It became the main source of inspiration for my songs.
November 5, 2020 8:15 AM Subscribe
Why I Painstakingly Built My Own Analog Drum Machine. Reverb.com interviews Sam Dust aka LA Priest on the years-long process of creating a drum machine from scratch, which served as both the rhythmic center of his new album's songs and its namesake. He named it Gene.
"As far as I know, it’s the only drum machine that you can change the timing of each beat down to the millisecond. It also uses 100% analog circuitry, which is sort of rare, and each one is built by hand, actually by my hands. There are lots of other things that are completely unique in its design: Every part of its circuitry was trial-and-error, so it ended up with some really unusual things going on in there. It's pretty hard to explain how some of it even works."
"As far as I know, it’s the only drum machine that you can change the timing of each beat down to the millisecond. It also uses 100% analog circuitry, which is sort of rare, and each one is built by hand, actually by my hands. There are lots of other things that are completely unique in its design: Every part of its circuitry was trial-and-error, so it ended up with some really unusual things going on in there. It's pretty hard to explain how some of it even works."
Analog drum machines are pretty incredible and it's difficult to describe why to people who are outsiders to synth stuff, and why it's so different than a sampled drum machine or even an FM synth machine.
Back in the 90s I knew someone with a Sequential Circuits analog drum machine that stored the pads/patches on an analog EEPROM or maybe plain old PROM chip and they did this crazy thing where they spent a long time crafting and recording their own analog drum pads created on a Sequntial Circuits Pro One analog synth, which they then burned on to a custom PROM chip, basically re-inventing the sounds of the original drum machine.
Since the waveforms are stored as analog impulse waveforms and not samples they can be tuned and stretched in ways that digital samples and FM synth drum machines simply could not do without significant distortion.
The bass kicks and sounds from that customized drum machine were just so full and so real with so much dynamic range and it was a fundamental part of his live PA act as well as studio recordings.
The major downside was that the drum machine was capable of putting out so much unfiltered bass that it would reliably brown out and pop breakers on the electrical systems at music venues to the point that it came to be kind of a known thing and quantity for their live hardware PA shows.
I remember seeing the at outdoor renegade parties run on generators and the load from the bass was so high that the electrical load on the generator would make it chug along to the bass kicks to the point you could hear the generator/alternator set totally bogging down and struggling with each bass kick in time with the music, like a poor, tired little robot going "Ok, that's heavy lifting. What the hell is going on here!?" and complaining about it.
It became like a running inside joke that at almost every show when they took the stage they might get the first bass kick or two out and everything would shut down, and since this was all early MIDI hardware live PA stuff sometimes this meant that they would lose all of their song patterns stored in a hardware sequencer and basically have to start live, from scratch, right there on stage and just rebuild tracks as fast as they could while playing and improvising.
These unplanned improv sets were some of the best shows for house/techno style music I've ever seen, heard or felt. People would absolutely lose their minds at these shows because the sound quality was just that good, that bassy and that loud.
I would trade almost anything up to donating body parts to see that live hardware PA jam happen on a modern dance/festival sound system like a Funktion One, KV2 or VOID system. All we had back then for bass bins were things like the Cerwin Vega L-36 Earthquakes or DIY W-bins that everyone was using back then.
We didn't have fancy DSPs and stuff like cardiod bass arrays or other bass music specific sound system tools back then. Hearing/feeling a mostly analog and hardware MIDI set on these kinds of modern speakers would blow your shirt right off your back.
posted by loquacious at 9:36 AM on November 5, 2020 [15 favorites]
Back in the 90s I knew someone with a Sequential Circuits analog drum machine that stored the pads/patches on an analog EEPROM or maybe plain old PROM chip and they did this crazy thing where they spent a long time crafting and recording their own analog drum pads created on a Sequntial Circuits Pro One analog synth, which they then burned on to a custom PROM chip, basically re-inventing the sounds of the original drum machine.
Since the waveforms are stored as analog impulse waveforms and not samples they can be tuned and stretched in ways that digital samples and FM synth drum machines simply could not do without significant distortion.
The bass kicks and sounds from that customized drum machine were just so full and so real with so much dynamic range and it was a fundamental part of his live PA act as well as studio recordings.
The major downside was that the drum machine was capable of putting out so much unfiltered bass that it would reliably brown out and pop breakers on the electrical systems at music venues to the point that it came to be kind of a known thing and quantity for their live hardware PA shows.
I remember seeing the at outdoor renegade parties run on generators and the load from the bass was so high that the electrical load on the generator would make it chug along to the bass kicks to the point you could hear the generator/alternator set totally bogging down and struggling with each bass kick in time with the music, like a poor, tired little robot going "Ok, that's heavy lifting. What the hell is going on here!?" and complaining about it.
It became like a running inside joke that at almost every show when they took the stage they might get the first bass kick or two out and everything would shut down, and since this was all early MIDI hardware live PA stuff sometimes this meant that they would lose all of their song patterns stored in a hardware sequencer and basically have to start live, from scratch, right there on stage and just rebuild tracks as fast as they could while playing and improvising.
These unplanned improv sets were some of the best shows for house/techno style music I've ever seen, heard or felt. People would absolutely lose their minds at these shows because the sound quality was just that good, that bassy and that loud.
I would trade almost anything up to donating body parts to see that live hardware PA jam happen on a modern dance/festival sound system like a Funktion One, KV2 or VOID system. All we had back then for bass bins were things like the Cerwin Vega L-36 Earthquakes or DIY W-bins that everyone was using back then.
We didn't have fancy DSPs and stuff like cardiod bass arrays or other bass music specific sound system tools back then. Hearing/feeling a mostly analog and hardware MIDI set on these kinds of modern speakers would blow your shirt right off your back.
posted by loquacious at 9:36 AM on November 5, 2020 [15 favorites]
Apologies if this should be obvious, but how are analog waveforms stored on digital chips without being written in discrete chunks (what I would think of as "samples")? When I think of analog, I think of things like magnetic tape or film.
posted by treepour at 10:36 AM on November 5, 2020
posted by treepour at 10:36 AM on November 5, 2020
Neat! ("Gene" for Mr. [Sing Sing Sing; Drum Boogie] Krupa, I hope.)
posted by Iris Gambol at 10:38 AM on November 5, 2020
posted by Iris Gambol at 10:38 AM on November 5, 2020
@treepour filling in some educated guesses here, but I assume the device was the sequential drumtraks, which like other samplers of that era ( e-mu sp-1200, akai s900, mpc60, etc.), audio was stored with 12 bits per sample, and that lower fidelity combined with usually an analog filter led to a characteristic lo-fi sound rather than a pristine reproduction - as well as different harmonic content when filtered and processed than a more faithful source.
posted by thedaniel at 10:45 AM on November 5, 2020 [1 favorite]
posted by thedaniel at 10:45 AM on November 5, 2020 [1 favorite]
Apologies if this should be obvious, but how are analog waveforms stored on digital chips without being written in discrete chunks (what I would think of as "samples")?
Not at all!
So, not all integrated circuits (chips) are digital. A plain transistor-based oscillator circuit is analog, and an integrated circuits can be nothing but basic passive electronic elements.
I'm not sure about the specifics of the drum machine or what kind of PROMs it used but I would imagine it's just an array of passive components that can permanently or semi-permanently store an analog value when burned in by a PROM programming unit.
Another way to think about this is that even all of our digital devices at the base component level are all analog to begin with, the "digital" part doesn't happen until you arrange them in such a way that they only accept threshold voltage values, which is why digital circuits were invented - to create (theoretically) more reliable/accurate circuits that allow for natural variations in values between, say, two individual transistors in the same circuit or even etched on the same integrated chip.
So when one active transistor is outputting 4.95 volts and the other one is outputting 5.05 volts, the digital circuit just reads both of them as 5 volts or a "1" as opposed to a "O", or vice versa depending on the circuit.
When a general purpose digital CPU or other "VLSA" type integrated circuit is churning through a lot of bits you can actually see this analog behavior if you plot and capture the voltage waveforms on a good oscilloscope and digital logic probe - it will show you and map the variations in voltages. In theory a digital circuit strives for square waveforms in voltage rise and fall, but in reality there's a slope over time of that voltage rise (and fall) to the accepted threshold level and it's not putting out perfectly square, instantaneous voltage changes. It's generally actually more like a trapezoid or rhombus in shape and the peak voltage isn't really flat, either, but has a lot of small wobbles and variations in voltage for the duration that it's "on" and passing through the output voltage.
You can see this analog variation between individual transistors in action when plotting response curves of bias/threshold control voltages and outputs. Really dodgy/cheap transistors can have a wide array of values and responses even within the same parts and lot numbers.
As I remember it, the musician in question had to burn through and trash a number of non-reprogrammable chips to finally get one with the values he wanted and could use. I'm not sure what the interstitial recording media was because it wasn't likely that he had his analog synth directly hooked up to a PROM programming unit, he probably had to record it to a high speed/quality analog tape deck first, compile and edit them to specific lengths and then probably had to sequence them as a series of tones/sounds with specific timing information or some kind of stepped programming control. This person also repaired audio equipment for a living at the time and had access to a lab for hifi equipment and electronic instrument repair.
I also recall that once he got it right he made a number of direct copies of this custom PROM chip so he had a bunch of backups in case the one he installed in the drum machine failed for some reason so he didn't have to go through the initial programming part all over again.
posted by loquacious at 11:04 AM on November 5, 2020 [4 favorites]
Not at all!
So, not all integrated circuits (chips) are digital. A plain transistor-based oscillator circuit is analog, and an integrated circuits can be nothing but basic passive electronic elements.
I'm not sure about the specifics of the drum machine or what kind of PROMs it used but I would imagine it's just an array of passive components that can permanently or semi-permanently store an analog value when burned in by a PROM programming unit.
Another way to think about this is that even all of our digital devices at the base component level are all analog to begin with, the "digital" part doesn't happen until you arrange them in such a way that they only accept threshold voltage values, which is why digital circuits were invented - to create (theoretically) more reliable/accurate circuits that allow for natural variations in values between, say, two individual transistors in the same circuit or even etched on the same integrated chip.
So when one active transistor is outputting 4.95 volts and the other one is outputting 5.05 volts, the digital circuit just reads both of them as 5 volts or a "1" as opposed to a "O", or vice versa depending on the circuit.
When a general purpose digital CPU or other "VLSA" type integrated circuit is churning through a lot of bits you can actually see this analog behavior if you plot and capture the voltage waveforms on a good oscilloscope and digital logic probe - it will show you and map the variations in voltages. In theory a digital circuit strives for square waveforms in voltage rise and fall, but in reality there's a slope over time of that voltage rise (and fall) to the accepted threshold level and it's not putting out perfectly square, instantaneous voltage changes. It's generally actually more like a trapezoid or rhombus in shape and the peak voltage isn't really flat, either, but has a lot of small wobbles and variations in voltage for the duration that it's "on" and passing through the output voltage.
You can see this analog variation between individual transistors in action when plotting response curves of bias/threshold control voltages and outputs. Really dodgy/cheap transistors can have a wide array of values and responses even within the same parts and lot numbers.
As I remember it, the musician in question had to burn through and trash a number of non-reprogrammable chips to finally get one with the values he wanted and could use. I'm not sure what the interstitial recording media was because it wasn't likely that he had his analog synth directly hooked up to a PROM programming unit, he probably had to record it to a high speed/quality analog tape deck first, compile and edit them to specific lengths and then probably had to sequence them as a series of tones/sounds with specific timing information or some kind of stepped programming control. This person also repaired audio equipment for a living at the time and had access to a lab for hifi equipment and electronic instrument repair.
I also recall that once he got it right he made a number of direct copies of this custom PROM chip so he had a bunch of backups in case the one he installed in the drum machine failed for some reason so he didn't have to go through the initial programming part all over again.
posted by loquacious at 11:04 AM on November 5, 2020 [4 favorites]
I made one of the first 100 x0xb0x kit clones of the analog classic TB-303 back in the early 00s. The designers went to great pains to recreate the original ("original transistors were analysed using big expensive curve-tracers to determine their characteristics") and it showed. But there's tons of good 303 clones now, apparently. Anyway it was great fun to build - I really should make use of the thing.
posted by exogenous at 12:20 PM on November 5, 2020 [1 favorite]
posted by exogenous at 12:20 PM on November 5, 2020 [1 favorite]
Julian Koster has a drum machine that's all cams, pulley and motors. Get the hammer and wrench out if you need to reprogram it, though.
posted by scruss at 12:31 PM on November 5, 2020 [1 favorite]
posted by scruss at 12:31 PM on November 5, 2020 [1 favorite]
I'm not sure about the specifics of the drum machine or what kind of PROMs it used but I would imagine it's just an array of passive components that can permanently or semi-permanently store an analog value when burned in by a PROM programming unit.
My idle speculation is that the drum machine probably consisted of a collection of analogue circuits (oscillators, filters, etc.) that could be connected or not depending on whether a wire had electricity flowing through it. The PROM's output pins would have been connected to these wires so a 1 or a 0 at in a particular byte in the PROM would determine what was connected to what. The front panel switch and/or a simple digital counter would generate the input address, resulting in a preset sound. Or a custom one if you replaced the PROM.
posted by suetanvil at 12:37 PM on November 5, 2020 [1 favorite]
My idle speculation is that the drum machine probably consisted of a collection of analogue circuits (oscillators, filters, etc.) that could be connected or not depending on whether a wire had electricity flowing through it. The PROM's output pins would have been connected to these wires so a 1 or a 0 at in a particular byte in the PROM would determine what was connected to what. The front panel switch and/or a simple digital counter would generate the input address, resulting in a preset sound. Or a custom one if you replaced the PROM.
posted by suetanvil at 12:37 PM on November 5, 2020 [1 favorite]
That is absolutely fascinating loquacious, thank you! And thank you too thedaniel and suetanvil
posted by treepour at 2:01 PM on November 5, 2020
posted by treepour at 2:01 PM on November 5, 2020
> "As far as I know, it’s the only drum machine that you can change the timing of each beat down to the millisecond."
Elektron's Analog Rytm can shift individual notes around by as little as 1/384th of a beat which is close enough to a millisecond. Totally not a reason to keep from building your own drum machine of course. Gene sounds awesome!
posted by technodelic at 2:06 PM on November 5, 2020
Elektron's Analog Rytm can shift individual notes around by as little as 1/384th of a beat which is close enough to a millisecond. Totally not a reason to keep from building your own drum machine of course. Gene sounds awesome!
posted by technodelic at 2:06 PM on November 5, 2020
My idle speculation is that the drum machine probably consisted of a collection of analogue circuits (oscillators, filters, etc.) that could be connected or not depending on whether a wire had electricity flowing through it.
I know this exists for a lot of synths that otherwise use analog oscillators, and this is called a DCO, a digitally controlled oscillator. One famous example of this is the MC-303 Groovebox, a reboot attempt of the 303 that was a sort of early hybrid drum machine, synth an sequencer toy in a box.
That's also the basic concept of a modular synth that uses a patch bay or jumperboard. You "progam" the arrangement of oscillators and modifiers by how you wire them together.
This doesn't really explain how my friend used his Pro One (not a drum machine) to design his own pads for the other device that was solely a drum machine, and as it was originally explained to me somehow the values of the waveforms or transients were actually on the PROM device in the as some sort of analog physical arrangement.
On the other hand you could use a synth to map out oscillator/modifier networks, write down the rough values of those settings and use a patch matrix as you're describing to control the logic of an analog synthesizer, but again as it was explained to me this is not what that chip did.
I'm deep into hand-waving territory at this point.
Still haven't ever heard another drum machine put out that kind of bass and sound. It was magic in a box.
posted by loquacious at 2:28 PM on November 5, 2020 [1 favorite]
I know this exists for a lot of synths that otherwise use analog oscillators, and this is called a DCO, a digitally controlled oscillator. One famous example of this is the MC-303 Groovebox, a reboot attempt of the 303 that was a sort of early hybrid drum machine, synth an sequencer toy in a box.
That's also the basic concept of a modular synth that uses a patch bay or jumperboard. You "progam" the arrangement of oscillators and modifiers by how you wire them together.
This doesn't really explain how my friend used his Pro One (not a drum machine) to design his own pads for the other device that was solely a drum machine, and as it was originally explained to me somehow the values of the waveforms or transients were actually on the PROM device in the as some sort of analog physical arrangement.
On the other hand you could use a synth to map out oscillator/modifier networks, write down the rough values of those settings and use a patch matrix as you're describing to control the logic of an analog synthesizer, but again as it was explained to me this is not what that chip did.
I'm deep into hand-waving territory at this point.
Still haven't ever heard another drum machine put out that kind of bass and sound. It was magic in a box.
posted by loquacious at 2:28 PM on November 5, 2020 [1 favorite]
The Sequential Drumtraks is 8-bit digital like the Casio RZ-1, so the EEPROMs have digitized waveforms. You can do some great stuff with digital sound sources and analog filters, as Sequential did with the Six Trak, which from the name I would assume was a sibling to this box and thus surely part of a product strategy. Reprogramming the EEPROMs is indeed essentially sampling, though with the worst possible UI. Reasonably usable sampling drum machines were still a couplefew years off.
posted by rhizome at 2:32 PM on November 5, 2020 [1 favorite]
posted by rhizome at 2:32 PM on November 5, 2020 [1 favorite]
This doesn't really explain how my friend used his Pro One (not a drum machine) to design his own pads for the other device that was solely a drum machine, and as it was originally explained to me somehow the values of the waveforms or transients were actually on the PROM device in the as some sort of analog physical arrangement.
The PROMs used in the Drumtraks were interoperable with the Linn Drum and the Oberheim DX & DMX, which I'm sure is what was being exploited even if unofficial. As I learned just now, there were apparenly only a handful of formats (via), so there were other groups of instruments with interchangeable chips using other standards.
posted by rhizome at 2:36 PM on November 5, 2020 [1 favorite]
The PROMs used in the Drumtraks were interoperable with the Linn Drum and the Oberheim DX & DMX, which I'm sure is what was being exploited even if unofficial. As I learned just now, there were apparenly only a handful of formats (via), so there were other groups of instruments with interchangeable chips using other standards.
posted by rhizome at 2:36 PM on November 5, 2020 [1 favorite]
Gene Gene the Drumming Machine?
posted by Saxon Kane at 3:18 PM on November 5, 2020 [1 favorite]
posted by Saxon Kane at 3:18 PM on November 5, 2020 [1 favorite]
but I assume the device was the sequential drumtraks, which like other samplers of that era ( e-mu sp-1200, akai s900, mpc60, etc.), audio was stored with 12 bits per sample, and that lower fidelity combined with usually an analog filter led to a characteristic lo-fi sound
That's still a digital / sample based drum machine, though...DSI *did* make an analog drum machine, the tempest, but that was much later. An analog drum machine, like the 808 would ususally basically be a synth, but with control data / patches to control the synth to approximate toms, kicks etc - pitching down/up the control parameters is sort of analogous to scaling vector graphics vs scaling a bitmap.
posted by Jon Mitchell at 8:01 PM on November 5, 2020
That's still a digital / sample based drum machine, though...DSI *did* make an analog drum machine, the tempest, but that was much later. An analog drum machine, like the 808 would ususally basically be a synth, but with control data / patches to control the synth to approximate toms, kicks etc - pitching down/up the control parameters is sort of analogous to scaling vector graphics vs scaling a bitmap.
posted by Jon Mitchell at 8:01 PM on November 5, 2020
I love that there's (of course) a nexus of synth nerdery here at metafilter--I guess I can recruit Ask in my endless "what analog polysynth to buy" question now too!
posted by thedaniel at 12:08 AM on November 6, 2020 [1 favorite]
posted by thedaniel at 12:08 AM on November 6, 2020 [1 favorite]
my endless "what analog polysynth to buy" question now too!
Pretty sure the answer is "all of them."
posted by soundguy99 at 4:47 AM on November 6, 2020 [1 favorite]
Pretty sure the answer is "all of them."
posted by soundguy99 at 4:47 AM on November 6, 2020 [1 favorite]
For me the answer is “a Juno but 20 years ago” :/
posted by thedaniel at 3:14 AM on November 8, 2020 [1 favorite]
posted by thedaniel at 3:14 AM on November 8, 2020 [1 favorite]
« Older 1.d4 d5 2.c4 | Presentations on fictional naval forces Newer »
This thread has been archived and is closed to new comments
posted by Beardman at 8:33 AM on November 5, 2020 [1 favorite]