First off, let me explain my music.  I write dark, experimental techno that draws heavily from my love for Detroit-influenced minimalist techno and electro.   I record in my home studio in a spare bedroom in my apartment in  New Hampshire.    My sound-generation gear is all hardware based, and about 75% of my sequencing is done on hardware sequencers.  My production style may be a little odd or "old school", but it seems to work very well for me.   

---

Sequencing:

     My main sequencer is an Alesis MMT-8.  This is an ancient old beast that many people seem to hate.  It doesn't have much power or memory, but it's very easy to use in live performances.  It works like an 8 track recorder with an on/off button for each track.  The MMT-8 controls my synths, sampler and a drum module.  Along side the MMT-8 I use a Roland R-5 Drum machine.  I use MIDI to synchronize the internal sequencer on the R-5 with the MMT-8 as a master.  I have it setup so that the R-5 triggers it's own internal sounds (I don't use it like a sound module with the MMT-8 sequencing it).

     When I  begin writing a new track, I generally record a drum track first.  It may begin with a cymbal or kick, or perhaps a sampled loop - just something to give me a good tempo or "groove" to start off with.  I generally create all of my kick sounds with my analog synth or an Alesis D4 drum module, and they are sequenced from the MMT-8.  Snares usually come from the D4, and I use both the R-5 and D4 for cymbals... All sounds from the D4 are sequenced on the MMT-8, and the R-5 sounds are triggered from it's own sequencer.  Finally, I may use my Yamaha A4000 sampler to add drum sounds; again, this is sequenced from the MMT-8.

   Soon after the drums are started, I like getting a melodic line written to set the overall feel of the song.  I may start with a bass line, or some background pads.  Just something to make it sound musical.  My synth sounds come from my analog, my sampler or an Alesis Nanosynth and all of them are sequenced by the MMT-8.  

    From this point, I continue to layer tracks until I fill all 8 tracks in the new pattern on the MMT-8.  After I complete a pattern, I move to a new pattern, and layer 8 new tracks.  I will usually copy important elements from the first track to the new one, and embellish or change them (so the song doesn't become too sterile).   By the end of the songwriting process, I'll have written a group of 10 or more patterns in the MMT-8 that work well together.   I use a similar process for filling up my R-5. I write groups of 5 to 10 drum patterns that go along with the patterns I wrote on the MMT-8.

     When I decide that I've written enough patterns, I'll perform my mix.  This is where I breakdown and actually use my computer.  I have a 550MHz Pentium III machine that I use to run Steinberg's Cubasis (A 'light' version of Cubase).  Again, the MMT-8 is the master clock generator, and the computer slaves to it.  Basically, I use Cubasis to record the output of my MMT-8.  I then perform the sequence "live" on the MMT-8.  I start it on the first pattern of the song, and allow the machine to loop.  I begin with only one or two tracks playing, but as I let the pattern continue to loop, I'll add more and subtract tracks to the mix.  As the song progresses, I'll scroll through the different patterns that I have written, and continue to arm and mute tracks to build the song.  I'll allow each pattern to play a few times before continuing to another.   This method of performing the tracks really gives me a lot of flexibility... I only have an approximate idea of what I want to do with the song up to this point.  By selecting the patterns in different orders, or muting the tracks differently I can create many unique mixes with the same dozen patterns.

    After I record the song, I'll listen to it and perform and corrections or additions in Cubasis.  I'll also add one-shot sound sound effects, vocal samples and other elements at this time.  Generally, this is also the stage where I add most of my sound modulations (filter resonance, volume and panning effects, etc).    Finally, when I'm happy with the song in Cubasis, I'll sync up the R-5 with the computer as a master.  I use the R-5's internal song sequencer to write a song out of the patterns I've written, so that it plays back in time with the MIDI from Cubasis. 

     By this point, the song is virtually finished.  All of my MIDI functions are recorded into either Cubasis or the R-5.  I'll then start up the mixdown deck (Minidisc or CD-R) and begin the playback of the sequences.  At this time, I add all of my tweaks to the analog synth sounds, as well as changing panning, volume and effects sends on my mixer.  These are all of the non-automated tasks that need to be done for the final mix.  These live tweaks can be hard to get right the first time, but that's one of the drawbacks of analog synths without patch memory...

Sound Generation:

     The key to writing techno is being as creative with your sound selection as you are with your pattern writing.  There are lots of good sounds in any of todays popular synths, but to really sound different, you need to alter your sounds to create new textures.  Always try to edit the stock parameters of the synth, and create your own unique patches.   Another important thing to note is that using a variety of instruments will add more depth to your sound than having all of your sounds coming from the same synth or sampler.  Each manufacturer seems to have their own unique sound and mixing them up will give you more interesting results.

     To give myself the most flexibility, I always use as many outputs as possible on my sound generators.  I use all 4 outputs of the D-4,  4 outputs on the R-5 and 4 outputs on the sampler.  The analog synth may be setup with one or more outputs, depending on the situation.  When using multiple outputs on a synth or drum machine, I use a pair of outputs as the main stereo out and they get connected directly to my mixer.  The remaining outputs are used more like effects sends.  Any patch that I want indepenently processed gets sent out it's own specific output and routed to the correct effects device.  The outputs of these effects devices are then sent to a mixer channel.

      I'm a big fan of using distortion boxes, multi effects processors and other gadgets to really warp the sounds coming from my various synths and drum machines.  I also use my analog synth as an 'effects box' by using the filters, VCAs and ring modulators on external sound sources.  A sampler is perhaps one of the best ways to really make crazy tones.  By pitch shifting, adding effects and resampling, you can alter the orignal sound beyond recognition.

    A great way to generate nice noises is by deliberately misusing your gear.  Try overdriving the input of an effects box by sending it a really hot signal.. some boxes will sound harsh and nasty, others will make a nice useable distortion.  Controlled feedback can also be fun.  I'll sometimes create effects chains by connecting  several stomp boxes and effecs units after each other.  By using effects sends on my mixer, I can route a small amount of the output of the effects chain back to the input.  After the chain starts feeding back a little, you can tweak the controls on each of the devices and "play" it like an instrument.  The result can then be either sampled, or performed realtime during the mixdown.  You can also send an external signal through this feedback chain and create spooky long sustaining sounds that eventually break into wild oscillation.  I recommend using a compressor at the end of this chain to keep your levels under control.