8081 - Accent
An accent feature is missing from the SoundForce 808 Snare. Adding one is a matter of replacing a 12V rail voltage with the accent voltage. And that’s where a rabbit hole entrance lies in wait.
So Common
The TR-808 has one global accent track where you program steps affecting all active voices with an increased trigger level. There is only one global level setting for the amount of accent. This accent is sent globaly on the COMMON TRIG line to all the instruments.
The common trig signal is passed through the global accent control, altering it's amplitude.
As you can see above, the accent level piggbacks the common trigger pulse and the level of the pulse that makes it's way into the voice circuits is determined by either the normal pulse level or the accent level.
This obviously means there are just two possible levels for a step. As long as the accent knob is left at one setting, of course. You can
twiddle that accent knob in real-time, but you can’t programme more than
two levels in step-time.
According to the Service Notes the normal trigger pulse is just 4v (approx) and the accent pulse ranges from 5V to the full-on 15V. That's quite a huge difference! A 10V range, in fact and that brings us to another 10V range...
So European!
In Eurorack clone modules the two-level 808 accent system translates to a gate/trigger input for the drum and another accent input with a control for the accent level. Tip Top Audio 808 modules work this way and although it's not actually passing the level of that accent pulse in to the circuit in the way the 808 does, the result is the same
The main difference to the 808 in modular is that you can control the accent per voice. There’s still only two levels per voice though. Again, you’re welcome to play the accent level knob and manually set levels, if that’s something you’re happy doing.
My modular drum rack is driven by an Arturia Beatstep Pro and a CV.OCD. The eight trigger outputs of the BSP trigger the first eight modules and then the CV.OCD handles the rest. It also sends trigger outs to several accent inputs. This is all via MIDI from the BSP and is configured to trigger those accents only when the velocity is 64 and above (halfway).
One gotcha is that some modules don’t have an accent control input. Not really. For those modules, 5V triggers are not high enough to get an actual accent. I’m using a level-shifter to boost CV.OCD triggers at 5V up to 12V for the Hexinveter AteOhAte Toms. The CV.OCD triggers are 5V only and even the CV outputs are limited to 8V so this is all I could do.
The AteOhAte Toms accent could then be said to have be a CV level. I’m just limiting it to one level and could patch in any CV signal if I chose to.
So controlling
This is a different way to look at this accent feature. Accent as a control voltage. Instead of a pulse input scaled up and down a continuously variable voltage can act as the accent level. Whenever a drum is triggered the level is set by that CV input.
This can also be handled as a velocity level by the CV.OCD, although the timing can be a problem. If the leading edge of the trigger arrives and the CV hasn’t quite got there yet then it doesn’t end up being the accent level you expect. In fact you probably get the previous step’s level and end up being one accent level behind through the sequence.
This is a problem particular to drums because the trigger pulse is so short - 1ms. As soon as that is over the bridged-T and decay circuits are off on their way and an accent to that trigger is either there from the start or is 1ms too late. Recall that the 808 handles this with a wider pulse for the instrument data and the actual timing pulse comes from the common trigger.
Excercise for the reader: Create a patch for the triggering of drum modules from a common trigger pulse ANDing the instrument trigger to guarantee perfect alignment. OR hook up a scope to the trigger outputs of the BSP or similar and measure the timing variations between triggers, then do the same for CV.OCD.
I have found workarounds for this velocity issue - adjusting the step triggers timing on the BSP - and it kind of works. To be honest with you though, it is a compromise solution which calls for some unsatisfying work to get a result.
And as I mentioned, the actual CV input might have be scaled up, if 12V is required to get to full accented levels.
The CV input works better with independent modulators, like LFOs - again with appropriate shifting and scaling. A squiggly voltage applying random, subtle variations in level can be quite effective. Notice though that this isn’t really an accent at all. It’s more that the drum hits have a human ‘feel’ be cause they aren’t all uniform. It’s more like a velocity setting.
As you can see, I have done some thinking about accents in modular! Before getting to a solution though there's a couple more things to note.
So, that's how it works!
Here's the schematic of the SF 808 showing how the accent input is tied to the 12V rail.
In my modified SF 808 I've cut tracks so that the 12V pin (1) on JP4 is now the accent input. I've then linked that back to 12V to test the rest of the module. Until now…
Let’s take a quick look at how this original 808 circuit, replicated be SoundForce, works.
The accent input is the 12V rail at R183 and the emitter of the PNP Q46. To add accent voltages I need to control the voltage here.
In fact this is more or less a VCA input for the trigger level injected at D1. Some would say that a VCA useding one CV to control another is more or less an AND funstion and if you want to logically AND signals then a VCA will do the job. I sort of agree but logic gates are open or closed not continuously variable, so be careful hwo you define this this still.
The PNP emitter is the control input here. The inverted trigger at the base switches the voltage at the emitter through to the collector. Zero volts at the emitter is thus no trigger and 12V at the emitter is a 12V trigger.
That's that cleared up. One more thing.
So much more
Replacing the fixed 12v at the accent imput with a variable resistor controlloing that voltage I noticed something about the level of accent and the oscillators.
Adjusting
the Accent input and observing the output with Snappy all the way down a
clear distorion on the first peak in the oscillator is visible once the
voltage goes above 8.3V. The peak is squared off as it settles out
around 20 Vpp. This obvioulsy alters the sound. Before that first peak
is clipped the sound of the oscillators is nice tight drum. After the
clipping there more of an overtone and the hint of the stick on the
skin.
Meanwhile
below 8.3V the snappy falls away much more quickly. You can bring it
back up with the snappy control or the decay control, but in terms of
dynamics it's less snare and more drum as you go up and vice-versa. Below around 3V there's really not much drum at all.
Now recall that the 808 goes from a 4V normal trigger all the way up to 15V for fully accented. In Euro levels equivalent that is 3.2 volts for a normal trigger. You can get things going at that level but the snappy has to be all the way up, otherwise you just get the oscillators pinging. There's a usable sound down there though and adding the full 8-10V of Euro CV would applying the same kind of boost as the 808 accent.
I tried this on the TR-6S. I set the accent on one step and off on the other. I truned the accent control all the way up. The normal step had no problem with snappy. Only turning snpappy right down cut the noise out and the accent had no reall effect on it's level. This is to be expected though, the real 808 doesn't struggle for snappy level when the accent is on normal triggers. So, why is it on the SF 808?
It's not the SF at all, it's my modification to the snappy decay circuit.With that mod bypassed the accent works just like the 808/TR-6S. OK, not great but I don't want to break that mod. The solution must incorporate this new distortion feature!
So What?
Which brings me to a fairly sophisticated implementation of accent.
To make this work well with normal Eurorack levels I'm going to need a circuit to scale triggers up, leave CV inputs unchanged or boost the CV as needed. That presents two options: amplifying or adding voltages.
I like adding because it means there's less chance of getting zero volts for unaccented triggers.
If the accent is a CV then you must have something on every trigger or it will be zeroed out and that might be difficult to manage. And, if you only want to accent on a trigger then at the accent there has to be something holding up a minmum level. You need to add it to something, in other words.
This thinking led me to the following:
No Accent Input
Accent input normalled positive supply rail = ~12V
Adder Voltage = 0V
Accent Control = 3V-12 V, Mid point = 7.5V
Sum = 0V + 3-12V = 3=12V
Note that the accent control is a voltage divder with another series resistor to set the minimum level - 3V.
Trigger Accent Input
Accent input connected to trigger input at 0 or 5V
Adder Voltage 7V
Accent control = 1.25 - 5V
Sum = 7V + 1.25 - 5V = 8.25 - 12V
CV Accent Input
Accent connected to CV input continuous CV voltage -10-+10V
Adder Voltage 7V
Accent control(0-10V input) =2.5-10V
Sum 0-10V input) = 7V + 2.5 - 10V = 9.5-12V
So
How to implement all that though?
Summing Up
To add voltage I need a non-inverting summing amplifier. I had wanted to avoid an op-amp in this design but to add means to get an average voltage first and then multiply back up by the number of things being added. And for that I really need an op-amp. For the two voltages there has to be a gain of x2 to restore the average of the two voltages to the sum of the two.
Dividing Up
With nothing plugged in at the accent socket the accent pot will divide a rail voltage into the summing amp. i.e. adding nothing to that.
(Vrail / pot) + 0V = 3-12V
e.g. 7.5V= 5/8s of 12V (0.625)
e.g 300+180 or 200+120 https://damien.douxchamps.net/elec/resdiv/
Taking Away
With the accent CV or trigger plugged in, the 12V has be removed from the pot and replaced with that Vinput, And then added to 7V such that the minimum accent level is maintained (unless CV is negative).
(Vinput / pot) + 7V = 7-12V
for +ve Vinput voltages.
The hard part is that there are two things to switch here. The 12V from the rail and the 7V into the sum. The 12V rail is switched out when the accent input is plugged in. Switching the 7V on as this happens is teh harder part. The presence of the 12V rail to disable the 7.5V and the absence of 12V, for voltages up-to 10V, to enable the 7.5 required some deep thinking.
This solution is centered on a 10V zener diode in reverse bias at the base of an NPN transitor. Voltages above 10V switch the transitior 'on' and effectively ground the collector node through the emitter. When the 12V rail is switched in, the output of this trasistor circuit athe collector node is therefore 0V. When the accent input is connected the 12V rail is switched out and up-to 10V maximum* is injected. Now the base is low voltage (or, whatever leaks through the zener) and the transistor is 'off' leaving the collector voltage at 7.5V which is now added to the accent.
And here it all is
*Eurorack levels should peak at 10V but if it is higher, then the last two volts will pass through the zener and switch the 7.5 volts off. This may actually be usful if for some reason you have a signal varying between 10V and 12V but will otherwise not be any kind of problem and can be disregarded.
It took me a while to come up with this simple solution, but I'm very satisfied with it's elegance, making two switches in one - the physical, socket normalling switch taking care of the main signal and the transistor handling adder volatge.
Comments
Post a Comment