with very little ring current.

The opto-coupler LED is driven off the phone line via a .47uF capacitor to block DC, a 22k resistor to limit the LED current, and a pair of back to back 22V zener diodes which prevent the LED responding to any low level noise. Furthermore, there is no loading of the speech circuit since the 22V is much higher than any audio voltage on the line. A 1N914 diode across the LED prevents it seeing reverse polarity when fed with AC. The circuit worked perfectly first go, and for the first time on the NBN the Gold Phone worked normally.

Ring circuit under test.

Power Supply.
The final part of the design was to provide 30 to 50V for the Gold Phone. The idea of running it off my DC solar supply was briefly investigated, but the switchmode converter I tried caused an audible tone in the line. Since I wanted to get the Gold Phone back in service quickly, I just settled for an ordinary 240V powered supply. The first likely transformer I pulled out of the drawer was a Jaycar MM-2007, 30VCT type. Rectified and filtered, it provides about 42VDC with the phone on-hook. With one 2000uF filter there was some hum audible, which was reduced when a resistor was connected in series with the rectifier output. It was completely eliminated with a second filter capacitor. I may have another look at the possibility of running the interface off 12VDC at a later date. Given the low current required, the only challenge is eliminating noise on the phone line. Powering the whole thing off the phone line is something I might also give some thought to.

Final Assembly.
Conveniently, as most of the circuit was already on a PCB, I only had to make up a board for the ring detector and power supply. I also added some LED status indicators. The whole thing was housed in a plastic instrument case.

Interface completed and working perfectly.

Circuit and Operation.

Complete circuit of the new interface. See notes further down concerning the installation of a 100R resistor in series with the RLA2 contacts.

The open circuit line voltage to the Gold Phone is around 42V, which is the rectified and filtered voltage from the 30V transformer. An LED shows the presence of this supply. The Gold Phone polarity is important, and the L1 connection must be positive.
When the Gold Phone is taken off hook, current flows through the coils of reed relays RLA1 and RLA2, through the normally closed contacts of RLA3 and into the L1 connection of the Gold Phone. From the Gold phone, the current leaves via the L2 terminal, through the other set of N.C. contacts on RLA3, and to the negative of the power supply.
In series with the reed relay coils is another LED which simply shows current flow, indicating the phone is off hook. It's shunted by a 470R since the line current can be higher than the LED ratings. Current is about 25mA off hook, and not much higher if the line is actually shorted. It therefore has a kind of constant current characteristic.

RLA1 and RLA2 contacts now close. Current from the L1 connection of the incoming phone line flows through RLA1 contacts, through the transformer primary, and back to the L2 connection via RLA2's contacts. Thus, the line is now looped and dialling can commence.
Since the Gold Phone dial interrupts the coil current of RLA1, its contacts interrupt the line current flowing through the transformer primary, and the line sees this as dialling pulses.
In conjunction with this, the 1k and 1.5uF RC network across RLA1's contacts suppress any excess voltage due to inductive components on the line, and provide a suitable pulse waveform.
RLA2 does not, however, respond to the rapid dialling pulses. Across its coil is a 100uF capacitor which keeps the coil energised during the loop disconnect period when dialling occurs. RLA2 functions as the equivalent of the gravity or hook switch on a phone - it loops and disconnects the line, but is not used for dialling. The purpose of it is to prevent the 1k and 1.5uF loading the line when the phone is not in use.

Since the transformer is carrying the line current, speech signals are coupled across to the secondary. These signals find their way into the Gold Phone via the 2uF and are shunt fed across RLA1. Since RLA1 has a high inductance and resistance, it does not impede the speech signals. The 2uF also prevents the low transformer winding resistance upsetting the RLA1 and RLA2 operation, as it isolates the DC conditions. Because the power supply has a low impedance (due to the 2000uF capacitor), all the speech signal across RLA1's coil goes into the Gold Phone.
Conversely, speech signals from the Gold Phone are coupled to the phone line in exactly the reverse manner.

At this point, the circuit can make outgoing calls. Across the incoming line is a ring detector based around an opto-coupler. The AC ring voltage causes the opto-coupler LED to pass current, via the two 22V zener diodes, the .47uF capacitor, and the 22k resistor.
When the LED illuminates, the photo transistor conducts, causing base current to flow in the BC337. This switches on the coil of RLA3, which has a 12V coil drawing about 37mA. The 560R resistor allows for 42V operation. Another LED shows the relay is being actuated, and indicates the presence of an incoming ring signal. Because of the low frequency ring signal, the relay would chatter during every half cycle, when the LED in the opto-coupler is not conducting. To prevent this, a 100uF capacitor connects across the BC337 collector-emitter circuit. The 100uF holds the relay closed whilst the transistor loses conduction on every half cycle. In series with the 100uF is a 330R resistor to limit the capacitor discharge current when the transistor conducts. A 100uF capacitor charged to 42V would be capable of damaging the transistor on its own.

When the ring signal is received, and RLA3 actuates, it switches the Gold Phone directly to the phone line, and the ring voltage feeds its ringer in the usual way. When the phone is answered and taken off hook, line looping occurs as described previously, the ring signal is stopped, and conversation can commence. If the Gold Phone is answered part way during the ring cycle, it will loop the line directly to cancel the ring. RLA3 will switch off, switching over the Gold Phone to the interface, which will then immediately loop the line. For this reason, the polarity of the interface power supply, and incoming phone line must be the same.

In summary, the Gold Phone is effectively being used as a local battery phone with a current and voltage that it's compatible with. The outside line conditions are now unimportant. Similarly, true dialling is made possible because the dialling relay provides a full loop disconnect.

Gold Phone plugs into socket at rear of interface.

Gold Phone Hints and Hacks.

Allow time for the super capacitors to charge when first connecting the phone.
Set the key to C for operation as a domestic phone. In position L, the display is blank. In position D, the display shows $0.00.
If the line disconnects in 3 seconds after lifting the handset, reverse the line polarity. White is positive; blue is negative.
If the coin container is missing, the phone is disabled by preventing the dialling and speech circuit working. No dial tone will be heard, even though the line is looped. To use the phone without the coin container, reverse the outer two wires connected to the micro switch. The connections are on the PCB to the left of the dialler, and labelled 'coin tin'.

Brown and Orange wires reversed to coin tin switch allow phone to work without the coin container.

The handset cable should run behind the locking cross member, otherwise it will be caught under the main mechanism when it's in the lowered down position. This will make reattaching the top and subsequent unlocking difficult - and that can lead to key breakage if forced!

Handset cable (orange) goes between rear PCB and cross member.

Some phones have a locally made surge arrestor module plugged in between the rear board and the mechanism.
The safe and mechanism locks (keys designated S and M respectively), can be made to operate with any key by removing the pins, plungers and springs from the locks. Of course, the phone has to be opened in the first place, requiring a mech key. The possibility of keying the mech lock to work with the safe key was investigated, but none of the pins were common.
The call charge key switch can be used without a key simply by changing the connections on the same PCB as the coin tin switch.
Although coins are not necessary, the phone will (should) register and retain them, refunding them when the handset is replaced.

Update 23/10/22.
After a recent storm, the modem had to be replaced. The new modem was eventually procured and set up. All seemed normal again, until I tried dialling from the Gold Phone. The Dialgizmo was unresponsive, yet it worked normally with all the other phones.
It was found the problem was with the interface, because a normal (300 series) phone plugged into the interface would also not dial.
From all tests, the interface was working but it was just impossible to dial from. It took quite a few hours to determine the problem. A clue was when I discovered that introducing resistance in the line in series with the interface did allow the Dialgizmo to work. That seemed strange, that increasing the line resistance allowed it to work.
It was then found that the line current from the modem was only 20mA. I can't remember what the previous modem was, but it was somewhat more than this.
Analysing this further, why would the Dialgizmo work with an increase in line current? That's because the voltage across the Dialgizmo is increased.
To cut a long story short, the interface presents a much lower resistance across the line than a normal phone. With the greater current of the old modem, enough voltage was produced to operate the Dialgizmo. Now, with less current that was not so.
Measuring the voltage made it clear. An 800 phone had about 3.6V across it, a 300 phone about 5.7V, but the interface was only 490mV. No wonder the Dialgizmo wouldn't work.
So, the line voltage needed to be jacked up. A 1.5V cell in series with the line actually worked and proved the point. Seeing that a cell would eventually discharge and/or leak, it was better to use a resistor. 100R is about right, and the line voltage is now 2V off hook.
The resistor is installed in series with the contacts of RLA2. A one watt resistor is adequate for the line current.

Surge protection has been installed at the phone output of the Dialgizmo, to protect it and the modem. A 90V gas arrestor is used along with 500mA fuse in each side of the line. The arrestor is earthed through the mains.

Coin Validation.
While it is not necessary to insert coins when the Gold Phone is used in a domestic situation, it should still register the coin and retain it. Since there are no metering pulses, the coin will be refunded when the handset is replaced.
However, it has been found with some phones that they do not always respond to coins - with the coin dropping straight through, and no value shown. It might be necessary to run the coin through, maybe 20 times, to get a response. In one instance, 20 cent coins would show up as 50 cents. The problem seems to be with the calibration of the coin validation unit. Since the sensors actually detect alloy content and coin size by inductive means, one would imagine the calibration is rather critical. There is a row of trimpots down the side of the board, under the black plastic cover. Unfortunately, as the coin validation unit was never meant to be serviced by Telecom technicians, there are no instructions on calibrating it.
The only suggestion would be to carefully (and accurately) mark the position of the trimpots original position, and then try each one in turn until a reliable response is obtained for each type of coin.
Fortunately, in a domestic situation, coin validation is not actually necessary, and really only serves as a gimmick.

Second Version of Interface.
This was built up for use with another Gold Phone. Since the parts were mostly all new and purchased for this project, details are given for those who wish to make their own. Some alterations to the circuit were required since some different parts were used. The reed relays used in the first version were of an unknown type, and it is doubtful that they would be available new in the present day. To use readily available relays required some modification to the circuit. Additionally, the Gold Phone line current was increased to improve speech performance.

New interface constructed with readily available parts.

Circuit of new interface using new parts.

RLA1: 12V SPST relay with 300R coil. (SPDT can obviously be used - and was in the actual circuit).
RLA2: 5V reed relay with 500R coil. Jaycar SY4036.
RLA3: 12V DPDT (labelled FRS6-1). Jaycar SY4061. With the 560R resistor, there is about 10V across the coil at a current of 39mA. A similar 24V relay could be used by reducing the 560R resistor.
Line isolating transformer. This was an old Arlec 45035 which is unlikely to be available. It is a standard 600 ohm telecommunications isolating transformer, and a modern type is Altronics M1000.
1.5uF capacitor should be rated at 200V at least, and must be non polarised (polyester or polypropylene). For example, Jaycar RY6951.
10uF capacitor is a 50V non-polarised electrolytic (Jaycar RY6810).
2.2uF capacitor is a 100V non-polarised electrolytic (Jaycar RY6902). A polyester or polypropylene type can be used if cost and increased physical size is unimportant.
The 560R + 0.047uF networks could probably be omitted, along with the 3.9V zener diodes. They are are legacy of the line isolating unit modified for the first interface. However, they are inexpensive components and were included. The capacitors were 100V polyester.
The 2200uF across the RLA2 coil only needs to be rated at 10V minimum since the coil receives 5V.
The 100uF, 1000uF, and 2000uF capacitors should be rated at a minimum of 50V.
The 0.47uF capacitor in the ring detect circuit was a 630V polyester. 250V would also be sufficient.
The non preferred value resistors shown can be replaced by the nearest preferred values; i.e. 10k, 68R, 220R.
Mains transformer is a Jaycar MM2007. It is rated at 30V 150mA.
LED colours chosen were red for "Off Hook", green for "Supply", and orange for "Ring". These LED's are not essential, but provide a useful diagnostic feature. For example, the Gold Phone will not ring if the interface is not powered up, even though the ringer current comes from the phone line. Therefore the user should be alerted to loss of supply. If the "Off Hook LED" is omitted, replace the 120R by a link.
Not shown is that the mains earth was connected to the socket for the Gold Phone, so the surge supressor can work correctly. Since the 40V supply is floating, the leakage problems which existed on the public exchange line do not exist.
Retail cost of all new parts comes to around $120 in 2024.

The circuit was built on Veroboard and housed in one of the ubiquitous "jiffy" or "zippy" boxes. The instrument case I used previously was no longer readily available in the same dimensions, and would have added at least $10 to the cost anyway.
While a PCB might have looked more professional, the labour to produce one at home would have doubled the cost of the whole project. As it was, I had the circuit assembled on the Veroboard in a day.

Initial testing of the 2nd version revealed a few modifications were needed. Both relays in the speech circuit were at this stage of development 5V 500R reed relays.

Some initial difficulty was had with the interface dialling wrong numbers. It was assumed that RLA1 would be the problem, since this creates the dial pulses seen by the Dialgizmo. Looking at the dialling waveform showed a very slow release time. Reed relays are supposed to be able to switch very fast, so it was at odds with what was expected. For testing, a conventional 12V relay was substituted, and this dialled perfectly. So why wouldn't the reed relay work? The relay was tested with a square wave function generator driving the coil through a diode. It easily switched up to at least about 200Hz. 10Hz from a dial should be no trouble!

First thoughts were that because the two relays were mounted close to each other (as in the above photo), the line looping relay (RLA2), was leaking enough magnetic field to the dialling relay (RLA1), to slow down the release time. Strangely enough, inserting a piece of steel between the relays to shield them from each other seemed to fix the problem. I then relocated RLA1 further away, but the problem was still there.

It was then found that the line current through the Gold Phone (about 25mA) was excessive for the relay coil, which should have only 10mA passing through it. Reducing this was achieved by connecting a 560R resistor in series with the reed relay coils. Since the line current was now inadequate for the Gold Phone, a 820R shunt resistor was connected across the relay coil circuit, giving about 35mA line current. It was felt that the 25mA used in the first interface was at the lower limit of being acceptable, and with the particular Gold Phone being used for testing (not the same one as the first), 25mA gave less than satisfactory speech performance.
With the line current now increased to 35mA, the resistor shunting the red "Off Hook" LED was reduced to 120R from the original 470R.

The hum from the power supply was felt to be higher than it should be, and was probably caused by the increase in line current. To improve this, a three stage filter for the 40V supply was implemented. Further reduction in hum was obtained by including the 33R resistor in series with the AC input of the rectifier. This reduces the input capacitor peak charging current. The supply drops to around 31V when the phone is off hook. An active filter/regulator could have been used, but it was felt the extra complexity was not justified.

The capacitor across the RLA2 coil was increased from 100uF to 220uF, since with the relay specified the hold in time was insufficient. This resulted in RLA2 responding to the dial pulses, whereas is should stay closed throughout the dialling period..
The ring relay (RLA3) holding network required the 330R be reduced to 100R, to prevent this relay responding to individual pulses of the ring voltage.

First lot of modifications were still not reliable.

Unfortunately, operation was still unreliable. It seemed the current through RLA1's coil was just too critical for dialling. A slight change in supply voltage was enough to cause erratic performance. At this point I decided to change this relay to a conventional type. This also resulted in an improvement with speech quality, because it had a higher coil inductance than the reed type used. This allows more audio voltage to be developed across its coil.
The relay chosen was a 12V SPDT type with a 300R coil. It is of enclosed construction, so it is felt that the contact reliability should be good enough. Certainly, the dialling performance is far superior. To restore the correct line current for the Gold Phone, the 560R between relay coils was removed and shorted out, and so that RLA2 is fed with correct voltage and current, a 200R resistor was connected in parallel with its coil. Line current is 32mA. Voltage across the Gold Phone off hook is 16V.
Yet again, wrong numbers were sometimes dialled. This was quickly narrowed down to the fact that RLA2 was no longer holding in long enough, and was also contributing to dial pulsing. Connecting the CRO across its contacts showed the contacts opening during dialling. The reason was that with 120R now in parallel with the coil, the time constant was much reduced. To restore this, the previously used 220uF was increased to 2200uF.

Final version working as it should. Note the dialling reed relay has been replaced by a conventional type.

RJ socket connects to Gold Phone. Grey wire to Dialgizmo.

Home