Digital Percussion Machine
Article from Electronics & Music Maker, March 1984
Hammond is hardly the sort of company that springs to mind in the context of the latest gadget for the rock musician, but there's no denying their ability at producing electronic organs with a human touch, and, like their fellow manufacturers, Lowrey and Technics, they seem to have got their digital voicing techniques successfully together in their latest and more up-market instruments.
The DPM-48 reflects this policy with a not inconsiderable degree of success - and it also looks good. Unlike the Drumulator, for example, which has a certain spartan quality in its pressed steel construction, the DPM-48 looks much more up-market with its nicely solid aluminium front panels and well laid-out, colour-coded buttons. Both these and the individual level faders at the back work positively, with none of that sloppy feel so often found on equipment manufactured on a tight budget. The unit is also more compact than some, measuring just 13.75"(L) x 10.4"(D) x 3"(max H).
Opening the unit up for servicing and replacing fuses or EPROMs (should the need arise) is simple, though there are a number of flying leads and ribbon cables to contend with. The processing side of digital drumming goes on on the PCB attached to the front panel, where the ubiquitous Z-80 type of processor is to be found doing its thing along with the pushbutton keys, LED display, and so on. On the track side of the same PCB there's a single 2732 EPROM, which a modicum of detective work suggests contains the necessary firmware for running the machine. The other ten 2764s and one 2732 EPROMs (a grand total of 84K) are found on the other main PCB, along with much macho TTL, one MCI408 DAC (8-bit, non-companding type), four AN6554s, and four LM13600s.
Because of the Japanese flavour of some of the chips, it's a bit difficult to be 100% certain of what's going on, but I'd hazard a guess that the AN6554s are dual sample & holds used in a multiplexing fashion like the Drumulator and MXR unit. Unfortunately, the manual is as silent as a grave when it comes to telling you about what goes on inside the DPM-48's head. Because of this, it's also impossible to tell how much memory is allocated to each sound. But 84K is a fair whack above most of the competition (64K for the Drumulator, 52K for the MXR, and 64K for Oberheim DX, as against 124K in the LinnDrum), and this shows up in the generally high quality of the sampled drum sounds.
If there's a single distinguishing feature between the DPM-48 and its more expensive brethren, it's the recess at the front where the RAM cartridge normally gets slotted. The DPM-48 is the only digital drum machine on the market to make use of this much-maligned storage medium, and I feel it's a strong point in its favour, although it's something of a mixed blessing. The problems with cassette storage are legion to anyone with a micro, and precisely the same problems of drop-outs, run-down batteries, and misalignment of tape heads applies to the loading and saving of drum-machine data. RAM cartridges provide a virtually instantaneous means of changing over from one set of songs to another, albeit at the cost of buying a set of cartridges. The DPM-48 RAM cartridges use two 5118 CMOS RAM (1K x 8-bit each), giving just 2K for song storage, and therein lies the only real problem with this unit, namely that you're limited to accessing only 3 songs at any given time. Whether or not that's a problem in practice is up to how you go about organising your creativity. In the studio, it's no problem at all, but on stage, you'd be well-advised to take time labelling the cartridges, not to mention keeping them out of beer mugs...
A word of warning: if you choose to change RAM cartridges in a quick-fire salvo fashion without switching the unit off first, you stand a good chance of condemning something or other to the great Silicon Chippy in The Sky. So, take heed of Hammond's warning: 'Turn off the DPM-48 before setting or taking out the RAM cartridge. Do not insert the RAM cartridge with unnecessary force. It could cause a great deal of trouble.'
Like the Drumulator and the MXR 185 Drum Computer, the DPM-48 has sounds digitally 'recorded' in EPROMs. The provisional (I hope) manual supplied with the review unit informs me that 'twenty-two voices were all digitally recorded, so DPM-48 recreates even the most sensitive overtones' (whatever that means). If that's really true, it's extremely unusual, because most digital drum-machines make do with a smaller number of drum samples than the actual drum sounds available from the unit, by using the tricks of altering the sampling rate to get different pitches and a post-DAC VCA to vary the dynamics for accenting purposes. Since there is a lot (84K) of EPROM sample space in the unit, it is conceivable that this is what they've done in part, but I'm sceptical of Hammond's claim that they've sampled as many as twenty-two separate voices, as that'd give less than 4K for each, and there's no way that a decent (and the DPM-48's is) crash cymbal can be squashed into that meagre pint-pot. Hammond also say that they recorded the total drum-kit sound for every individual drum sample, the idea being to pick up whatever sympathetic vibrations are engendered by a mighty thump from the bass drum. But does preprogrammed crosstalk 'make DPM-48 produce the liveliest sound', I wonder? Well, here's what actually emerged from the DPM-48 in practice...
Toms 1/2: Tuned in thirds, a good duration of about 0.4", and plenty of membrane bounce, but suffers from a halo of (quantisation?) noise, though this wouldn't be obvious when buried in a mix. I'd say that both sets of toms would benefit from being re-recorded at a faster sampling rate at the expense of length of sample (which is probably too long for its own good). As Ian Gilby said in his review of the MXR unit (E&MM, October '83), the tom sound that's in vogue at present is the African tom, its main features being a gut-gripping impact tone and a fast decay, and perhaps Hammond would do well to go for this rather than what they've got at the moment.
Toms 3/4: Lower pitched and a slightly longer duration of about 0.5", with a good solid tone. Again, on the noisy side. Because the deeper tones of this set of toms aren't proportionately longer than toms 1/2, and since the sound is that of a larger set of drums rather than small ones put through the digital mangle, it's obvious that toms 3/4 come from memory that's separate to their higher-pitched counterparts. On top of that, this second set of toms goes through a separate D/A channel, so you've got the facility to play two toms (from the two groups) simultaneously, unlike the LinnDrum and Oberheim DX, which have just one tom sample doing the time-sharing bit.
Bass 1/2/3: A nice meaty kick to this, though EQ would help to make it a mite 'clickier'. No noise to speak of. Bass 2 just adds a few more dB to the output to accent it, but Bass 3 is a curiosity; I think it's meant to be a third dynamic level, but it actually sounds as if someone's laid some snares across the drum skin. Bearing in mind what the manual says about separate drum samples, this is undoubtedly a different sample to the other two bass-drums. The problem with Bass 3 is that it seems to have lost its percussive identity. In fact, it almost sounds as if the sample is being clipped in some weird and wonderful way in getting to the outside world (or perhaps on recording?). Mind you, it's still usable; I'm just not sure when and where.
Hi-Hat Closed/Accent/Open: For once, the accented version of the closed hi-hat isn't just the latter at a louder dynamic level. This time, the sound actually changes quality (the benefits of the multi-sampling approach). Ten bonus-points to Hammond... What's more, the open hi-hat sample is also excellent.
Snare 1/2/3: A good, personable snare, this, and guaranteed to work its way through to the top of the mix. Like the bass-drum, this is as clean as a whistle as far as noise is concerned, which makes me wonder why Hammond let the toms get away with being so dirty. Snare 2 is just a few steps louder than Snare 1, but Snare 3 is another different sample, this time an effective 'near the edge' sort of sound.
Ride 1/2 & crash cymbal: This is the really tricky territory where less than angelic machines fear to tread. To be or not to be a crash cymbal, that is the question... Well, last things first, this is a good crash; it's not perfect, but it does provide a darn good secondful of sound that decays with grace rather than being caught in flagrante delicto. Judging by the layout of EPROMs on the voicing PCB, I'd say that Hammond are using three 2764s for this one sound. 24K is a lot to splurge on a single sample, but its worth is well and truly proved in this case. Ride 1 and 2 (the latter just a notch up in dynamic level from the former) don't fare quite so well as far as abrupt cut-offs are concerned, but in the context that a ride cymbal is normally used, I don't foresee any problems. The important point is that they perform well while sample data is there for D/A conversion.
Cabasa 1/2 & Clap: The organ pedigree is showing here. Cabasas are those gourd-like things filled with dried peas and shaken to death by organs pretending to be of Latin-American descent. Actually, they're quite effective if used sparingly and well back in the mix. The DPM-48's cabasas lend themselves well to this requirement because they've been recorded with a cautious eye on the meter. The clap, on the other hand, deserves an entire paragraph of criticism (don't worry, I won't). This definitely wins my nomination for the worst clap of the century. If it sounds like anything on the face of this blighted planet, it's the bark of a dog being force-fed with worming pills - a sick dog, in other words. I'm sure the main problem with this sample is that it's being played back at too slow a sampling rate, but, of course, without any sort of tuning option on the DPM-48, it's impossible to test that theory.
Agogo 1/2 & Rim-shot: An agogo is a set of two cowbells, tuned apart by a third. Fortunately, that concurs with Hammond's idea of an agogo, a point which saves a lot of potential egg on the face. These cowbells are actually rather nice. Still, cowbells are cowbells - alpine symphonies and all that sort of thing. On the other hand, the rim-shot is just what the doc ordered for that crisp, clean feel when you're tired of the ride.
As drum-machines go, the range of sounds on the DPM-48 is very generous, but why, oh why, is there no tuning control, either overall (as on the MXR) or for individual voices (as on the Oberheim DX - admittedly not possible with the multiplexing approach)? And as for the 'clap', the sooner the DPM-48 gets a shot of a megaunits of penicillin the better!
If I've a general point to make about the DPM-48's drum samples, it's that they're fairly traditional voicings of the sort you'd expect to find in a drum computer coming from the electronic organ stables. To cater for the quirky demands of the discerning rock musician, Hammond might consider offering alternative EPROM sets that go a bit further into the sort of new-wave percussive territory explored by Peter Gabriel and others.
One of the sensible things about the DPM-48 is that each of the twenty-two sounds has a separate (blue) button, so the unit can be played in real time with just the fingers if that's what takes your fancy. In comparison, the MXR has twelve voice buttons, the DX eighteen buttons, and the Drumulator only six buttons, with a seventh to switch between conventional drums and the more exotic side of percussion. And with eight separate output channels, going via individual faders to separate outputs or a stereo or mono mix, straight digital input to those twenty-two buttons can make for a very impressive bit of bombast.
Like just about every other digital drum-machine around, the DPM-48 allows this sort of real-time input to be recorded for posterity, and provides a metronome click-track for that purpose. Incidentally, though the metronome level is fixed when it comes through on the mono or stereo mix, a separate output is also provided. A drum pattern is built up event by event, sound by sound, and once that's done, it can then be subjected to what Hammond call their 'automatic correcting system' a process whereby the timings of your manually-entered events are quantised to a greater or lesser extent depending on the minimum duration you choose to program into the machine.
As with other digital drum-machines, tracks are constructed by assembling Patterns into Songs. The yellow keys immediately under the LED display and down the right-hand side provide all the necessary programming functions. The DPM-48 allows 48 patterns to be stored, with up to 32 events in each. Precisely what those events are depends on (obviously) what drum sounds are used, but you also have to take into account how far you want to go with subdividing the crotchet. The options here are 1/2, 1/3, 1/4, 1/6, 1/8, 1/12, 1/16, 1/24 and 1/32, or, in other words, going from a plain and simple quaver to the sophistication of a demi-semiquaver. All this is accomplished by using the COUNT (sub-division of a crotchet) and LAST STEP (number of steps in a pattern) keys in conjunction with the grey numeric keys, and it's the combination of these that determines in what shape or form a real-time programmed pattern sees the light of day.
The alternative approach to entering drum events is step input, where the '<' and '>' keys are used to move through the steps of a pattern, and the appropriate combination of drum keys are pressed accordingly for each step. The ERASE key provides the wherewithal for scrubbing either Patterns, Songs, or individual voices by pressing the key in conjunction with other keys. Considering the fascination of the keyboard fraternity with non-real-time methods of input, it's curious that the only other sampling drum-machine to provide this lazy drummer's option is the Oberheim DX. Mind you, the 'quantisation' option on this and other drum-machines makes entering drum scores so straightforward as regards timing goofs and so on that it's quite understandable why most makers have stuck to their real-time guns.
Once a number of Patterns have been put together, these can then be turned into honest-to-goodness Songs by telling the machine which Pattern you want played for each of up to 256 sections. An INSERT key is also provided so that one Pattern can be changed in favour of another within a Song. The one big (well, fairly big) drawback is that the unit only has room for just three songs because of the limited memory in the eponymous RAM cartridge.
What's left of the front panel controls include the RUN/STOP key, REPEAT (which switches Pattern playing to the next highest numeric value - excellent for setting up segno/dal segno repeats, fill-ins, improvisations et al.), the coarse and fine tempo controls, and the mixed volume.
A mark of a good machine these days is the number of sockets on the back panel, and the DPM-48 gets good marks on that front (if you'll pardon the reversed pun). Aside from the separate 1/4" jack outputs for the metronome, two sets of toms, bass-drum, hi-hat, snare, cymbal, perc 1, perc 2, headphones, mono and stereo mix, there's also a 5-pin synchro in/out working on the 24 pulses-per-crotchet standard of Roland and Korg, and sockets for footswitching of the REPEAT and RUN/STOP functions.
In addition, there's a 12-pin 'D' connector marked 'Triggers', which presumably allows a micro to take control of the voices or the driving of the unit from drum pads. Mind you, the manual seems to turn a blind eye to this side of the DPM-48's interfacing potential.
One further point worth mentioning is the 'write protect' switch on the RAM pack. If you value your sanity, it's absolutely vital to remember what you're doing with this!
Overall, this is a superb digital drum-machine at a remarkably cheap price. Everything works as it should, it's easy to program, and the majority of the sounds are very good. In fact, it'd fit very happily into my own studio. What it could do with is a larger capacity RAM cartridge (using 6116s, for instance) and the option of different sets of EPROMs to suit different users. If the DPM-48 doesn't break Hammond into the rock market, there's no justice in the world...
Review by David Ellis
Previous article in this issue:
Next article in this issue:
mu:zines is the result of thousands of hours of effort, and will require many thousands more going forward to reach our goals of getting all this content online.
If you value this resource, you can support this project - it really helps!