WHEN DOES A DIGITAL DELAY LINE become a sampler? Is a MIDI socket enough? Is it okay just to chop the ends of your sample, or does it only make the grade if you can tidy up the beginning as well? And where does looping come into it.

All these questions and possibly some of those questions over there in the corner, will be picked up and rattled firmly within the next few hundred words.

What is the Korg SDD-2000? ("Qu'est-ce-que c'est le Korg SDD-2000?"). It is a digital delay line. ("C'est un leigne echoes du digitale.") With a maximum 4368ms delay. ("Avec un arrete premier de 4368ms.") And a modulation section for producing the usual chorus, flanging and ADT effects. ("Et un division... er... du modulation pour... er...[qu'est-ce-que c'est le 'flanging?'!"). But which can store up to 4368ms of sound and replay it, monophonically, from a MIDI equipped keyboard. ("Eh! Le Replay? C'est pour le machine de pinball.") And a top frequency response of 18kHz at 1092ms. ("Sacre Bleu at donnez-moi le vino".)

As a delay line the Korg behaves impeccably. A rack mounted creation (the army issue 19in), it's destined primarily for recording systems. All the sockets are at the rear, and there's only one jack input so the designers were logically presuming other inlets from a mixer. That said, there are live aids, (no, not you Geldof), such as footswitches to step up through the 64 programs or MIDI operated program selection. Choose patch 28 on say, your DW6000 and 'echo' 28, ideally matched to your keyboard sound, will leap into place on the 2000.

An inspection of the front panel should lead us to a greater understanding, overall. There's a four LED headroom bargraph plus volume control to set up the best input level, closely followed by a level control for the direct output, a switch to bypass the delay completely, and a Record Cancel indicator (footswitched from the rear) that 'escapes' you from the effected half of the signal.

Next are five buttons and an LED, all to do with the sampling section which we'll overhaul later. This is neighboured by the programmer — all the controls to set the modulation effects — and a six figure LED display to show you where you're at, soundwise.

The controls are logically logo-ed and easily placed, with the exception of the last, the incremental controller. This is the knob that selects, the programs manually, or which can step, say, the delay time by increments of 1ms. It's a constantly rotating knob — goes round and round; each click is an increment. Now most manufacturers faced with such a situation would either give you an alpha numeric keypad to dial in the numbers (luxury), or push buttons that would glide through the values, depending on how long you press (basic but practical). Korg's solution is not only old fashioned but tedious and long winded. To climb to the longest delay takes ages — not a fast turn of the knob but virtually a cranking action where you're going 'round' the control four or five times. Too slow and these days relatively easy to avoid.

Out back, where we keep the chickens, there's the single input (flickable between -35dB and -10dB gain), direct, +Mix and -Mix outputs (the latter being an out-of-phase combination of direct and effected signals) four footswitch sockets (record, program up, record cancel and bypass), a drum machine trigger socket, master tune control (for bringing the 'samples' into line with your keyboard), and MIDI In and Thru.

Setting up one of the 64 memories to the delay time of your choice involves the programmer section. You select effect level, feedback amount, time, modulation frequency and modulation intensity by touching the appropriate button. The LED screen then shows the present value, and turning the incremental control will step it up or down. (Most parameters have values of 0-31.) No complaints at all about the degree of control, it's very fine, especially the time which goes up in steps of 0.1ms from 0.0 to 9.9ms. That's useful for selecting phase or flange effects with a particular resonance.

In the x1 mode, the delay time runs up to 1092ms at the best frequency response of 30Hz to 18kHz, and it does behave cleanly and clearly. Getting to longer delays is achieved by punching the x4 button, and the sums are obvious. Max time is now 4368ms, but you don't get something for nothing, and the frequency response is consequently divided by four — 4.5kHz is now your top whack, and you notice it.

I shall now tell you my favourite Vangelis, 'Blade Runner' setting — lucky readers. Echo time of about 300ms, effect and feedback levels fairly high (to taste, as they say), then add a very slight, very slow modulation to produce a wavering, vibrato quality, not to the original note but to its fading echo. Best done on dreamy, monophonic bits.

Before we get to sampling, there is an additional trick the Korg performs. It can accept a trigger from a drum machine, and use that to dictate the delay time so echos will fall in perfect rhythm with the drum pattern. A MIDI clock can be applied in the same manner. Essentially, the first trigger starts the delay running, the second stops it, and in between is the correct phrase length.

The Korg samples in two modes. Sequence takes the sound you've stored and keeps playing it over and over, as long as your finger is on the key. Sample plays it once. Both are velocity sensitive with 5-bit resolution.

When recording in either position, the 2000 will wait until it hears an input that crosses the +3dB mark on the headroom indicator. The sampling then starts automatically, either until the memory runs out or until you punch the record button again, telling the 2000 to lay off.

This, of course, is wonderfully simple when you're recording an instrument that you twang once. If you're attempting to rip something off tape/LP — unbelievable but some musicians do do it — you'll have to tap the sample switch just as the right part comes up. That will need practice but it's not impossible.

Now the 2000 is primarily a digital delay with sampling thrown in as a bonus, and some compromises are to be expected. But when the actual recording task has been made so easy, it seems a shame that they have made the rest fiddly.

The number of keyboard keys across which a sample can be played relies on the time mode you opt for — at x1 it will play over an octave; at x4, over three octaves. To a certain extent you can move these octave and three octave areas around on the keyboard, but it is restricted. For example, you might want a three second sample to occupy the bottom octave and a half so you can play sampled mono bass lines against a different synth sound in your right hand. (As every new note retriggers the Korg, you have to isolate 'zones' or it will try to follow all you do.) Unfortunately, it won't work. You can have a 1 second sample on the bottom octave, or up to 4 seconds over the bottom three, but the system won't permit any finer divisions down that end. Similar problems arise if you attempt to declare a solo-ing patch at the top end. In the middle you're okay.

Korg call these areas 'Supported Note Ranges', but have more to add. If your sample came out at its normal pitch on the lowest note in the range, you'd have three octaves of higher playback; if normal on the top note, you'd have three octaves lower, anywhere inbetween and you'd have a mixture. This 'Sampling Note Setting' can also be fixed to your taste.

The manual hardly helps demystify these actions, but at least they're not actively irritating. You can't say the same for the Calibration procedure. Each time you make a recording, the 2000 will recalibrate itself, to make sure the sample is in tune with the MIDI information it's receiving from the keyboard. It takes five or six seconds to do this in Sample mode, and 15 seconds in Sequence mode. In other words, every time you record a sound, there's at least a five second delay before you can press any key on the keyboard and hear if it's been successful — not advisable for the impatient among us.

Not that any of the above impair the quality of sampling, which is very good in the x1 mode (perhaps less impressive at the longer times). They just slow the process.

Where serious samplists would begin to show concern is in the editing and looping arrangements. Once you've made your recording, you can splice off the unwanted end by reducing the echo time with the incremental control. (The manual somehow infers that you then lose the edited section forever. You don't, it's still in the machine and can be reclaimed.) But there's no way of editing the beginning or selecting a chunk from the middle — common desires among the samplist fraternity.

The Sequencing option automatically loops the end of the sample back to the start, and repeats it. Again, without full editing facilities, you're lucky if you get a sound without a click as the join is made.

Finally, the structure of the internal memory seems such that you can only use the 2000 in two ways — samples of up to 1092ms with a frequency range of 30Hz to 18kHz, and samples of up to 4368ms with a frequency range of 30Hz to 4.5kHz. Spot the problem? Samples that are unlucky enough to be 1100ms long will have a top range of 4.5kHz plus 3268ms of unused memory to their credit. There's no way of efficiently sharing out memory to get the best response for each sample — a fault not uncommon among delay/sampling dual machines to be honest. Perhaps a third option of x2 (9kHz top whack) would have helped.

So for someone who wants a powerful digital delay (which it is) and wouldn't mind spending a few extra bob for some sampling ability — albeit somewhat limited and a mite long winded — the 2000 is worth investigation. Four seconds worth of sampling time is not to be sniffed at and the velocity sensitivity is very handy. However a serious samplist, intent on doing a lot of playing around with noises as his prime objective could rapidly exhaust its possibilities.

KORG SDD 2000 sampling/delay: £799

CONTACT: Rose-Morris, (Contact Details).