Personal web pages of

Tim Stinchcombe

CAVE! Isn't it just typical—the day after I upload this I discover that what I thought about the power dissipation/REN of these circuits seems to be upside down from what I was expecting! Initial measurements of the power dissipation of a few of these boards suggests that those with smaller series resistors actualy have lower RENs, and not the other way arround as my comments below suggest. Hopefully when I've conducted a load more measurements I shall be able to correct my statements, based on a greater understanding of what actually is the case. [Updated 21 Sep 2026]

Trimphone Tone Ringer boards and variants

My interest in telephone electronics was first piqued a few years ago: quite by chance whilst trying to get to grips with LC tank oscillator circuits I came across the original analogue DTMF 'Touch Tone' circuit, which is rather elegant, especially given the low component count, and that it oscillates at two different frequencies at once! (Here is a basic schematic.) Then more recently, due to having my landline moved over to VoIP, I retired an old 80s phone, and on pulling it apart I was intrigued to discover that the ringer IC in it (an H31002P) actually contains two oscillators, the slower of which modulates the other to switch between two higher frequencies.

This immediately raised the thought in my head as to how the 'ringing' was done in original Trimphones, the first phone to not use bell ringers: since the sound the IC makes mimics the distinctive warbling/chirping of the Trimphone, but that back then when the Trimphone was released it would have been unlikely to have contained any integrated circuits at all, how was it actually implemented in discrete circuitry? Finding the answer was much harder than I thought it would be: given that they were first released around 60 years ago, this seemed plenty of time for the official documents to have been appropriated from somewhere, so I was hoping I would simply find the 'Definitive Page of the Trimphone Ringer' on the web, containing a schematic and stuff. But alas if such a page does exist, I have yet to come across it, and worse still, whilst it was known that there were several versions of the tone ringer board in existence—in particular the 5A and 8A—the GPO/BT only seem to have released the schematics for the 5A, and so the circuit of the 8A, likely the most populous variant in existence, was not readily to be found.

Thus I started acquiring not only complete telephones, but also individual tone ringer boards, so that I could work out the circuits for myself. It quickly became apparent that there are more variants to be found than my initial searching led me to believe—this page attempts to catalogue those variants discovered so far.

There are broadly two areas in which the circuits differ. The first is how the DC biasing voltage is derived, using either ordinary diodes (in a chain), zener diodes and/or full-blown bridge rectification. The second is how the crescendo volume control is implemented, with the volume starting at a quiet (SOFT) setting, and then increasing in volume until it becomes LOUD (this is denoted by the symbol on the volume dial, between the SOFT and LOUD settings). In some circuits thermistors are used for this function; in others the charging of a capacitor in a simple RC network is used. In order to be able to distinguish between these different variants I have adopted qualifiers to go with the variant number, namely -Therm for thermistor types, and -RCnet for those using the RC network. Thus far I have identified four distinct variants all carrying the same '8A' designation: judging by the large list of manufacturers used by the GPO to make their equipment, it doesn't look like they were much bothered about enforcing strict configuration control of PCBs etc.

The oscillator part of the circuit on each of the boards is essentially the same across the variants, and is fairly basic: a tuned LC tank in the collector of an NPN transistor, consisting of the parallel combination of a capacitor, the primary winding of the transformer, and the sounder itself, which is also inductive (i.e. basically a small speaker). The transformer couples this through to the secondary winding on the base, and this produces the positive feedback necessary for oscillation (the oscillation frequency is around 2.4kHz). I was a little disappointed when I realised that the oscillator simply turns off and on during alternate cycles of the approx. 25Hz of the ringing voltage, i.e. unlike the chip mentioned at the top of the page it is only running at a single frequency, as opposed to the modulation seen from the IC.

The following sections itemise/detail each of the known tone ringer board variants.

The order of the evolution of these circuits seems quite hard to fathom out. Due to the selenium bridge rectifier in the 8A-Therm circuit seeming a little 'old hat', it is tempting to think that this may have come first, and then that the use of a thermistor was dropped in favour of the RC net variants, with their order (possibly) being 8A-RCnet-00, -560 and -3k6, caused by the apparent desire to lower the circuit's REN (Ringer Equivalence Number), by steadily increasing the value of the high-wattage resistor in series with the main part of the circuit. However that idea is negated by the fact that the 8A-Therm already had a 220Ω series resistor in it! I doubt the precise order will ever be known, or the more specific reasons for so many versions, and indeed, it is quite possible that there are other variants still out there that I have not come across.

8A-Therm [Reference example: manufacturer GEN, date code 80, host phone 786; Identifying characteristics: 2x BJT, 1x thermistor, bridge rectifier]
DC biasing voltage is generated by a combination of a selenium full bridge rectifier and zener diodes. Transistor TR2 just turns the whole oscillator, which is built around TR1, on and off per half-cycle of the 25Hz ringing voltage. In the 'off' volume switch setting the whole oscillator circuit is shorted, and so remains off; in 'soft', resistor R1 restricts the current into the oscillator; in 'loud', R1 is shorted, so as to maximise the current into the oscillator; in 'crescendo' thermistor RY1 is in parallel with R1, so the total resistance decreases the longer the ringing signal is applied, thus allowing more and more current into the oscillator to allow the volume to build up. Simulation, backed up by direct measurement on an oscilloscope, shows that at the switch-over from the forward half-cycle to the reverse one, transistor TR2 is actually turning on backwards, i.e. with the functions of its collector and emitter swapped over. I have no idea what this might do to the reliability of the circuit, but it does make you wonder if this could have been part of the reason for re-designing the circuit?

8A-RCnet-560 [Ref eg: manu STC, DC 79, host unknown; Id chars: 2x BJT, many diodes]
DC biasing is now just left to a couple of zener diodes, which only have effect during the forward half-cycle of the 25Hz ringing voltage. TR1 controls the volume by letting more or less current through to the oscillator based around TR2. In the SOFT volume setting current through C1 (in the forward half-cycle) allows C2 to charge, holding TR1 base at a low voltage, but allowing some current through to the oscillator, so it sounds at a low volume. The LC resonant tank action, of C5 coupled with the transformer winding in the collector of TR2, means the voltage there can swing through quite a large range, certainly well above the 15V rail generated by zener D2 (which is needed to bias TR1). In the LOUD and crescendo volume settings, this voltage is fed back through the volume switch, then either of R5 or R6, and hence into C2: in LOUD the lower value of R5 means C2 quickly reaches a significant voltage to allow TR1 to switch on hard, and so we get increased swing of the oscillator and lots of volume; in crescendo, the larger value of R6 means it takes a while for the voltage on C2 to build up, which thus takes place over several ringing cycles, with TR1 turning on harder and harder as time progresses.

Also note that the large series resistor R10 is now 560Ω when compared to the 220Ω R9 of variant 8A-Therm above, perhaps indicating the start of the drive to lower the load of the ringer (that is its REN) on the line.

8A-RCnet-3k6 [Ref eg: manu EET, DC 81, host unknown; Id chars: 2x BJT, many diodes]
This variant is nearly identical to the 8A-RCnet-560 above: the only difference is that the large 560Ω resistor R10 has now increased to 3.6kΩ, and has moved from before the main circuitry to after it—I'm supposing that this change was solely to lower the REN of the circuit. (The -3k6 schematic shares that of the -560 above, but with the changes for R10 as noted on it.)

8A-RCnet-00 [Ref eg: manu TCH; DC 71; host not clear, perhaps a 2/722; Id chars: 2x BJT, many diodes]
This is clearly a fourth distinct variant bearing the designation '8A', and which clearly doesn't physically match any of the three above. The photo comes from Sam Hallas' useful website, which includes a page where he does a teardown of a Trimphone (it is a slightly cropped copy of 'Image 9' on that page). The description there identifies it as using the 'RC net' method of crescendo volume control, and as all the resistor values apparently match those of the -RCnet-560 and -RCnet-3k6 above, it is a fair bet the actual circuit is the same as the other two. However it is also apparently missing the larger-wattage 560 or 3k6 resistor in series with the whole circuit which is present in the other two (and indeed the 220Ω, R9, of the 8A-Therm circuit), and it doesn't look like any such resistor is to be found on the main board—thus I have assigned it the designation '8A-RCnet-00'. It is tempting to believe that this may be an early incarnation of the 'RC net' crescendo control method, but this seems at odds with the fact that its date code pre-dates the 8A-Therm variant I have by nearly a decade; the apparent lack of the necessity of including a large series resistor is also puzzling.

5A [Ref eg: manu TCH, DC 68, host 722; Id chars: 1x BJT, 2x thermistor]
This variant is apparently the only tone ringer for which an official schematic exists, drawing N692. The drawing is disappointing in that the outline for the rectifier device 'MR' doesn't actually tie-in with that fitted on my example, an '8XBL' device (and I have no idea whether it might do so for the alternative devices specified, but if it does the physical layout of the board must differ). Accordingly the schematic alongside the photo is an amended version of N692 which matches my board (which also in fact does not have capacitor C1). I have been unable to find any useful information on the 8XBL, but the implied long chain of diodes, when forward biased, provides enough voltage—measured at around 12V—for the oscillator to run.

2A [Ref eg: manu TCE, DC 66, host 712; Id chars: 1x BJT, thermistor (piggybacked on a capacitor?)] (The photo has been culled from an eBay auction.)
This appears to be how the circuit started out, as used during the field trials of the Trimphone. The board is notably sparser than the later ones shown above: there is just a single transistor, and as there appears to be a thermistor, it is a fair bet this is being used for the crescendo volume control. One assumes the black plastic-encased bar across the top of the board is the transformer, and the black cylindrical device at the bottom looks to be a 'CSD11' rectifier, perhaps similar to the alternative devices listed for the 5A variant schematic (document N692 linked above).

There are a few other variants of tone ringer boards which are known to exist, but which appear in phones other than the Trimphone, namely in the Loudspeaking Telephone, generally designated as the 'LST4' (note that several different models exist), which was a large desk unit allowing multiple people sitting around the desk to participate in the call:

9A A reverse-engineered schematic of this variant, made available by Tony Duell in this post in the telephony section of the Vintage Radio forum, shows it to be identical to the 8A-Therm variant above.

3A Another board, again reverse-engineered by Tony Duell in the same post as above, shows it to be very similar to the 5A above, but several component value differences can be seen, and the long chain of diodes providing the forward biasing voltage is not identified (an easy oversight to make, as the voltage drop across them will not be revealed by simply using the diode checker on an ordinary DVM). (And clearly the version traced has capacitor C1 present, unlike the 5A variant in my possession.)

6A I came across this whilst searching for 'tone ringer' in general. Model 4D (drawing N4475) of the LST looks like it had an option for 'vibrator ringing': Note 7 on page 1 mentions 'vibrator or transistor ringing' (my emphasis); Notes B & C on page 2 again mention 6A in connection with 'vibrator ringing'; final mention in Note D on page 8.

Ringing voltage circuit: To produce a decent ringing voltage so as to be able to make the ringer boards sound, or to ring a complete phone, I produced this simple little circuit based around a KS0835F SLIC module (apparently a cheap clone of the Silvertel AG1171), driven by a couple of chips: a 556 (dual 555 timers) and a 4013 (dual D-type flip-flops). The fixed resistors as shown gave me timings good enough for my purposes (400ms on, 200ms off, 400ms on, 2s off; repeat), but it would be easy enough to replace the key timing resistors with a resistor and/or a trimpot if greater timing accuracy were required. (It runs on 5V, e.g. a USB supply, and produces ±80V for the ringing; all diodes are 1N4148; a slight drawback of using the 555 is the slightly longer first period: it is just noticeable that the first '2s off' period is slightly longer than those that follow.)

Finally, the luminous tube illuminating the dial in the 722 phone I bought amazingly still has just enough oomph to glow a gentle green colour, even after 58 years (though it did take a 30 second exposure to actually capture the image!):

[Page last updated: 21 Sep 2026]
[Warning about REN comments added]