Coquelet technology was well documented in the designers' house magazine, ACEC REVUE No 3-4, 1970, in the article "Les Téléimprimeurs, Téléchiffreurs et Transcodeurs ACEC - Système Coquelet". This important paper includes an excellent analysis of why it is advantageous to limit the number of symbols per character to one or two.

"Predicted-wave Radio teleprinter", Electronics 1954 No 27 is apparently another paper on Coquelet by M.L. Doelz. If you have a copy of this article, please let ZL1BPU know!

Coquelet was conceived as an electromechanical system, using tuned reeds to generate the tones, and the same reeds as receiver filters. Because the tones were in two sets, low and high, one for each tone of the pair, sync could be recovered directly. Coquelet was designed to operate directly with existing ITA2 teleprinter equipment, so ran at the equivalent speed. The original system was asynchronous (i.e. stop-start; there could be gaps between tone pairs with hand sent text), and the equipment converted the signals directly from and to conventional teleprinter asynchronous signals. ACEC also manufactured ITA2 to Coquelet two-way converters.

The first tests were at Crowborough in Sussex, where an underground high-powered radio station had been built in the Ashdown Forest. (A photograph of the tests is extant). This station was built to send propaganda to Germany during WWII. Two way communications was first established with Delhi in 1962, with Beirut in 1964 and then with Singapore in 1965, using the 32 tone system, one tone per letter of the alphabet! (Actually 33 tones, as an idle tone was used to maintain synchronism). The system proved extremely successful, and communications between Britain and Singapore often continued for an hour after standard teleprinter links had faded out. To quote from David White G3ZPA, who worked with the 33 tone system at the time:

"I was allocated to take part in the Singapore tests from our London station. It was amazing, it was so superior to conventional radio teleprinter systems as almost to be beyond belief, and was quickly used for regular traffic. When we opened the circuit at 6 A.M. every morning, the Piccolo would give solid copy immediately, but the RTTY was so poor it was not till 7 A.M. before it was usable for traffic. It was the same in the evening, because after 6.30 P.M. the RTTY would fade out to become unusable, but the Piccolo still kept giving solid copy for another hour. The amazing thing was that we could no longer hear the signal on the receivers, but the tone oscillator detection system could still produce an S9 signal so long as it could detect it even though the signal was under the noise level.

"It was so good that we quickly adapted all 64 radio stations around the world to change over from RTTY and Morse until by 1967 Piccolo had taken over. It handled all government diplomatic traffic for the next 22 years, until it was decided [to change to] an even narrower band system, that was even faster than the 100 baud 33 tone Piccolo, which was expensive and needed dedicated Piccolo technicians to maintain it. Existing radio teleprinter engineers had no knowledge how to maintain it".

Piccolo Mk 1 to Mk 5 were all 33 tone systems with no error correction. The system was so robust that none was needed, and tests in 1968 comparing Piccolo with the new experimental SITOR system (known to Amateurs as AMTOR) showed that when copy became difficult the SITOR link slowed down or stopped, while the Piccolo link continued at full speed with just the occasional error. Since the system was generally used with five character cypher groups, errors were easy to spot. Later systems were designed to be adapted for error correction, but it was never needed!

Initial transmissions were on AM, soon to be followed by SSB. The 6 and 12 tone systems were introduced in the late 1970s to further save bandwidth. The 6 tone system is still in limited use today. The most recent equipment was made by Racal. Like Coquelet, Piccolo is a tone sequential-tone system, using six or twelve tones, however Piccolo was from the outset a completely electronic system, with the exception of the symbol clock, which was at first motor driven!

The final Piccolo transmission took place on the 4th July 1993, and the whole of the U.K. Government's radio transmission system on high frequency was then scrapped. One remaining original 6-tone Piccolo system remains in working condition at the Bletchley House museum, lovingly restored by David G3ZPA, who kindly provided much of the historical information about Piccolo.

The early 32 tone version is well documented in the paper "Multi-tone signalling system employing quenched resonators for use on noisy radio- teleprinter circuits, published by J.D. Ralphs et al, Proc. IEE Vol 110 No. 9, September 1963.

This paper describes the use of quenched lossless LC filters for reception and tone generation, and has an interesting discussion of how to achieve orthogonal signalling in a non-coherent FSK environment. The discussion is in the time domain, and nicely parallels modern frequency domain arguments about tone spacing, which reach the same conclusions in different language.

The early Piccolo system related each of the ITA-2 characters to a single tone, so operating at 10 baud achieved a remarkable 100 WPM. Like Coquelet, the later Piccolo systems are two-tone techniques, i.e. transmitting sequential tone pairs. Synchronism is achieved in all systems by 10% AM modulating the transmitter with a character rate square wave.

Unlike Coquelet, Piccolo uses the minimum tone spacing for orthogonal detection (spacing = baud rate) and has always used electronic tone generation. The integrate and dump detection system allowed this level of performance, and the receiver filters were also used as triggered oscillators to generate moderately phase synchronous tones. Many versions of Piccolo exist - the most well known versions can be summarised as follows: