Tuesday, February 5, 2013

SSL Double Deka -- full review

Geez, only eight posts last year?  I've got to do better than that.  I'll kick off 2013 with a full review of the SSL Model 1130 Double Deka VCO that I previewed last year.  Now that I've finally had some time to experiment with it, I've got some audio clips to share.  And a photo of the thing, finally powered up:

First of all: I mis-stated and grossly over-complicated the theory of operation in my preview article.  Here's what the Double Deka basically does: Use a bank of sliders to draw a waveform, and the Double Deka reproduces that waveform, more or less.  That's it.  How it actually does it is rather complex -- there's a VCO that runs at ultrasonic frequencies, and it steps from one slider to the next on each cycle.  By doing so, it re-creates a waveform that looks on the scope like the shape that you created with the sliders.  

Now, some background: The Double Deka is a design created by renown module designer Ian Fritz.  The 1130 is a further development of the basic design, which adds some new features and also re-formats the module into the Dotcom module format.  (A photo of the older version can be seen on the Bridechamber web site here.)  My 1130 is from SSL's second production run, from late 2012; this run fixes a circuit bug that the original run had which prevented the sync input from working properly.  The 1130 mounts in a standard Dotcom-format case and has the standard Dotcom 6-pin MTA-100 power connector.  It's a large module, 4U wide. 

Creating Waveforms with the Double Deka

The first thing everyone notices about the 1130 is the cool lighted sliders.  The two banks of green sliders generate two waveforms, each driven by the master VCO.  The orange sliders select which octave each bank operates in, using a deviously clever method that will be explained later.  You create a waveform by "drawing" it with the sliders.  As a simple example, here is a sawtooth wave set up in the A bank of sliders (the B bank is not used in this example):

And here is the resulting waveform, as captured using the oscilloscope tool in MOTU CueMix:


 Similarly, here are the sliders set for a narrow pulse wave:

And here is the resulting waveform:

You'll notice that there is some sagging present in the waveform.  The 1130 has a switch that can be used to select AC or DC output; when the switch is in the AC position, DC is filtered out.  As it happens, the tops and bottoms of a square or pulse waveform look like DC to the filter; hence the tendency for the waveform to drift towards the axis when it should, in theory, be perfectly horizontal.  I did try capturing some waveforms with the switch set to DC, and I still saw sagging, although not as much.  However, I suspect that my recording chain (MOTM-890, Mackie mixer, MOTU 828 Mk III Hybrid) is not DC-continuous, so that could account for that.  I have not tried putting an actual scope on the output -- I might do that next.

Here's a triangle wave, with the sliders:

And the resulting waveform:

Notice that in this one, and the sawtooth wave from above, the stair-step nature of the waveforms.  Obviously, ten sliders is nowhere near enough to do a high-resolution waveform.  The Double Deka isn't intended to be a wavetable oscillator; it's intended to be a device that allows experimenting with waveforms and creating a variety of waveforms, from a somewhat restricted set, on the fly.  The stair-stepping does have some implications for the sound, which I'll talk about in a bit. 

And finally, we have a waveform resulting from some random slider settings.  I didn't photograph the panel for this one, but you get the idea from the resulting waveform:


So what do these waveforms sound like?  Here is an audio clip, containing five waveforms in this order: square, sawtooth, triangle, pulse, and random slider settings.  I didn't try to record a sine wave because I couldn't set the sliders precisely enough to come up with anything that sounded remotely like a sine.

You'll notice that there's a certain tonal flavor that is common to all of the waveforms.  That, I think, is due to the stair-stepping between slider values; it tends to impose an odd-harmonic structure on all of the waveforms that are output.  I was able to get rid of that tone by running the output through a lag processor.  Linear slew limiting might be even better, but it's hard to do in analog circuitry; I'll have to play with coding it up in Csound and running the 1130's output through it.  I thought at first that a built-in lag processor, or lowpass filter that could be offset according to the frequency, might have been a useful addition.  Then again, there's already a ton of circuitry crammed into this module, and not much room for any more jacks or controls.  And there is such a huge variety of lowpass filters available on the market that I realized it would not make sense to incorporate one into the DoubleDeka itself.  Instead, leaving that choice to the performer makes sense.

I've found that there seems to be a bit of the unexpected in how the sliders interact with the processor.  I'm not sure what it actually is.  The first thing I noticed: if you set all ten sliders to the zero value, you would expect to get silence.  Instead, there is a low-level buzzing noise.  I wanted to see what this looked like on the scope display, so I cranked the gain way up at my mixer, and got this:

It's a digital-type noise that seems to be impossible to totally get rid of.  Where does that come from?  Is it something to do with the calibration of the sliders?  I don't know.  Another thing I noticed, possibly related to this, is that all of the sliders don't seem to behave in the exact same way.  For example, on the A bank, if you set all the sliders to zero and then set the first (leftmost) slider in the bank all the way up, you get what you'd expect -- the sound of a narrow pulse wave.  However, if you zero all the sliders and then turn the second slider way up, there's something of a difference: a prominent third harmonic that doesn't seem to show up on the scope, but you can hear it.  Other sliders in the bank did one or the other to varying degrees.  In the B bank, the same thing happens, but on different sliders.  

Exploring the Panel

Anyway, let's move on to explore some of the other features of the 1130.   And there are a lot of features here.  Getting past the obvious: At the upper left are knobs for coarse and fine tune.  These control the base frequency of the VCO and hence the frequency, modulo octave settings, of both banks of sliders.  (It would have been cool if there were a way to offset the frequency of the B bank from the A bank, but that would have required  a second VCO.)  If you move the coarse tune knob slowly, it steps audibly; I'd say about 1/8 tone (25 cents) per step. This is the price to be paid for having a coarse tune control with a very wide range.  The range of the fine tune control is about eight half steps. 

The two slider banks each have an individual output, and then there is also an output for the mix of the two.  There is a knob for controlling the mix, and also a control voltage (CV) input.  With the knob full counterclockwise and no CV connected, the output is A only; at full clockwise on the knob the output is B only; in between the two are mixed.  (Minor quibble: This knob should have been indexed on the panel with values ranging from -5 to +5, or some such, rather than 0-10 which is a bit confusing for a cross-mix control.)  A positive voltage into the CV jack moves the mix more towards the B bank.  This is one of the features added in this new version of the Double Deka that was not present in Ian Fritz's original design.

Here is a sound clip that illustrates using the mix CV to vary the mix between the A and B slider banks:

Octave Selection and Control

Each bank has a slider that allows selection of one of six octaves.  Twiddling the octave sliders, one notices right away that they are not slide switches but linear potentiometers.  There is a reason for that: each bank also has a CV input for octave selection.  The CV is added to the slider setting and then the sum is quantized to determine what octave the bank should play in.  If you input a signal from an LFO into the CV input, and then move the slider up and down, you find that the pattern of octave changes varies continuously as you move the slider.  That's why it isn't a slide switch: by careful setting of the slider, you can get sequences of octave changes that would not be achievable if the octave slider only had discrete values.  This is an addition to the original DoubleDeka design, and a very nice design feature.  Feeding low-audio-rate waveforms into the CV inputs can produce quite startling results.  Here is an example:

Modulation Options

This is an area where the 1130 really stands out.  The available modulation inputs are:
  • 1 volt/octave, standard scaling for playing the module in scale from a keyboard or a MIDI/CV interface.
  • Exponential FM, with a panel knob for controlling the ratio.
  • Linear FM, with a panel knob for controlling the ratio.
  • Sync input
  • "Ring" modulation input
The 1V/octave, expo FM, and linear FM inputs all do what you expect.  The available FM modulation index values are very high if you crank the knobs up.  The result tends to trend towards noise very quickly; I suspect this is another aspect of the stair-stepping in the output.  It's a harmonically rich output and applying FM to it creates harmonic chaos pretty quickly.  There is a a switch for the linear FM input that allows a DC-blocking filter to be switched in or out.  A linear FM example:

Linear FM

The "ring" modulation input is kind of strange, but also capable of doing some really neat things.  It isn't really a conventional ring modulation circuit.  Here's what it does: It "squares up" whatever input waveform is applied to the "ring" input jack, so the result is either "high" or "low".  When the result is "high", the output is cut off.  When the result is "low", the output passes through normally.  A consequence of this is that a sine wave, triangle wave, and sawtooth wave of the same frequency will have pretty much the same effect.  The most interesting results occur with pulse and multi-pulse waves at the ring input.  I created a patch with a Synthesizers.com Q106 VCO feeding the ring input from its pulse wave output, and an LFO driving its pulse width modulation across nearly the full range.  At one extreme, the output almost disappears, but at the other extreme you hear an un-modulated output for a moment. Ian calls this "digital ring modulation".  Here is a rather long example of the ring modulation, starting with a low-frequency modulation and then moving into audio frequencies.  Note how the modulation starts by simply cutting the audio in and out, and then begins to resemble more typical ring modulation as the modulation frequency increases:

Ring Modulation

The sync I haven't yet figured out.  There are two modes, a "harmonic" mode and an "aharmonic" mode.  There are some demos on Ian's web site, which I haven't yet managed to reproduce.  I'll post something about it later. 

Access to the VCO Core

The "HF Out" jack provides you with the raw waveform from the ultrasonic VCO core.  You could use this to, for example, feed an octave divider and generate additional waveforms that could be used to generate audio signals to be mixed with the DoubleDeka's outputs, or to do weird sync tricks.  (Don't run it directly to your amp -- it'll fry your speakers!)  It would have been nice if there were also an input that overrode the VCO core's signal to the A and B divider banks.  That would, for example, allow a second DoubleDeka to be slaved to the first, possible at some odd division of the first unit. 

Design and Contstruction

The 1130 appears to be well constructed.  The panel is of the same mechanical design and same thickness of aluminum that Synthesizers.com uses.  There are three circuit boards: one that carries all of the jacks, one that carries all of the sliders, and one that carries the VCO core and other circuitry.  The slider board has some surface-mount components on it; I didn't look closely to see what they are, but they are probably multiplexors for stepping through the sliders.  The VCO is tempco regulated, and trimmers for adjusting scale of the 1V/octave input, and high frequency compensation, are available.  I have so far not needed to touch the calibration.  The rotary pots are from Alpha and feel very nicely damped.  And I didn't have to re-index any of the knobs; they were all spot on.  Max depth from the back of the panel is 2.75", or slightly less than 7 cm.

The panel is well laid out despite being crammed with stuff.  I think this module wins the prize for most items on the panel -- 22 sliders, 5 pots, three toggle switches, and 12 jacks.  The area around the mix knob could get busy if you are using all of the A and B octave CV inputs and the mix CV input.  The A octave slider is a bit close to the coarse tune knob -- I bumped it a few times while tweaking that slider.  But these are all minor quibbles.  You might not think it's possible for a 4U width module's panel to be "full up", but this one is, and SSL did a fine job of laying out the panel so as to keep everything comfortable.  The panel graphics are the same as what Synthesizers.com uses, and the silkscreening is high quality.


You may have some doubts about whether you want to dedicate 4U of precious case space to one module.  In this case -- do it.  Everyone needs one of these; it's a great route to exploring and finding new timbres.  Use the modulations, apply generous amounts of low pass filtering, and will be pleased by the results.  I leave you with one more sound sample.  This is a pulse wave from a Synthesizers.com Q106 VCO being fed to the Double Deka's ring mod input.  The Q106 is being pulse-width-modulated by an asymmetrical sine wave from a Synth Tech MOTM-320 LFO.  The range of the LFO is such that at one end it drives the Q106's pulse width completely into cutoff, and at the end of each sweep the un-modulated waveform peeks out for a moment.


And visit Ian Fritz's Double Deka page for more demos.