Electronic organs - circuits, experiences and interests

[Lucien Nunes]

airflow turbulence presumably ensures phase incoherence between the various notes

It usually does, although it is possible for pipes to synchronise spontaneously by accident or design, both through ambient coupling and perturbation of windchest pressure. There are different physical layouts of pipes on chests that offer different compromises in coupling and resulting tendency for pipe to pull at each others' pitches vs. constructional and tuning convenience. Chromatic (pipes in ascending order) and Diatonic (two symmetrically-disposed chests with alternate pipes in each) are the most common, with others such as Augmented (adjacent pipes in minor thirds) and the compact but uncontrollably pull-y Octaves aligned.

Complex non-linear beats can arise even between two pipes, let alone a few dozen. The superposition of:
* harmonics within the pipe timbre locked to their fundamental,
* mutation ranks whose fundamentals speak nominal harmonics of the unison pipes but unlocked from them,
* inharmonic partials,
* nearby tempered intervals that beat audibly,
plus the complex spatial relationship between the pipes and the building that you refer to, result in a nightmarish challenge for the electronic constructor seeking an accurate simulation.

In the 1930s this complexity was beyond synthesis. The best one could hope for was to create approximately the correct 'composite' timbre, approximately in tune regardless of the weather, and get a reasonable amount of it coupled into the auditorium with modest harmonic distortion and intermodulation.

The next stage of evolution in the 1950s and 60s was to either attempt to disguise the locked-phase sources using acoustic FM and AM at the loudspeaker or in a spring reverberator, or to provide a financially and technically arduous multiplicity of oscillators (see my comments on that in post #36).

By the 1970s it was still preferable to generate locked-phase waveforms, but then replicate them shifted cyclically in pitch (hence reproduce them arbitrarily displaced in time) using variable electronic delays. The key to the success of the 'chorus generator' was Sangster and Teer's 1969 invention of the 'Bucket-brigade' analogue delay. Within a few years these were common in many types of instrument, archetypally the string synth, but ubiquitous in organs too.

Below is a pic of the 'presence modulator' in the Wurlitzer 950TA of 1979 introduced upthread. This is based on three Reticon SAD5120 512-stage BBDs*, ICs 3, 7 & 10. Their respective clock oscillators are ICs 4, 6 & 11, which are in turn modulated by the low-frequency oscillators ICs 5 & 12. The outputs of the two LFOs are mixed in three different amplitude ratios to provide the modulation to the clocks, and it is the correct relationship of these ratios to the three mean deviations that is critical to the overall distortion and musicality. The three outputs are not electronically re-combined; they proceed to three independent channels of amplification and speakers.

With the move away from analogue synthesis to digital and scaling of both processing power and economy of DAC channels, it has become possible to generate a closer approximation to a pipe-by-pipe simulation. Large numbers of audio channels are now commonly used (dozens - even up to 96) both to avoid intermoduation and to provide a spatial diversity of acoustic inputs to the environment. There is however an interesting analysis that suggests the extreme configurations offer little real advantage, here: here: https://www.colinpykett.org.uk/number-of...organs.htm

* As a diversion I will mention that this presence modulator is a slightly later board, not original to the organ. The original was based on the MN3001 and 3002, which turned out to be somewhat unreliable over time. They were however used in some popular guitar stop-box designs, hence remainder stocks have been snapped up and they are now effectively unavailable, or only at exorbitant prices. It is a classic situation where a much more popular but less demanding application has deprived the organ maintainer of essential parts. The original presence mod in this organ had issues with both ICs and my route of least resistance was to replace the board as the later BBD was not a direct replacement. There are some minor issues with level matching that I have not got around to resolving, but might once I get my head under the stoprail with the general sortout I promised earlier.

[Lucien Nunes]

Here's something rather high-end, the Yamaha E50 from 1978. This was near the pinnacle of 2-manual analogue home entertainment / multipurpose organs in its day, selling for over $12,000 and surpassed only by the E70. These two models used Yamaha's patented digitally-controlled analogue sound engine termed 'PASS-VCF' which generated four source waveforms (sine, square, pulse and sawtooth), filtered them e.g. by digitally-controlled VCFs, then combined them with VCAs to complete the synthesis. What is interesting about the system from a wider electro-music perspective is that Yamaha's highly respected CS-80 analogue synth, which was the firm favourite of Vangelis and competitor to such greats as the Sequential Circuits Prophet, uses the same sound engine as the E70. The key difference is that the CS-80 presents all the parametric controls explicitly to the user, whereas the E70 presents them only as presets via the conventional organ controls. The E50 only has one synth core compared to the E70's two, so it cannot generate as many different timbres simultaneously, but in other respects they are near identical.

I mentioned earlier the notorious unreliability of some organs due to connector problems, where the quality of the connectors was inadequate for the complex and extensive purpose of constructing an electronic organ. Also, that Wurlitzer and others tended to avoid them where possible, creating  tightly tethered boards festooned in wire-wrapped looms and spread liberally over all available surfaces. The E50 shows how it really should be done, in Yamaha's characteristic, sensible, Japanese style. Admittedly, with the price tag of these instruments, they did not have to scrimp on components. The digital end of things lives on about eight boards in a completely enclosed card frame with a backplane and high-quality 2-row connectors. The remaining 18 or so boards that are not sensitive to longer wiring runs, are hard-wired into the loom and slotted into a card frame. But they can be easily withdrawn and placed on top for access, as each has its own branch of the loom with sufficient slack. Thus, the number of connectors is minimised but where they are essential, they are up to the job. Only the amps and PSUs are outside the rack.

[joebog1]

What fun Lucien!!!. Replacing 500 little horrible .47 uF, 1 uF 2.2 uF etc etc capacitors. Been there, done that. Makes the old clunky Hammond very attractive.

Joe

[Lucien Nunes]

Yes, there is that. However at 44 years old this Yamaha has not needed a single capacitor replaced so far and still works 100% and sounds fantastic, so the cost and effort of a recap will be spread over half a century of use. Unlike the Wurlitzer that started this thread, the E50 has also escaped failure of any custom semiconductors, which are often the ultimate limiter of lifespan of these more modern instruments. By contrast the Wurlitzer is only still playing thanks to cannibalisation. It has had at least five custom or difficult-to-source ICs replaced so far.

The one that worries me most from the point of view of becoming irreparable through age-related deterioration is the Dereux. The rotary electrostatic generators are truly excellent with very high resolution photo-lithographed waveform traces, but they are intricate and fragile and there is some visible incipient debonding of the traces from the stators that I have not been able to investigate yet. It still plays fine, but I fear they might need to be dismantled and carefully scanned, then reassembled and played on an archive-quality audio recording to provide a permanent record of what this remarkable instrument was able to achieve. Because it's a materials stability issue, it's likely that once one starts to fail, they all will, and without a record as of now there might suddenly be no fully intact examples to use to re-construct unless some have been kept at consistently low and stable temperature which is likely to have assisted with survival.

See attached pics of the Dereux. Some aspects of the instrument are a bit basic and clunky, notably the playing surface which falls short of the fine, craftsman-built Comptons. But the tone generation, really the heart of the organ and the bit that sells it as a musical work of art, is impressive both from the methodology and the execution. The principle is very similar to the Compton, using proportional DC keying voltages to command from the generators the correct proportions of different waveforms. It does not rely on complete additive synthesis; there are complete pipe tones present as there are in the better Compton models. Also corresponding to the upper tier of Comptons there are some refinements, such as a breaking mixture, implemented in rather more streamlined ways (in this case by a diode matrix instead of a separate rank that breaks back) and a speaker with a rotating treble director (Compton and Allen both moved the drive units, Dereux moves only the diffuser.)

The generators are in a subchassis with six on each side, hanging on a spring inside the console. The cable looms run to 12 boards, one per generator, where voicing switches are operated by rods controlled via Bowden cables attached to the stopkeys. To maximise the SNR, this organ uses very high polarising voltages, 900V IIRC, so ones does not want to go poking around on those boards without one hand in the pocket.

[Lucien Nunes]

I have been quiet in this thread for a while, but the truth is that I haven't touched an electronic instrument for weeks. I did however repair a Hammond chassis for a friend, which had all sorts of issues. The organ is an E100 from the late 1960s. 79-pitch tonewheel generator, plus some percussion effects etc which are solid-state, with 2-channel valve amplification (main & reverb). Most, but not all, of the audio guts apart from the percs are on a single chassis. During initial testing and evaluation after acquiring the organ, they had identified most of the problems as being on the main chassis and had started to sort some of them out. However, we knew that some of the valves were below par and wanted to crank them through the AVO, so it was more convenient simply to bring the whole chassis down and pop it on the bench here.

The main power amp uses a pair of 7591s connected tetrode-fashion with fixed-bias, driven by a concertina phase splitter using a 7247 (ECC832.) The reverb amp is smaller and uses a pair of ECL86s with cathode bias. These are fed respectively by a GZ34 and an EZ81, and our first problem (which we were already aware of) was that the GZ34 was low. But neither of us reckoned on half of it being completely dead, so a new valve was fitted forthwith. We had tested the 7591s on the AVO Mk IV and found them to be low and a poor match, with one taking barely half the standing current of the other. This was no surprise, because one of the first tests Jamie had done on receiving the organ was to check the cathode currents and found the bias control set so far out that one valve was barely conducting. We replaced them with modern ones which were a bit hot, so we had to increase the overall bias slightly to tame the current. This was exaggerated because with the rest of the organ disconnected some of the many HT rails were rather on the high side. The bias VR is only balance, not total, so we changed one divider resistor to get the cathode current down for now.

The 7247 in the power amp section turned out to be completely dud, with a near short g1-k on the phase-splitter half. Luckily we had one new 7247, and installing it got the power amp back to normal. There are two more of this valve type in the chassis, in the reverb driver and recovery amps. One was OK, the other turned out to be not a 7247; it's a Mazda with partly rubbed-off markings but 3 minutes with the AVO confirmed it to be an ECC82. A 7247 (ECC832) is half an ECC83 and half an ECC82, so the reverb gain was slightly out. Didn't have another spare 7247 so that has been ordered. During this process, I ran into an error in the AVO valve data, even in Gerry Horrox's version with many of the known errors corrected. The grid bias setting for the '83 half is paired with the anode current and gm of the '82 half and vice versa. I expect that when I have tested them before I have used the '82 and '83 settings from memory, but this time I couldn't recall which of A1 or A2 was the high-gain half so I looked them up in the book and found the error.

The ECL86 amp was not too bad, the valves are not perfect but one does not want to be chucking out ECL86s when they are only a little bit down. The voltages around these valves look a little odd until one remembers that the combined output cathode current passes through a 22V zener to generate the regulated supply for other bits of the organ. A few rogue R's and C's later and all was good, with 35W output from the main amp at 1% distortion. It simply remained for Jamie to trim the bias once the chassis was back in situ in the Hammond with the HT rails loaded as normal.

In a fortnight's time we are going to install multiple outputs for Leslie cabinets, with Jamie tracking down various crackles and niggles in the preamp in the meantime.

[joebog1]

I dont know that the link will work, but its a good story.
https://www.abc.net.au/news/2021-05-23/g...c_news_web&utm_content=link&utm_medium=content_shared&utm_source=abc_news_web


Joe