Way Out Ware TimewARP 2600

Can this second in-depth software emulation of the ARP 2600 improve on Arturia's existing 2600V?

This new instrument from Californian company Way Out Ware joins Arturia's 2600V in emulating ARP's 2600 semi-modular synthesizer. Installing and authorising the software is straightforward if you do it on a machine connected to the Internet, but WOW provide a couple of workarounds if you prefer to keep your studio computer solely for music use.

Oscillators & Noise Generator

As you would expect, TimewARP' s oscillator section is modelled on that of the original ARP 2600, the only differences being that VCO 1 and VCO 3 each have a sine-wave output. (This is in sharp contrast with Arturia's 2600V, which offers additional octave selectors, oscillator sync, and scaling controls for the keyboard CV inputs.) So how true to the original do TimewARP' s oscillators sound? LIning up my own ARP 2600 and 3620P keyboard alongside my Mac running TimewARP 2600, I started by comparing the the sawtooth waves, and was encouraged to find that the two instruments sounded very similar. This is surprising, because the waveforms show marked visual differences on an oscilloscope. Indeed, WOW seem unable to tell the difference between a sawtooth wave and a ramp wave. Immaterial at audio frequencies, this makes a huge difference when using the oscillator as an LFO. However, don't be too harsh on the developers — I only recently discovered that the ARP 2600's oscillators produce sawtooth waves at audio frequencies, but a misshapen ramp (that changes markedly when you start patching it to different destinations) in LF mode.

The sine waves appear more similar visually, but listening to them reveals that the ARP 2600 is bright, whereas TimewARP sounds like the 'pure' tone that a sine wave should be. Repeating these experiments with the triangle wave, I found that TimewARP' s is again more 'true' than the ARP 2600's, and I obtained equivalent results for the square and pulse waves. This implies that WOW have modelled their waveforms on ideal waveshapes, not those derived from real ARP 2600s with real components and less than perfect calibration.

However, there's an even more significant way in which TimewARP 2600 differs from the real thing. When tuned to the same pitches, the phases of all three oscillators are always the same! Without phase-locking (as occurs by accident in some vintage synths) this would only happen by chance on an analogue synthesizer, even if you could tune the oscillators perfectly. By locking the oscillators' phases and eliminating any movement, WOW have significantly affected the nature of any two- and three-oscillator sounds, eliminating the warm, organic movement that occurs on a true multi-oscillator vintage instrument.

Filter Section

The TimewARP 2600 filter section sticks to the configuration of the ARP 2600, with the exception of an added KBD CV Amount control. (On the ARP 2600 itself, this is always 100 percent). Testing the ranges of the two instruments' filter cutoff frequencies, I found that the ARP's filter ranges from 12Hz at the low end to way, way off the scale at the upper end. (When last I saw it, it was heading up toward 30kHz and still going strong.) In contrast, TimewARP' s filter ranges from 10Hz at the bottom to approximately 10.2kHz at the top. Despite WOW's claims to the contrary, I was unable to make it increase beyond this, no matter how many positive voltages I applied to its CV inputs. This means that at maximum cutoff frequency with no resonance the soft synth audibly attenuates high frequencies to a far greater extent than the hardware. I suppose that you could hypothesise that WOW modelled TimewARP 2600 on a later version of the ARP 2600, one that used the ARP4072 filter. Due to a miscalculation, the 4072 had a bandwidth of about 12kHz, so ARP 2600s from 1976 onward had a duller sound than earlier units.

The TimewARP 2600's sine-wave spectrum (bottom) and the ARP 2600's sine-wave spectrum (top) for a 440Hz tone. While the emulation is almost mathematically pure, the hardware synth is much brighter.It was while making these tests that I noticed that the cutoff control's Tool Tip shows the wrong frequency values. When I set up the filter resonance so that the filter was self-oscillating and the Tool Tip read 440Hz, the actual output measured at 621Hz. That's six semitones too high! To obtain a true 440Hz output, I had to set the filter cutoff frequency to 309Hz.

Transient Generators, VCA & Output

TimewARP' s transient generators, VCA, output mixer, and reverberator follow the design of the ARP 2600 precisely, with no additional frills. However, WOW have taken the opportunity to extend the envelope times of TimewARP 2600. I measured the longest Attack as 56 seconds, but more important, of course, are the minimum Attack, Decay, and Release times. With all the ADSR controls set to minimum, the TimewARP' s transient was quicker and better defined, which resulted in more precise clicks and snappier transients.

To the furthest right of the control panel you'll find the output Mixer and the Reverberator. Again, I can see no difference between WOW's implementation and the original synth. However, the sound of the reverb is nothing like that of a spring. Don't get me wrong... it's a usable lo-fi reverb, but the definitive 'boyoyoyoyoinnggg' of the ARP is completely lost.

Modular Toys

We've now come to the end of the ARP 2600's conventional signal path but, of course, this is only part of the story. Underneath these modules lies a set of Voltage Processors and a powerful Sample & Hold and Electronic Switch section, all of which are recreated accurately on TimewARP.

To the far left of the panel lie the Ring Modulator and the external signal input modules — a Preamplifier and an Envelope Follower. I rather like the Ring Modulator on TimewARP; as it offers both AC- and DC-coupled operation, and sounds much like the device on the ARP 2600. The external signal input is, however, an improvement on the original. When you launch TimewARP 2600 in a suitable application, you can choose from none, one, or two signal inputs. If you choose the last of these options, the two channels of audio presented to TimewARP appear independently on the Preamplifier's Line 1 and Line 2 outputs, ready for directing anywhere within the synth. If you like to use synthesizers as signal processors, you'll like this a lot.

Keyboard & LFO

If we ignore the Blue Meanies and Grey Meanies, the earliest ARP 2600s were supplied with the ARP 3604P keyboard, which provided nothing more than pitch, trigger, and gate information. ARP later replaced this with the ARP 3620, a sophisticated duophonic unit that provided single- and multi-triggering plus a flexible triple-output LFO with delay. As you would hope, TimewARP' s keyboard section emulates the ARP 3620, although it does so with 60 keys rather than 49. The layout is totally different from that of the original, but the facilities appear to be the same, with the unavoidable absences of the footpedal inputs for portamento on/off and the interval latch. This means that the ARP 2600's unusual duophonic mode is retained, allowing you to play two pitches (although only one underlying timbre) using a single voice.

The graph (top) demonstrates the differences between the low-pass filter responses of the ARP 2600 (pink trace) and TimewARP 2600 (green trace) using the same noise signal— the modelled processing audibly attenuates the high frequencies when compared with the original hardware. The amplifiers (bottom) also respond differently, with the TimewARP 2600 (red trace) giving faster envelope times, and therefore better defined transients.TimewARP' s keyboard LFO is, in one way, a significant improvement on the ARP 3620's, because its maximum frequency is 100Hz rather than 20Hz. However, as found elsewhere in TimewARP 2600, the waveforms are a bit too close to the ideals and, unlike the 3620, TimewARP' s LFO sine wave is gated by the keyboard — on the original, it is free running and ever-present at its output, which is much more flexible.

New Friends

In common with other soft synths, TimewARP allows you to control parameter values (but not patching) over MIDI using velocity, aftertouch, the mod wheel, and one assignable Continuous Controller. You can determine the minimum and maximum ranges of each controller, select one of three response curves for each, and invert the effect if you wish. While this system does everything that you might reasonably wish, only three-quarters of its settings are saved in the patch — the Continuous Controller is part of a global map that you must save and load independently. That's daft.

Click on the TimewARP 2600 logo to the bottom right of the panel, and a small menu appears. This allows you to save and load the global MIDI maps that you create, as well as providing access to the micro-tuning options and the MIDI Clock synchronisation window. The micro-tuning has some interesting scales, some of which I have not seen before. Unfortunately, you cannot create new scales of your own.

At the top of TimewARP 2600 you'll find the inevitable menu bar, and this gives you access to the final set of facilities. Most of these are concerned with the saving and loading of patches. The Patch Manager is a hierarchical (Group-Category-Patch) system for importing, exporting, organising and, if desired, reordering your sounds. Unfortunately, it does not lend itself easily to selection with MIDI Program Change commands, because only the first 128 patches in any Group are available. This means that you have to limit the number of patches in upper Categories if you want to be able to 'reach' patches lower down the list. Furthermore, despite allowing you to move around the lists using the cursor keys, the highlighting does not move with you, which instantly leads to confusion. Alongside the patch management controls, there's the polyphony selector (one to eight voices), a reset button that re-patches TimewARP 2600 to a basic default and silences the thing, a MIDI indicator, output level meters, and a CPU usage meter.

Absent Friends

Unlike the Arturia 2600V, TimewARP does not dispense with the ARP 2600's (in)famous VCA thump, which occurs when you play notes with the Attack and Release set to their minimum values. Likewise, the Initial Gain at the top of the VCA section is retained. I was also pleased to see that I could use TimewARP' s Voltage Processor sliders as CV generators (which was a trick missed by Arturia). Everywhere you look, it's apparent that TimewARP 2600 is more faithful to the ARP 2600's 'look and feel' than is the Arturia 2600V. But this also means that the Arturia synth offers far more in the way of goodies and extras not found on the original.

These 'extras' are non-trivial. I've already mentioned those found in the 2600V' s oscillators, but TimewARP 2600 also loses the Arturia's four additional filter profiles (12dB/octave low-pass, high-pass, and band-pass, plus notch), the dedicated chorus/delay effects section, and powerful extensions to the Voice Processor architecture. Likewise, the Tracking Generator (a programmable modulation generator that allows you to create unusual modulators and direct them anywhere you choose within the 2600V) is gone. Then there's the loss of the 2600V' s 10 additional patching VCAs.

But much as you may (or may not) miss these, the absence that will probably cause WOW the greatest headache in terms of competing for your cash is that of the ARP 1601 sequencer. Not just confined to producing Giorgio Moroder-esque bass lines of the 'I Feel Love' variety, this offers all manner of polyrhythmic effects and complex sound-shaping possibilities. Now that the world has become accustomed to it being there, I can't imagine many people being blasé about losing it.

Let's Do The TimewARP...

To test TimewARP 2600 in a musical context, I ran it as a stand-alone synth on my G4 Powerbook and as an AU plug-in within Plogue Bidule. Having launched the program, I set up the obvious stuff like MIDI channel and polyphony (all easily done), and immediately discovered that the CPU meter is an ever-present worry. This is because TimewARP 2600 — especially in its stand-alone incarnation — has a humungous appetite for CPU power. On a good day, I might suggest that this is due to a careful implementation of every nuance of analogue circuitry, resulting in extremely power-hungry algorithms. Alternatively, I might suggest that it smacks of badly optimised code that chews power unnecessarily. Either way, the result is the same. If I wanted absolutely reliable and glitch-free results on a 1GHz computer, I had to set the polyphony to... one!

To be fair to WOW, my Powerbook is close to the minimum specification recommended for TimewARP 2600, but I was still shocked. With simple patches, I tried to get as many as three notes, but even then I couldn't use long release times, because the software generated digital hash as it tried to arbitrate between the tails of previous notes and the onsets of new ones. There was also the digital equivalent of mistriggering, as the software tried to decide which new notes to play! Strangely, the program proved to be less greedy when running within Bidule, and I could run multiple soft synths with less glitching than generated by TimewARP alone in stand-alone mode.

When I set up the same simple patches on both hardware and software synths and listened to the results, there was little similarity. I had the original ARP 2600 patch book to hand as I wrote this, so I tried programming its patches on TimewARP 2600, and these did not result in the correct sounds. Nevertheless, as I tweaked patches — tailoring the envelopes by ear and adjusting the filter settings to match the response of the real ARP — the timbres started to move closer together, as did the articulation of the notes. Before long, I found that I could make the ARP 2600 and TimewARP 2600 sound quite similar, even though TimewARP lacked the brightness and sparkle of any sounds that were supposed to be extra bright and sparkly.

As I started to programme complex patches, the differences between the original and the copy became more obvious, so I next attempted to increase the polyphony — such as I could — and to create some interesting polyphonic patches. I found that TimewARP 2600 is rather good at percussive sounds such as electric pianos, which is to be expected given the snappiness of its envelopes. However, I failed to obtain the warmer ensemble sounds that I had hoped. The factory sounds bear this out, with a heavy concentration of leads, basses, and effects, but few polyphonic/ensemble offerings.

It was around this time that I really started to miss the additions provided in Arturia's 2600V. Processing external signals is less fun when you don't have a MIDI-sync'ed sequencer to manipulate complex parameter changes in synchronisation with incoming sounds, and the loss of oscillator sync and the additional filter profiles (while irrelevant in any comparison with the original 2600) is a shame.

I also found TimewARP' s user interface not always to be as slick as the 2600V' s. For example, there is no equivalent of Arturia's Move Away Cables facility, which causes patch cables to move out of the way of the mouse pointer so that you can adjust parameters that would otherwise be hidden by electronic spaghetti. Neither are there the useful colour-coded patch cables and patching menus. But I'm also worried about WOW's quality control. Has anybody else noticed that LFO Speed is written as LFO Seed on the keyboard section? This makes me worry about other oversights.

Happily, the operation of TimewARP 2600 was pretty robust, with just a couple of freezes during the review period. These tended to happen when I was doing something innocuous, such as selecting patches using the Patch Manager. In contrast, over-stressing the CPU usually just caused ghastly noises.

Conclusions

Way Out Ware were going to have to work hard to live up to their claims that they have 'perfected the software synth version of the ARP 2600'. The company's marketing blurb also claims that its developers started 'with an accurate digital emulation of every module from the original ARP 2600, and then added new features', but given the differences between the original synth and their emulation, I'm not sure that they were accurate enough.

I'm also unhappy with WOW's claims that they 'spent two years reinventing the mathematics behind analogue emulation, developing proprietary and optimised virtual analogue oscillators and filters'. It seems far too grandiose to claim that they have reinvented modelling, and, given the CPU drain occasioned by TimewARP, the word 'optimised' appears to be misused.

Some people have suggested that, while less enhanced than the Arturia 2600V, TimewARP 2600 's sound is closer to that of the original ARP. I beg to differ: while the graphic design of TimewARP is truer to its inspiration, it exhibits a similar degree of sonic authenticity, or lack thereof, as other soft synths. Whether this is a problem or not is open to debate — many of the issues I've raised have no bearing on whether TimewARP 2600 is a good synth or a bad one; they simply tell you whether it is an accurate recreation of the ARP 2600 or not. So let's view TimewARP 2600 as a synthesizer that has the look and feel of the ARP 2600, and which shares much of its character. It then fares quite well, although I can't help feeling that the better-specified Arturia 2600V fares rather better.