Modules - 4ATTMIX

yeah, Ive noticed similar things to this before, when I did some explorations on how CV was being treated by various modules.
(Ive got some notes somewhere, which I cannot find atm) 

anyway, so I took a look at what's happening on MY 4ATT/MIX according to my oscilloscope... 
(I used the 2AATT with fixed voltages as a kind of reference) 

it'd be interesting to get some feedback from robertlanger on this... 

so underlying I think the 4ATT/MIX acts as 4 attenuators into a mixer,
however, I think the characteristics of the mixer, are something like a unity mixer, that is averages voltages, 
but its a bit different perhaps biasing around 2.5v


so I did a whole bunch of tests ( I wont bore you with the details as they are really clear) , but there are some observations

a) there is some kind of averaging done, which depends on how many of the inputs you mark as 'mixed'
(and beware, if you have something marked as mixed, and the input floating this will add if you have the attenuation up.)
b) it clips at around 4.4v
c) it appears, that it seems be be using 2v-2.5v as a zero point, 
e.g. if I send it in1/in2 at 2v, the in3/4 at 0v, then I get a mix out of 2.5v

---

ok, so here is my assumption (strawman for robert to burn down in flames ) 

the DC switch is used for the attenuators, and this is so they attenuate around 0v , rather than 2.5v which is required for audio. (this is an area discussed before with the old VCAs) 

the mixer however is designed more for audio signals, which need to centre around 2.5v, and also a unity mixer behaviour is preferred.
i.e. you don't really want clipping, what you prefer to do is to be adjust gain relative to one another
... if it simply summed, you would have to seriously attenuate every audio signal as it came in... since even 2 signal would clip.

(Im assuming the mix switches, are helping to decide what the offset is, as you can only average if you know the total possible gain?) 


in fairness, this entire area is a bit tricky for unipolar system (ae etc) , compared to bipolar (eurorack) ,
because with bipolar system, audio and cv are kind of identical, they all are based around zero.
but in a unipolar system, cv is 0v based (*), and audio is based around 2.5v (or 2v)  



this highlights an area for future modules which I'd previously raised.
really what we probably need is a few 'CV tools' for AE module, if we want to deal with CV voltages precisely.

a buffer multiple.
if you take a single output to multiple destinations, the voltage will drop, this is not an issue for modulation, but is a big problem when you are using for pitch.

a precision adder.
this has the behaviour I think you wanted from the mixer,  these sum voltages without loss of an voltage, nor averaging.

a precision subtractor?
this is a bit odd, unless the precision adder assume bipolar cv (e.g. 2.5v = 0) , then a subtractor is possibly needed.
... but perhaps the precision added, could have a DC switch, on its inputs... given the 2ATT with its INV output already, has this notion of < 2.5 = negative offset



in summary, Ive tended to find dealing with multiple CVs usually a little confusing in AE, modules have a tendency not to react as i expect.
I usually end up breaking out a voltmeter/oscilloscope to see whats going on.

I think largely this goes back to the beginning, where as Robert has pointed out, originally he was designing an 'abused' noise machine, 
so precise pitch tracking (v/oct) was not a goal, and by implication i expect combining CV voltage precisely (for pitch) was also not a requirement.
it was more about 'doing it by ear', and using what sounds good  

but things have moved on, some are using for melodic use, and robert has introduced much better pitch tracking on oscillators, and things like the env/amp have been improved DC behaviour.
so perhaps things like buffered mults and precision adders  whilst not the most 'exciting' of modules, could be the next step in this area?

thoughts?



 

---

(*) interesting , I guess a 0..5v system , could also notionally decide 2.5v is also  zero point, but I'd assume that make every module more complex! (=more expensive) 
... whats interesting is the 'inverted' out of 2ATT assumes this! 

Both your reply, thetechnobear , and the note from robertlanger saying that he will look at this, were really helpful.

For me, I accept that the 1V/Octave was an addition to an exceptionally useful modular (as do you, and others). You are also right that I am likely pushing the modular into harder territory, so it's useful to have some boundaries. For me, it was initially a question of , "Should it do this...?", and if the answer is "No", at least I could stop scratching my head   I did even think about getting a voltmeter out myself to see just what the effect was - admittedly, my ears did a pretty good job in the end.

My apologies, I wasn't implying that you were implying that 

I agree, it's really a case of using the modules within their design parameters and understanding those removes misunderstandings when things don't quite work as you expect. I also agree that as the modules increase, some of them will be these more "bread and butter" ones that don't appear glamorous, but help with some of these audio experiments that we all perform.

To be clear from my perspective, my intention is to dig in and work out how to do what I want within any design constraints, that's the fun of exploring for me. Knowing that you're asking for something that needs a further module is helpful.

I did do some different experimentation with the AE Modular when I couldn't quite master the 4ATTMIX, with the thought that answers/advice will arrive soon... and it has