1. Can compressors colour the sound?
Just as an ideal amplifier ("straight wire with gain") shouldn't alter the sound other than make it louder, so an ideal compressor would just act as a perfect automatic level control. However, as we know, a guitar amplifier is used creatively to alter (hopefully enhance) the sound, not just through the use of tone controls, but also adding saturation, harmonics and intermodulation products to change the whole timbre of the sound. Similarly, a compressor can be used to alter the sound by changing the threshold, attack and decay times, and driving it out of its linear operating range, though the "old school" engineers at Abbey Road would probably not have approved of the process (at least until it became more widespread practice with pop music, especially after the Beatles came along).
The circuits of compressors like the Altec 436 and Abbey Road RS124 derived from the Altec are similar in some ways to Vox guitar amplifiers in that they have no overall negative feedback that is normally used to reduce distortion. This has two effects: they can fairly easily be driven beyond their linear operating regions, and when this is done the clipping characteristic is quite "soft" and the distortion more pleasing to the ear than an amplifier that has a lot of feedback and a hard clipping characteristic that doesn't sound as nice. Furthermore, they have coupling transformers on the input and output and these can again be driven harder to produce distortion effects. Changing the dynamic response of the compressor (threshold, attack and decay) can also change the sound by emphasising or de-emphasising the initial attack of the notes and changing the sustain, for example. Finally, it is the tubes themselves that have a particular response to overload that gives vintage compressors such as these their particular quality.
So compressors like the 436 and RS124 certainly do have the potential to enhance the sound.
2. Differences between the Altec 436 and RS124.
A couple of RS 124’s at Abbey Road.
The engineers at Abbey Road modified their Altec 436’s before they were deemed to be suitable for use in the studios. While there is no definitive statement as to the nature of these modifications, there are a number of sources of information when taken together probably give a good idea of what was done.
It is generally accepted that the main EMI modifications were:
1. To make the “recovery” more flexible by adding a new front panel control with 6 settings and
including the “hold” function between each setting.
2. Replace the input level pot with a constant impedance stepped attenuator.
3. Add a stepped output attenuator control.
4. Modify the release times to cover 0.5 Sec to 1.5 Sec.
5. Later on to add a balance control to allow for tube imbalance and a test tone circuit to facilitate
Looking at some of the circuits available on the web that purport to show the EMI mods, other changes might have been modifying the time constants to also get a shorter attack time (25 mSec for the RS 124 compared to 50 mSec for the 436). Just changing the storage capacitor in the time constant circuit from 1.0 microfarad to 0.47 microfarad (as indicated on some schematics) would bring the two very close in spec. Also indicated are small grid stopper resistors on the second amplifier tube and some extra power supply smoothing to reduce any residual hum. None of these would have altered the basic character of the sound compared to the original 436’s. In fact, the extra smoothing would have reduced the headroom even further giving earlier clipping/limiting.
Another potential source of information is from the RS 124 replicas and plugins that claim to be accurately modelled on the originals. The three RS 124 plugins from Abbey Road (now discontinued) that were supposedly modelled on the actual units at Abbey Road are all slightly different and seem to have attack times of somewhere around 15-35 mS, 25-60 mS and 50 -90 mS. Also, the RS 124 reproductions advertised by Chandler also have three different attack times (as well as some others).
Finally, there is the archive of tunes recorded while the RS 124’s were in regular use at Abbey Road. Some of these, like Atlantis recorded by The Shadows, bear very characteristic compressor effects and comparison with those recorded with similar equipment gives an insight on the nature of the sound modifications.
3. Some Test Results.
A colleague Phil Coggan recently acquired an Altec 436C and demonstrated that, with other suitable equipment including a 1963 Strat and TVS3 echo unit, it can indeed accurately reproduce the recorded sound on tunes like Atlantis. In order to try to better understand how this happens we ran a number of tests on his Altec using some test signals and a range of settings. He sent me the wave files and I have been able to analyse them to see what is going on, and it turns out to be very interesting and a bit unexpected!
1000Hz test signal envelope (all transitions at zero crossing of the waveform).
1000Hz Test signal spectrum.
All the normal dynamic parameters for his Altec were pretty much as in the printed specification: attack time 50 mSec, decay times 0.3, 0.8 and 1.3 Sec. The total harmonic distortion was around 2% for signals in its linear range and it responds normally to the threshold with gain reduction from about 2:1 to 4:1 depending upon the threshold setting.
However, what the specification doesn’t mention and what is so important in getting the Atlantis sound, for example, is tube saturation. In a nutshell, before the first signal arrives the 436 amplifier has a gain of 56dB. So any signal will initially be amplified by about 1000 times and anything but the smallest signal will drive the tubes into saturation, clipping the signal very effectively (but it is a soft clipping with rounded waveform peaks, not hard clipping). This is clearly evident in the test signals right at the start, and also in the waveforms on Phil’s wet Atlantis lead track. As the delay time kicks in (50 mSec) the gain is reduced according to whatever compression setting you have and the amplifier becomes more linear again. This saturation effectively removes the high initial transient you would expect from a normal compressor with such a long attack time. As the amplifier moves from saturated to linear it goes through various stages from complete cut-off, through symmetrical extreme clipping to asymmetrical clipping and finally linear, with the operating curve of the amplifier changing continuously and this also changes the compression ratio etc! All these stages of recovery have different distortion spectra and have major impact on how the initial character of the sound develops. When used on a single instrument like a lead guitar where there are often spaces between notes that are greater than the recovery time, the whole process repeats itself again with each note.
Altec 436 envelope with extreme tube saturation on initial transient. The wiggle at the end of the initial burst is due to the asymmetrical waveform causing a shift in the dc level.
Altec 436 envelope with less tube saturation
Altec 436 envelope with initial tube saturation just at the beginning of the transient.
Altec 436 waveform with extreme tube saturation
Spectrum of Altec 436 with extreme tube saturation. The high background is due to all the numerous intermodulation products.
Altec 436 symmetrical saturation waveform.
Altec 436 spectrum of symmetrical saturation.
Altec 436 asymmetrical saturation waveform.
Altec 436 asymmetrical saturation spectrum.
Altec 436 waveform just coming out of saturation.
Altec 436 spectrum of unsaturated tail (roughly 2% THD).
It is only at quite low inputs where the 436 is starting to act as a linear compressor. However, the levels of highly transient signals like the guitar are hard to ascertain and it is really only by ear that you could set the input level to get the required sound.
All this also probably explains why the Abbey Road engineers fitted the “Hold” function to keep the amplifier gain down and avoid the initial saturation effect that would have sounded horrible on normal program material.
So I think it is a lucky happenstance that the 124/436 was used to give the large degree of compression on tracks like Atlantis. It was a critical part of achieving the sound heard on the records. A bit like the Meazzi echo units actually, which just happened to also have a circuit design that gave a particular distortion and its own style of compression that also became an essential component of “that sound”.
4. Can plug-ins do the same job?
Now that we have a better understanding of the complex behaviour of the Altec 436/RS 124 in processing transient signals like a guitar, it would seem a very tall ask for a digital model to achieve the same outcome. Nevertheless I thought it might be interesting to see how close some of the readily available plug-ins performed.
I ran the same test signals that we used for the Altec 436 through four compressor plug-ins: The Abbey Road RS 124 (supposedly an emulation of the actual units at Abbey Road, expensive but no longer available), the OverToneDSP FC70 (supposedly a Fairchild 660 emulation), the Nomad Factory LM662 (another Fairchild emulation), and the essentialFX (a generic DAW plug-in, typical of most of the freebies).
The RS 124 plug-in has roughly the same attack and decay times (in normal operating mode) as the Altec 436 but, not having any provision for tube saturation, the initial transients were way too prominent. This because the compression was not starting to take effect until well into the 50 mSec attack period, by which time much of the initial attack had passed through without much attenuation. In the “hot” mode (shorter attack time) things were a little better, but still way off the mark.
RS 124 plug- in envelope.
The FC70 has a fixed short attack time and so the initial transient is attenuated to a fair degree and with a suitable decay time the wave envelope was a bit closer to the 436. Like the first two, the essentialFX does not include any provision for saturation and so a short attack (0.2 mSec) is again necessary to catch the initial transient and with a suitable decay time the performance was much the same as the FC70.
FC70 envelope. Note the very fast attack capturing all but the first half cycle of the initial transient.
Unlike the others, the LM662 has provision to include a saturation effect (the slider marked “12AX7”) but it was nowhere as effective as the 436 tube saturation and so it was again necessary to use a short attack time (0.02mSec) and long decay time to get close to the observed waveforms of the 436. However, the added saturation did go a long way to reproducing the change in distortion components measured on the 436 during all stages of the attack and recovery. Those components were of course missing from the other three compressors as they made no attempt to model the saturation.
LM622 envelope with high degree of tube saturation. Note very fast attack capturing all but the first half cycle of the initial transient.
LM622 spectrum during peak (with high degree of tube saturation).
LM622 spectrum during lower amplitude signal (with high degree of tube saturation).About 8% THD.
It would be quite feasible to put a tube saturation module in front of a compressor module to get a similar effect to the LM662. While this is a worthwhile improvement, none of them can reproduce the very complex tube saturation effect of the 436, involving as it does a continuously changing operating characteristic that depends upon the signal content.
Listening tests confirmed the above measurements: the guitar attack was way too prominent with the RS124. The FC70 and essentialFX had better transient attack but sounded quite sterile compared to the LM662. While the LM662 was clearly the best of the bunch (and incidentally was the one we used for the Atlantis track on the jukebox on the TVS Specialty Products website), it still fell somewhat short of the detail of the 436 sound, particularly on the transients.
Actually an accurate Fairchild 660/670 model is not the ideal starting point for the 436/124 since they get over the transient problem by having a very fast attack time and generally don't have to cope with tube compression as a dominant feature. However, other than the discontinued Abbey Road RS 124 plug-in, I haven't come across any other 436/124 plugins and certainly no other generic ones that can model the tube saturation evident in the Altec/RS124.
The only way to come close to a real 436/124 would be to construct a model involving each and every stage of the entire 436 circuit, including the complex non-linear behaviours of the transformers and tubes and all their interactions. However, to operate this in real time would require considerable computing power and it would be far easier (and probably even better) just to build a real one with transformers and tubes, much the same as we did with the TVS3 echo unit that reproduces the sound of the original Meazzi’s!
TVS Specialty Products
The audio and video files available on this site are presented for the purposes of non-commercial research, private study, criticism and review.
This use is pursuant to U.S. Code Sec. 107, Fair Use provision of the US Copyright Act and the Fair Dealing provisions of the Copyright Acts in other Commonwealth countries.
© copyright TVS Specialty Products - All Rights Reserved.