To those that build them, electronic music kits are an endless source of satisfaction, despair, relief and a whole host of other emotions. Now a confirmed addict tells of his experiences with one well-known kit in particular.

I didn't need much convincing. I'd already built several of Tim Orr's circuit designs, had a great interest in MIDI through having a DX synthesiser, and been wanting a cheap(ish) way of getting into sound-sampling for some while. So, when I saw the photograph of the Powertran MCS1 on the cover of E&MM September '84 and found that the acronym stood for MIDI Controlled Sampler, 'sounds interesting' was an understated version of the thoughts that came into my head.

After a quick call to Powertran to clarify certain points, I was convinced that I must have one. Convincing my bank manager took a little longer. However, on October 29 of last year, the requisite cheque and order were despatched to Andover, and I sat back in eager anticipation.

But as Mrs Beeton might have it, 'First catch your kit'. Due to, ahem, early production problems, my kit wasn't delivered for another eight weeks, and even then I was short of a few 'minor' items like the EPROM and User Manual. But having waited that long, I wasn't going to be deterred by the absence of instructions - I decided to start building the thing that evening.

If you're planning on building the MCS, I suggest you prepare yourself for saying goodbye to quite a few evenings. Construction took me about 15 hours, all told. Mind you, that was without the guide, so you might manage it in 10 or 12, patience permitting.

The Beginning

The first (and most pleasant) task was to unpack the bits and throw away all the polystyrene chips, something I regretted later when I reflected on how great they would have been for putting in people's crisp packets in crowded pubs. Ordinarily, the next job would have been to check what I hadn't binned against a component list, just to make sure there was nothing vital missing. This wasn't actually possible given the absence of such a list, but that didn't stop me having an educated(?) sniff around.

The components turned out to be to Powertran's usual high standard, with the exception of the IC sockets and key-switches, more of which anon. There are three PCBs in all: one a single-sided affair and the remaining pair two-faced, as it were. But the MCS1's piece de resistance is its main logic board. Quite simply, this is the best PCB I have ever seen in a kit, and quite a bit better than a few ready-built items, too. It looks like a computer-aided design job, but I'm assured that it was laid-out with the assistance of nothing more revolutionary than the human brain. It also comes with a label informing you that it's already been electronically tested - before you've even unpacked it.

The main logic board shares two excellent features with its sister PCBs, namely a silk-screen print of the component layout and solder resistance on both sides. The former (obviously) helps you put all the bits in the right place, while the latter takes the form of an insulating coating that prevents you shorting address lines and the like when you come to soldering them in; it also stops bits of stray swarf and other metallic rubbish having a similar effect, long after you've finished any soldering. It doesn't, however, offer any protection against misplaced glasses of Mouton Cadet '58, so think before you drink before you build.

But the IC sockets are disappointing - cheap versions of what is, after all, an absolutely crucial component. The main problem is the inordinate amount of force you need to apply in order to insert the chips into them properly: it doesn't make electrical connection any more reliable and it puts unnecessary strain on the PCB. And the fact that an altogether superior quality socket is available at only a penny or two more only serves to make the inclusion of these things an even greater silliness on Powertran's part. Ah, well.

The Building

Let construction begin. I spent the first night assembling the case, the front and back panels, and the front and the back PCBs, and doing the mains wiring. This took me two-and-a-half hours. A day later, I put the resistors and the IC sockets on the main logic PCB. This involved the best part of 2000 individual solder joints, and took me just under five hours. During the next session, I finished the logic PCB by fitting the capacitors, semiconductors (though not the ICs) and power supply components. This took me four hours. I spent one, final session wiring up the PSU, the PCBs and the front panel controls, which took an hour and a half.

The above schedule might sound a little on the lengthy side, but the whole process was quite easy to accomplish with the aid of the PCB silk-screening and the relevant copy of E&MM.

So there was my finished - though currently unpopulated - MCS1 sitting in front of me, and there were still a few postal days before Christmas. Would the missing EPROM arrive before the festivities so that I could show off the sampler's abilities to my brother, or would he get the chance to bore me to tears with his QL chess program instead? Well, I'm afraid to say the Royal Mail assumed me to be a company closing down on December 20 and wanting no more post. And sure enough, first post on January 2, my EPROM and elusive MCS1 manual were delivered to my door. If there'd been a decent MIDI interface package for the QL, my Christmas would at least have been bearable...

And so to the 52 million dollar question. Would it work? Before trying to find out, I checked the values of the components I'd put in against those on the overlay. What's this? Capacitors of 0.47mF instead 4.7mF - four of them. Poor printing quality on tantalums is my excuse.

But nothing else seems to be amiss. Flinchingly, I turn on the machine with the help of a long stick. No bang, no puff of smoke, no wailing or gnashing of teeth - just the quiet hum of an underloaded transformer. I check a few voltages: all present and correct. Time to have a cup of coffee to see if anything blows up while my back is turned. One cup of Nescafe later, I check the voltages again: still no problems.

Scarcely able to believe my good fortune, I populate the board as per Mr Orr's instructions. Still nothing goes bang. I'm getting worried. To make matters worse, the unit actually seems to work. At least, all the front panel controls - with the exception of one keyswitch - seem to be behaving themselves admirably.

I decide there's only one thing for it. I dig out my trusty oscilloscope, and set to work looking for erroneous waveforms. Eventually, I discover that the PLL frequency is below spec. Aha! Another wrong capacitor, this time 470pF instead of 47pF. Obviously, Powertran had had the same trouble with powers of ten that I do, supplied me with the wrong component, and I'd fitted it without noticing. Anyhow, a quick dip into my component stocks soon solves the problem, and the PPL frequency comes out something like normal (680kHz compared with TO's quoted 700kHz - sounds close enough for me).

I'd found something wrong and felt a little more at ease, so I went to bed (it was 3.15am). But I was still worried that things were going far too well. Next day I checked the audio side of my MCS1, and found that it was all rather superb. Serves me right for being such a pessimist, I suppose.

The last job was to calibrate the MCS1's CV/octave relationship, something that proved fairly difficult to get a grip of without my reading through the procedure a few times. I attempted some semblance of accuracy with an admittedly inadequate 3½-digit Beckman multimeter, though the whole process took me a good couple of hours. But imagine my surprise when I plugged in my Pro One and found it to be decidedly ill-tempered. No two semitones were alike, and even after setting-up by ear, results were disappointingly unpredictable.

Thinking the error might lie in Sequential's wondrous monosynth, I checked the Pro One by plugging it into a multimeter and then into a Transcendent Polysynth. This done - and several hours later - the Pro One was satisfactorily recalibrated and put to work on the MCS1 once again. The results were a little better, but there was still some slewing between notes - regardless of where the CV/gate information was coming from.

To be fair, though, this is a property of the VCO that can't easily be remedied without introducing other problems, and in any case, I suspect some owners will make use of the facility without even noticing something is amiss.

The Quibbling

I've already mentioned the below-par IC sockets, an economy measure on Powertran's part that's difficult to justify from where I'm sitting now; maybe things are different in economic reality. Either way, it's not the only fault I can find in the way this kit is packaged. The major quibble surrounds keyswitches. Frankly, I can't decry those on the MCS too much, even though their dreadfulness is more than compensated for by the excellence of the PCBs. The switches each comprise four separate parts that have to be assembled before you can solder them in. That wouldn't be so bad if the MCS1 didn't require 24 of these awkward little buggers. Tedium sums up the main emotion one experiences on going about fitting them.

They also lack feel. In fact, one of them had so little it refused to work at all; I had to modify it before it would function properly. Part of the problem lies with the LED holders (each of the keyswitches has one built in, in itself a useful enough feature to have), which turned out to be very variable in size and prevented their parent keys from being depressed.

I fear this is another false economy on Powertran's part, as these switches are - unsurprisingly - some of the cheapest available. What complicates matters further is the fact that half the switches have to be assembled with their LEDs the other way around, and the only reason I can think of for this is that it made designing the PCB easier.

Another niggle is that the PCBs, especially the front panel one, could do with being better braced mechanically. A couple of extra pillars in the middle of the keyswitch section would stop the keys retreating whenever you press them, a feature that contributes to the difficulties described above. A similar provision somewhere in the midst of the logic board would probably help, too.

And another thing. A lot of the screws that come with the MCS1 kit are obviously too long for the job. The main offenders are the logic-board fixing screws, but the voltage regulator/heatsink items are also guilty.

Then there are the tinier, less significant points that could be cleared up on future production MCS1s. Why, for instance, can't the front panel PCB be extended to include some more panel components, like the pots and the two LEDs? And why don't Powertran adopt ribbon cable connectors between the boards? They'd make construction a lot easier, and also allow PCBs for ready-built machines to be tested prior to assembly.

Well, I guess that sounds like an awful lot of flaws. It isn't. In most respects, the MCS is a really well thought-out and executed kit design. But as with a lot of things, the faults that are present draw attention to themselves simply by virtue of being few in number.

The Judging

Despite a sampling orgy involving cats, milk bottles, the postman and a cast of thousands, I'm still getting to grips with my MCS1. It isn't the easiest bit of gear to use. However, it is obvious, even to my meagre intellect, that Mr Orr has excelled himself with this one.

The MIDI performance is exemplary, the effect of aftertouch-induced vibrato simply stunning to behold. The delay line mode is also superb, with both sample rate and memory size being variable over a range wide enough to give the machine an awful lot in the way of versatility. As Powertran themselves admit, the CV/gate section is the one area of performance that really brings the MCS down. It's usable, but it isn't perfect, and I reckon most people will end up using it primarily for special effects purposes.

The kit is mechanically OK and, in spite of its size, surprisingly easy to build. After all, if I can do it without a manual, it must be a doddle. Talking of manuals, this one lacks the notes on component identification, hints on order of component assembly, and tips on how to create a neat appearance that were once de rigeur for Powertran literature. Now, that sort of thing isn't of much interest to a seasoned pro like me, but I imagine it would be useful to mortals of lesser kit-building experience and confidence.

Does the omission imply that Powertran don't think their kit is suited to beginners? If it is, it's a mistake, because as I've just intimated, the only thing about putting the MCS1 together that's in any way daunting is its sheer physical size. Admittedly, the equipment required to test the machine and calibrate the CV/octave relationship (oscilloscope, frequency meter and multimeter) aren't exactly average hobbyist material, but it is possible, assuming the machine works first time, to set it up with no more than a multimeter and a pair of ears.

Anyway, I'm looking forward to linking mine to a BBC Micro and getting to grips with what could, given availability of the right software, become a powerful - and gratifyingly cost-effective - computer music system in its own right. The sound dump software is out now, but what are the 'future facilities' hinted at in the MCS brochure? Actually, I think I know already, but I'm keeping it under my hat. So there.