Originally recorded in 2005 and released in 2008 as an edition of 50 CD-Rs. This release expands the original with some additional related material.


I’ve sampled the drums from the PSS-380. This should also cover the PSS-280, 290 and 390 (these use the YM7116AA OPK and YM7129 OPK2 ICs). The drum samples are stored on the IC and use the onboard DAC. They’re not the best drums ever, and it’s a little ridiculous that there are Latin patterns, etc. using only 6 drum samples. The playback sample rate for the drums is 18.643 kHz. They can be played individually from the test mode (hold the highest B and C keys while switching it on), so the sampling process was straightforward. There’s also an accent, but I didn’t record the accented/unaccented sounds.

My latest acquisition is a Technics SX-K450 keyboard. I’ve made a multitrack recording and sampled the drum sounds.

The first line of Technics home keyboards consisted of the SX-K100 and K200 in 1982, followed by the K150 and K250 in 1984. The second generation included the SX-K300, K350 and K450 (1985-86), as well as the cheaper SX-K50. These all have sampled drums, although the cymbal and hihat sounds are still analog. I think this means that the K100 and K200 were the first home keyboards to use sampled drums. They may have first appeared in the SX-U90 in 1981. The K200 includes an analog solo section with a VCF. The K250 uses a sampled solo section that seems to have been introduced in the E series organs in 1983. The main tone generator is basically analog divide-down. All models have an excellent 3 BBD stereo string ensemble chorus with several modes, again derived from the full size Technics organs.

The main tone generator for the polyphonic section in the second generation models is the MSM5232, like the Korg Poly 800, SAS-20 and PSS-50. The master clock is 2 MHz, and this is fixed with no vibrato at all for the polyphonic voices. The older models may use a more conventional top octave synthesizer, but I haven’t seen any documentation for those. The cymbal and hihat sounds in the first and second generation models also seem to be different, but otherwise the later models have the same drums, chorus, and the sampled solo section from the K250. There’s a lowpass state variable VCF for the guitar, brass and accordion sounds, and a second one for the guitar sound in the accompaniment section. There’s also a monophonic bass section generated by an 8253 programmable divider. In total it’s 10 note polyphonic and 4 part multitimbral over MIDI. There’s an auto-accompaniment section and a 5 part sequencer with a very small memory. Strangely, both the drum and auto-accompaniment patterns have some amount of randomization that automatically varies the patterns. I haven’t encountered this in any other keyboard. The effect can be interesting, but on several of the patterns it sounds erratic and rather broken. The timing is also pretty sloppy compared to most keyboards.

The models worth looking out for are the SX-K200, K350, K450, and possibly also the K50. The K450 is the flagship model, but other than having a shorter keyboard the K350 doesn’t omit any notable features. The K100, K150 and K300 omit the solo section and various other features. The K50 is much simpler and may use very different hardware (I haven’t seen any documentation it).

The sampled drum section is based around the MSM6202 ADPCM IC. This is a general purpose IC made for playing back things like voice recordings or sound effects from an internal mask ROM. The process is described in detail in the 1987 OKI Voice Synthesis LSI Data Book, although the MSM6202 isn’t included here since it seems to have been discontinued by the time of the data book’s publication. ADPCM results in fairly low quality sound, since it compresses a waveform with 8-12 bits of dynamic range into 3 or 4 bits, but using this IC to play sampled drums is cheap, simple and actually very clever. Yamaha introduced sampled drums in their organs and portable keyboards in 1983, but they didn’t have a comparably simple single IC implementation until about 1988 (this IC, the YM3419 RYP7, seemed to use ADPCM as well).

The MSM6202 has 16 different drum samples and probably 4 channels. The clock is 62.5 kHz and output sample rate is 15.625 kHz (4 clocks per sample). It seems to store about 1.4 seconds of waveform data (roughly 22.5 kSamples). There’s no accent, and no ability to transpose, scale or envelope the sounds in any way. The onboard DAC has two outputs for the positive and negative portions of the waveform. An external differential amplifier is used to reconstruct the complete waveform. I think this is done to make a sign-magnitude DAC with a very simple implementation using a single-ended power supply and no virtual ground.

The analog cymbals and hihats use 6 square wave oscillators, similar to the Roland TR-808 and other drum machines. There’s an accent here too, although it only seems to be used on the hihat. The first generation Technics keyboards sound somewhat different, similar to the SX-U90. I suspect these use a different design, possibly mixing all 12 outputs from a top octave synthesizer to obtain the metallic noise (this was commonly done in organs in the 1970s).

The output of the SX-K450 is too noisy to sample directly, and anyway the cymbals and sampled sounds overlap, so the individual sounds can’t be isolated. I made some temporary modifications so that the outputs of the ADPCM IC and cymbal/hihat section could be recorded directly. Several of the sampled sounds are still never played in isolation, so I used an Octave script to separate them. This will be described in another post.

The sounds are all sampled at 96 kHz. There are 8 samples each of the cymbal and hihat sounds for round-robin sampling. I think the ADPCM sampled sounds in particular are excellent. They’re extremely compressed and sharply truncated.

Another new album, again everything is Pure Data, 2013-2017.

It seems to be common for people to lose the power supplies for the Seiko DS-101, DS-202 and DS-250. They use a DIN connector with no pinout indicated on the keyboard itself, so there’s no readily available replacement, or even an obvious way to build one. But they’re not really special or irreplaceable. It’s just an ordinary 13.8 VDC, 1000 mA unregulated supply with a 4 pin DIN plug. Pins 1 and 2 are positive and pins 3 and 4 are negative. At least on the DS-250, the positive terminals are toward the rear of the instrument. There’s a photo of the power supply here, including the pinout.

There are several possibilities for replacement. The best option might be to replace the DIN jack on the keyboard with an ordinary barrel connector so that it doesn’t need a special power supply. I don’t think it’s really possible to solder a barrel jack in place, but it would work well enough to use a female barrel connector with a lead. It would similarly work to attach a DIN plug to an ordinary adapter. Here’s the DS-250’s jack board with the DIN jack highlighted and the terminals labeled:


There’s also a simpler workaround: Alesis and Digitech used the same 4 pin DIN plug, with the upper and lower pins on each side connected together in the same way. Alesis and Digitech used 9VAC power supplies, so these can’t be used with the Seiko, but there are adapter cables available to convert a standard barrel connector to a 4 pin DIN plug. These should work for the Seiko, but I don’t know how this adapter is wired internally, so I don’t know what polarity the input power should be.

This material is from 2012 to 2016, made entirely in Pure Data.

I’ve got a large backlog of material that I’d like to see released. I’m going to try to release one album every two months until I’ve exhausted it all. I also realize that I don’t have a clear and complete discography anywhere that collects everything in one place. So, to give an idea of both past releases and the plan for things to come, here’s an overview:

Things that will never come out:
Blue Tape (1999)
Yellow Tape (1999)
Pink Tape (1999)
Black Tape (1999)
Untitled CD-R (2000)

Early netlabel releases:
shapes with and without corners (2001-2002)
blue music (2002-2003)
gray music (2002-2003)

Things that I might release/re-release eventually:
saccades part 1 (2003)
saccades part 2 (2004)
saccades part 3 (2004)
saccades part 4 (2004)
saccades part 5 (2005)
TBD compilation (2004-2007?)
TBD compilation (2001-2007?)
TBD (2008-2009)

Current and future releases:
eisegetic waiver (2005) [released 2008, revised and re-released 2017-09]
airburst parcel (2007) [released 2013]
impulse, vagaries and travel (2009-2010?) [to be released 2018-09]
sounder/resounder (2010-2012?) [to be released 2018-07]
agglomeration and homothety (2011) [released 2017-03]
crepitus weathering (2012) [released 2012]
flutter straight on the nine miles (2012) [released 2013]
hg iterated glass entering 3e2 (2013) [released 2013]
acheiropoietic ansätze (2012-2016) [released 2017-05]
intangible pavilions (2014-2017) [released 2017-07]
TBD (2016-2017) [to be released 2017-10?]
adventitious auscultation (2010-2017?) [to be released 2018-05]
wave and lattice (2013-2017) [to be released 2018-01]
TBD [to be released 2018-03]
aleatoric aubades (2012-2017?) [to be released 2017-11]

Not really an album:
miscellaneous synth demos (2004-2012) [released 2012]