Korg Volca Beats Snare
Analysis



The Volca Beats snare kind of sucks. Korg’s designers may have expertly and earnestly designed the snare to sound more natural than the electronic snare of the Roland TR-808 or TR-909, but many users of analog drum sounds want an unnaturally snappy and punchy snare. Several attempts to mod the snare circuit have been made. The most common mod is to add a 100nF capacitor to the empty C78 pads on the PCB. Another mod increases the noise component, sometimes using a 10K pot to allow adjustable noise level.

I added a 100nF capacitor to the empty C78 pads and it improved the quality of the snappy component, but the snare remained underwhelming. In order to make meaningful modifications to the snare circuit it would help to have a schematic. I reverse engineered the Volca Beats snare drum circuit and determined semiconductor part numbers by looking up SMD codes. The result is a schematic of the Volca Beats Snare drum circuit. It is likely there are errors, although I'm fairly confident of the fundamentals.  Resistor values were printed on the resistor packages. Capacitor values were not determined (I wasn't inclined to desolder and measure every capacitor...). Resistor and capacitor references are impressively comprehensive considering the board density, however some references are not printed on the silkscreen.

First, a little preamble: I've compared 3 different Volca Beats. They all sound a little bit different. For example, there was a noticeable difference in the maximum snare decay time between my Beats and my friend’s machine. Additionally, the minimum and maximum pitches were slightly different. Based on the SMD codes, the resistors are most likely 5% tolerance and the capacitors are likely in the 10-20% range. In the same way no two 808s sound exactly the same, it would seem the same applies to the analog sections of the Volca Beats.

For a detailed analysis of snare drum synthesis, the venerable Gordon Reid has the final word. Briefly, a synthesized snare drum consists of a 'membrane' component and a 'snappy' component. Using the TR-808 snare as an archetype, the membrane component is synthesized using a 'T-Bridge' resonator whose frequency and decay are determined by the resistor and capacitor values in the feedback network:


(T-bridge resonator from TR-808 service manual)

In the case of the TR-808, the membrane component consists of a mixture of two T-bridged networks. One resonates at about 180 Hz and the second one resonates at about 330 Hz. The 'tone' control mixes between the two resonators.

The snappy component consists of white noise whose amplitude is controlled by a simple VCA and envelope generator. The 'snappy' control attenuates the trigger into the noise envelope generator. The membrane and snappy components are summed together to produce the snare sound.

The Volca Beats snare circuit seems to follow a similar topology: a single variable pitch T-bridge resonator is combined with a white noise pulse to produce the characteristic electronic snare. The problem is that the Volca snare sounds terrible, at least relative to the TR-808 and TR-909 archetypes. Why is this?

MEMBRANE

The Volca snare consists of a single resonator with a variable pitch (IC10C) . In contrast, the TR-808 snare consists of two resonators with fixed pitches that are mixed together with the tone control.  I measured the Volca snare's membrane pitch to be variable between 250 Hz and 570 Hz. Based on the T-bridge frequency formula, the feedback capacitors are likely around 10 nF. To my ear, the Volca snare membrane frequency is too high. Again, using the TR-808 as the archetypal electronic snare, my impression is not inaccurate: the 808 snare membrane frequency varies between about 170 Hz and 330 Hz. But here is an interesting aside: the value of the feedback capacitors were changed over the course of 808 production. The original values specified in the service manual schematic produced frequencies of 250 Hz and 500 Hz, which are very close to the range of the Volca snare. However, the majority of stock 808s (at least the 4 that I measured) have a lower frequency range, The frequency of the Volca's T-bridge network seems to conform to the formula in the TR-808 service manual. Changing R183 to 10K changes the membrane frequency range to about 164 Hz to 232 Hz. To my ear, this produces a more satisfying snare.  Further changes could be made to the Volca snare to get it to better conform to the frequency range and resonance of the TR-808 snare. In fact, it might be possible to produce the same membrane sound as the TR-808 with the same frequency range and resonance by changing the two resistors and two capacitors that make up the network. However, unlike the TR-808 snare, the Volca snare has a variable decay. The decay is determined by the amount of feedback controlled by the decay pot, not entirely dissimilar to the decay circuit of the TR-808 kick drum. The snare decay becomes important later...


(increasing R183 to 10K reduces snare membrane frequency)

SNAPPY

Now onto the snappy component. The first major revelation about the Volca snare is that the white noise seems to be generated digitally. What looks like a train of random high frequency pulses from the microcontroller is filtered into white noise by IC9B. IC9B forms a second-order low pass filter, at least according the bible of active filter design techniques. Changing the frequency response of this filter will alter the character of white noise.




Replacing  R134 with a lower resistor value (eg. 1.5K) or even a 10K potentiometer will increase the gain and change the frequency response of the filter. Modifying the noise filter has been previously described. I measured an increase in the noise output from about 1.3V to 3.3V when R134 was changed to 1.5K. Other changes to the noise filter could also be done to change the character of the Volca snare's noise source. It seems the noise is also used for the hats and the toms.


(R134, replaced with 1.5K resistor)

Now, how is the envelope generated for the snappy component? This is where the Volca snare makes an interesting departure from the TR-808 snare. The TR-808 has a fixed membrane decay and an independent fixed snappy decay. The Volca snare has a variable decay, which works by varying the amount of feedback of the T-bridge resonator. As far as I can tell, the output of the membrane component is fed into the snappy VCA through C91. The membrane component of the Volca snare is used to shape the white noise that makes up the snappy component. Thus, the snappy decay is determined by the membrane decay. This seems like a clever and efficient use of the variable membrane decay to control the snappy decay without requiring independent snappy envelope circuitry. But, to many people's ears, the Volca snare's snappy component sounds fucking awful. It's 'distorted' and 'grainy' and here's why: the output of the membrane is buffered and rectified by IC9A but it isn't filtered. Here's what it looks like on a scope at the cathode of D14:


(snare snappy envelope without C78)

The envelope is chopped up by the frequency of the membrane (the first ~15 ms pulse is the trigger signal coming through D13). This amplitude chopping is likely the cause of the ‘graininess’ and 'distortion' in the snare snappy that many users (myself included) complain about, which is especially prevalent with longer decay settings. The noise is being chopped up at 250 - 560 Hz, dependent on the snare pitch. C78, the legendary missing capacitor on production Volcas, has the job of partially filtering out the membrane frequency:


(snare snappy envelope after 100 nF C78 installed)

A 100nF capacitor soldered to the C78 pads doesn't completely get rid of the membrane frequency component, but at no point does the envelope return to 0V until the end of the decay. The result is a much more pleasing sounding noise that is less 'grainy'. A smooth contour can be had by increasing C78 to 470 nF, however this produces a noticeable snappy decay when the snare decay is 0 and just the gate pulse is triggering the snappy VCA. There isn't much of an audible difference: 100 nF seems to be the 'sweet spot' for filtering out the membrane frequency without adding an excessively long tail to the gate pulse.  Alternatively, a 47 nF capacitor can be soldered across the 100K resistor in the feedback network of IC9A to produce a nicely shaped contour with almost none of the frequency component apparent from the membrane. This does result in a slightly longer minimum decay, though.

Why Korg left C78 off the production Volcas is up for debate. To many users ears, it improves the quality of the snappy noise and it was clearly designed for this purpose. It does produce in a slightly longer minimum snappy decay.

DECAY

The amount of decay available on my Volca's snare drum is satisfactory, but two other Volcas I looked at had noticeably shorter maximum decay times. The adjustable feedback that provides decay control is handled by IC10D. Making changes here (eg. increasing gain of feedback amplifier by increasing R163) will increase the maximum decay time.

CONCLUSION

To my ears, the Volca snare suffers from two problems:

1. The membrane frequency range is too high.

2. The snappy noise is of low quality, suffering from an unpleasant 'graininess' or 'distortion'.

Both of these issues can be remedied by just two resistor changes (R134 for amplitude/frequency of noise and  R183 for membrane frequency) and addition of C78 (100 nF). The result is not quite as good as the taught and snappy TR-808 snare, but is a vast improvement over what the Volca offers out of the box.

It helps to have a hot air rework station to do these mods.