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.
From another site, the service and installation manual, and the operating instructions can also be downloaded.

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.

I can only put this down to something coming from the modem (dial tone perhaps?) that has caused the Gold Phone to think it should be charging for calls. If there is no earth connection in this situation the call will drop out as an anti-theft measure. I haven't fully investigated what exactly was causing the phone to be unhappy with the modem signal, but with the key in the D or L  position, it held the line. Again, this makes me think the phone is looking for something which allows it to charge for calls.
Perhaps even more interesting is that with the Dialgizmo out of the circuit, it would also hold the line in the C position. Perhaps the Dialgizmo was introducing something, but I can't imagine what, at this point.

As for the inability to dial out, again I'm not fully clear as to why this is so. My guess is that the capacitor charge current is causing the Dialgizmo to think the loop has not been disconnected during the dialling pulses. It is not to do with it being an electronic dialler, as I confirmed when I tried a Gondola phone - another decadic dialling phone with push button dialler.

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. 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.

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.

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.