Ah ha. This I understand.
This brings up some interesting questions about the ability of innovative SW implementations to address some of these HW-specific design tradoffs. e.g. If you are not emulating a phsycal circuit then feedback should not be involved. It seems there should/could be some killer algos out there...one day.
In fact, the irritating component variation aspect means that things like stereo tracking can be improved, but for other devices like the LA-2a,component aging means that the unit ends up behaving very differently with age.
The T4b module for example, has an enormous influence on how the compressor actually COMPRESSES, and the chemistry of the cadmium sulfide optical cells means that with age, the time constants (whcih are determined SOLELY by the cells... there are no external controls to affect attack and release etc) therefore modify significantly with age. -In fact some T4 modules actually 'overshoot' slightly, and 'nip' into the envelope slightly, punching a small dip in...
My good friend Steve Sykes told me a good few years ago that when you go into a studio with several LA-2a's in the rack, you should listen to find out WHICH of them you like the behavior of. -My contention is that -aside from poor maintenance and things like aging capacitors messing up the signal- the largest effect, and certainly the biggest factor acting upon the actual compression characteristic (as distinct from the output tone, distortion and other characteristics unconnected with actual compression) is the T4 module.
In this sense, the 'modeled' virtual units may not be modeling a characteristic which one actually LIKES... for certain applications. -As part of a recent analysis and investigation into what goes into these things, I've built a box which switches between multiple T4 modules instantly, and allows side-by-side comparison and analysis of the T4 module alone, without necessarily having to have multiple LA-2a's... And even though I DO happen to have a few of them, I'm interested in identifying and observing what characteristics are SOLELY due to T4 modules, and cannot be ascribed to unintentional or unnoticed variations in the 'host' unit.
It seems that people often have a "this one works great on Bass, but this one rules for female vocal" type of experience with owning several of the original units, (as opposed to the reissues, which -being newer- don't tend to have acquired the aging variations) which I believe can -most likely- entirely be ascribed to variations in the T4 module. -unfortunately, the modeling just gives us a single 'unified model' option, and -while these can often be pretty good- they remove the 'individual character variations. This lack of control option seems to be 'necessary' though, because introducing time constant control options on an LA-2a for example, would make the control interface (and therefore the user experience) too dissimilar. -And that's not what will sell to the vast majority of home studio buyers who can't readily own or use the real thing.
Now, let me -in the interest of all fairness- declare right here and now that a driving force behind my present interest in T4 analysis is because I am interested in manufacturing and selling a T4 type module... possibly with some form of switchable characteristic, if that proves possible.
With this in mind and if anyone wants to learn further about these aspects, I'd be happy to start a separate thread about LA-2a's... I've built a few of those in my time also, and they too have grown to become a 'classic' over the years; just as the 1176 has.