Billy Cobham & George Duke giving it welly. Hell yeah!
And there’s always…
Billy Cobham & George Duke giving it welly. Hell yeah!
And there’s always…
Awesome performance from Rory. I remember this being beamed live across Europe on TV and Radio in 1977. I was five and my brothers moved the stereo (a precious and rare thing in the west of Ireland back then) into the living room.
I had a scratchy cassette tape of this show for years – wore it out. Happily it was made available on DVD a few years back.
Still hoping to make the Rory Festival in Ballyshannon – would love to see Wilko Jonson in action.
Wheeeeeeeee! I have some money again, though not a lot of time, but I’m determined to finish and further expand the projects I started with such gusto last year.
Both the Tempest and FS1r are blurred memories, and I’ve been getting my kicks, whenever possible, with a Monomachine/Blofeld Combo.
The great thing about this pairing is that I can sequence 6 channels of Blofeld multimode goodness alongside 5 Monomachine tracks, with the 6th track acting as the FX machine for the Blofeld, which is routed back through the MnM inputs. A nice, self contained, and gloriously digital setup. Did I mention that MnM does killer drum synthesis?
On the the MnM it is possible to input chords using an external MIDI keyboard for the Arps and external sequencing.
I have an cheapo Akai LPK USB mid keyboard that would make a nice compliment to this mini setup. But MnM only accepts 5-pin DIN MIDI. One existing option is the Kenton USB MIDI Host, but that goes for over 100 euro.
I got to thinking that I could use my similarly-abandoned and superseded Raspberry Pi 1, and a MIDI 1×1 USB MIDI cable that never got used because it kept causing bluescreens on my Windows PC. What if I could use the Pi to route the output of the LPK through the 1×1?
Well yes, it’s possible, works perfectly, and is really easy to get running.
Any flavour of linux will do, in my case I’m using a Raspberry Pi v1 with an optimised Raspbian Wheezy image I downloaded from here. I’ve also got this to work on a Rpi2 using the official Ubuntu ARMv7 distro.
The instructions for both are the same.
Obviously, with only 2 USB ports on the Pi v1, there is no room for wireless, so I needed to login over ethernet.
Once a command prompt is available, it’s a matter of installing Alsa:
sudo apt-get install alsa alsa-utils
Now connect the *class compliant* MIDI devices, in my case the Akai LPK25 and E-mu USB MIDI 1×1.
To show all connected MIDI devices:
sudo amidi -l
Show connection status and port numbers of connected MIDI devices
sudo aconnect -i -o
Look for the device ID, which is in the format x:0. In my case, the LPK25 was 20:0 and the Emu 1×1 was 16:0. So to connect the output of the LPK to the output of the Emu, just go:
sudo aconnect 20:0 16:0
…and voila! Works a treat here, no latency and I’ve sent boatloads of MIDI through it.
To dump all midi message to the screen,
sudo amidi -d
Naturally, we will want this connection to happen automatically every time we start the Pi. Of the several ways to do this, I opted for the laziest, which was to make a root crontab.
If you’re not root already,
at the end of the file, enter the aconnect command that works for you to run at reboot, e.g.
@reboot aconnect 20:0 16:0
Sunday morning Youtube trawls FTW!
Here’s Bjorklund. He’s been dropped, stood-on and attacked by cats, yet survives. I was pissed drunk when I made this but hopefully it will provide a first inkling of the little fella’s potential:
Euclid’s algorithm (circa 300 BC) describes a method for computing the greatest common divisor (GCD) of two integers. It is one of the oldest numerical algorithms still in common use.
In 2003 Eric Bjorklund extended the Euclidean algorithm to address timing issues in Neutron Accelerators. He wanted to solve the following problem:
‘distribute n pulses over m “timing slots” in the most even way possible, even though n may not necessarily be an even divisor of m.’
In 2004, Godfried Toussaint demonstrated that the resulting binary patterns mirrored many familiar ostinatos (repeating rhythmic phrases) heard in diverse range of musical styles.
African rhythms are well represented and, naturally, have since appeared in South American music, modern jazz, pop, rock and dance.
Bjorklund is a Bjorklundean sequencer.
8 independent tracks, each with their own:
– Tracklength of 1- 64 steps
– 1-64 Pulses, feeding into the algorithm.
– Full control over current playback position (rotation).
– Clock dividers (Whole note to 1/32 ).
– Random Velocity offset – Humanisation (+/-)
– Accented notes every 1-16 steps (max velocity)
– Assignable MIDI Note.
– Assignable MIDI Channel.
– Randomization per track or all tracks.
Fine control over Master BPM (+/- 3, 1, 0.1).
Next version will bump-up to Teensy 3.1 and will feature…
Groove – Tick Quantization and Tick shift
Extensive preset management and fast switching
Apologies for the shitty video quality. But does it matter?
I’ve got the beer on ice, so tonight I will make a video demonstration of the finished item, and it has far surpassed my expectations!🙂
Some quick snippets…
But first, here’s a catch-up on construction and materials used. Rather than explain in minute detail each step, I will for now just point to the resources that provided the most direct and useful information. They are the nuggets gleaned from numerous trawls through crappy/broken code and bad schematics.
I keep the power supply on a separate board. It’s a common diode-protected, RC filtered 5v regulated design…
MIDI output is easy-peasy…(but remember to map the pins of the ATMega)
The LED matrices are just too big for the breadboard, so I propped-them up on stackable headers to allow me access to the pins. I’m using 2 x MAX7219 driver chips. Connecting these to the ATMega and to each other was a breeze, however wiring the matrices themselves was a little tricky. I mostly followed the instructions here, however this wiring was not correct for my particular LED matrix (common cathode – from Tayda). I found the wiring diagram here to work for me.
The encoders were connected and via the MCP23017 I/O port expander. I got these encoders from Reichelt – they are quite stiff and they have been problematic.
Firstly they wouldn’t stay on the breadboard (legs too short!) so I made a breakout. Those funny protrustions are my attempt at hardware ‘debouncing’. I soldered sockets to the pins so that I could experiment with different capacitor values. In the end, 100n capacitors connected between each pin and common ground were best, although the encoders were still somewhat jumpy. I used a healthy dose of contact cleaner on each and this helped significantly. These encoders need to be ‘broken-in’.
In the code I am polling these controls according to this helpful post. This works well but is quite the resource drain. To circumvent this I tried using interrupts, however they proved even more problematic because they were detecting even the slightest tap of the encoders. Also, setting-up interrupts via the MCP23017 is no trivial task. I managed to fry two chips for my troubles. Interrupts are for another day….
In the next version I will abandon the mechanical encoders in favour of optical. I think the additional cost will be more than compensated by time spent, in frustration, fixing a problem that is beyond my control.
Finally there is a RGB LED which provides instant access to 7 colours as menu indicators. I’ve coded a state machine that allows me to turn 4 encoders into 40 controls, depending on what sequence of buttons is pressed. The mode is indicated by the single-or-mixed colours of the RGB. Thus there is always a clear visual indication of which menu is currently in use.
Overall, a fairly small parts count, easily obtained. Here is the Bill of Materials and sources:
1 x ATMega 328-PU
1 x 16MHz Quartz Crystal
2 x 22n Ceramic Capacitors
3 x 47K Resistors
1 x MCP23017 I/O Port Expander
2 x 8×8 Dot Matrix LED display – Common Cathode
2 x MAX7219 LED Display Drivers (Tayda’s are much cheaper, but apparently counterfeit. They work!)
4 x 24/24 Encoders (optical would be sooo much nicer)
10 x 100n Ceramic Capacitors
2 x 10uf Electrolytic Capacitors
1 x RGB LED 5mm Common Anode
3 x 220 Ohm Resistors
1 x 27K Resistor
Lots of jumper wire!
1 x MID DIN Socket
1 x 220 Ohm resistor
When I finish the code I will publish it, along with a proper schematic (hey, I’m still learning).
Read it and weep. I know I am😛
This was the most challenging piece of code to date. Outwardly simple, and with many existing examples in other languages, not to mention the C code example in Bjorklund’s paper, it should have been a doddle to implement on the Arduino, right?
But no. I realised the big downside to using the AVRISP: no access to the serial interface, which makes debugging impossible. Thankfully I could use the Teensy 3.1 to test my Code. But this in itself brought heartache, as there were some library conflicts with Teensyarduino. A troubling story to be sure, but In brief, I battled through ’til dawn and eventually cracked the bastard.
My overriding concern was – how much of the ATmega’s CPU time would be taken-up by using recursive function? I would have expected this to be a resource hog, and I worried that it wouldn’t handle 8 tracks simultaneously, slowing everything down. I’m happy with the current state of the MIDI clock (99.98% accurate), and this was something I didn’t want to compromise. I also witnessed some talk on the forums about Arduino’s limited stack size.
In the end, performance wildly exceeded my expectations. See the example outputs below.
Note that I have reversed the order or the patterns in alignment with the examples in Toussaint’s paper. See chapter 4 for many more examples. An interesting read for sure.
In the end, the pattern directions don’t matter, because I’ve implemented track rotation and track reverse in the main application. So…
For 8 steps and 3 pulses, the pattern is:
Bjorklund took 8.00 microseconds.
Cuban tresillo / Habanera rhythm. “It can often be heard in early rock-and-roll hits in the left-hand patterns of the piano, or played on the string bass or saxophone. A good example is the bass rhythm in Elvis Presley’s Hound Dog.”
The tresillo pattern is also found widely in West African traditional music. For example, it is played on the atoke bell in the Sohu, an Ewe dance from Ghana. The tresillo can also be recognized as the first bar of the ubiquitous two-bar clave Son given by [x . . x . . x . . . x . x . . .].
For 8 steps and 5 pulses, the pattern is:
Bjorklund took 10.00 microseconds.
Cuban cinquillo. Used extensively in jazz, 50’s rockabilly and West African traditional music. e.g hand-clapping pattern ‘Hound Dog’.
For 12 steps and 4 pulses, the pattern is:
Bjorklund took 10.00 microseconds.
The 12/8-time Fandango clapping pattern. (I know, I know)
For 3 steps and 2 pulses, the pattern is:
Bjorklund took 5.00 microseconds.
E(2,3) = [x . x] is a common Afro-Cuban drum pattern. For example, it is the conga rhythm of the (6/8)-time Swing Tumbao. It is also common in Latin American music, as for example in the Cueca.
Lastly, a silly example to check performance again:
For 83 steps and 56 pulses, the pattern is:
Bjorklund took 63.00 microseconds.
This is going to be damn sweet.
I’m usually the one to grumble when I come across all those meaningless, grubby 20-second videos on Youtube. But today I’m adding my own. What the hell.
Here is Bjorklund just come to life:
At this moment I have a tight MIDI clock running on the ATMega328 at 24ppqn, with the option to go to a higher resolution if and when I trade-up to a better processor(Teensy 3.1 here I come!).
The 4 encoders and buttons are multiplexed using a single MCP2317, and the LED matrices are driven by 2 MAX7219s (both apparently counterfeit but working perfectly).
What you see are the available tempo divisions from slowest (each quarter note) to fastest (each MIDI tick) flashing across the screen.
I’ve come-up with an elegant way to deal with the note queues, rotation and sending MIDI. I can send notes on 8 tracks, remaining stable to master BPM of >300. All Looking good so far!
All that’s left to do is implement Bjorklund’s algorithm and feed it’s output to the note queue.
Then finally the controls/menu system and that’s it! It’s all optimization from there. It seems the most challenging part is documenting the process, but I’m gathering my notes…
I should be wiring-up the 4x4pole, but here I am fiddling with breadboards. Suddenly got it into my head that I must build an Euclidean Rhythm based MIDI/CV sequencer. Bjorklund’s algorithm will be used:
The problem reduces to the following: construct a binary sequence of n bits with k one’ s, such that the k one’ s are distributed as evenly as possible among the zero’ s. If k divides evenly (without remainder) into n, then the solution is obvious. For example, if n = 16 and k = 4, the solution is [1 0 0 0 1 0 0 0 1 0 0 0 1 0 0 0]. The problem of primary interest is when k and n are relatively prime numbers , i.e., when k and n are evenly divisible only by 1.
Toussaint discovered (only in 2004!) that the resulting patterns of spaced pulses often represent the rhythms of many word music styles. This presents a simple way to generate new and interesting drum patterns.
There are already other Euclidean generators out there, and even one in the MIDIbix Seq4, but I have my own nascent ideas, even if I only a little knowledge on how to implement them. Another distraction!🙂
The logical conclusion to this effort should be a PCB for a unit that can act as a standalone MIDI/CV sequencer in a box, and one that is small enough to be integrated into a Eurorack setup. After some research I decided to start with Arduino. It comes with MDI libraries as standard, and already there are lots of interesting projects out there to reference.
In keeping with the DIY philosophy, I forewent the purchase of an Arduino board and instead opted to set-up a development environment using ATmega microcontroller and the Arduino libraries: – essentially to create an Arduino clone on a breadboard.
I am using the ‘standard’ Arduino chip – the ATmega328p (used in the MIDpal). As I had a spare ATmega644 (used in the Shruthi-1), I decided to breadboard it as well. The 644 is not officially supported by Arduino, however there is an Arduino-compatible bootloader called Sanguino.
There is plenty of information out there on how to breadboard these chips. The most useful video I found was this one from Notes and Volts, which helped me setup the 328p in no time.
In the process, the concept and function of the pins became clear, and I could easily map the ATmega pins to their Arduino equivalents using the following diagram:
For the 644 I used this reference and this pinout diagram:
Once hooked-up, I needed a way to program the chips. Fortuitously I had an Atmel AVRISPmkii to hand. I bought this to program my Shruthi chips and used it for all of 5 minutes a few years ago.
There are some issues to watch out for when using the AVRIPSMKii with Arduino:
OK, so after overcoming the AVRISP driver issues, the next task was to set the fuses and install the Arduino boatloader onto the virgin chips. This requires hooking-up the AVRISP to the correct Arduino/Sanguino Pins. Referencing the pinout diagrams above, making the connections is an easy task, but one that must be approached slowly.
Referencing the AVRISPmkii pinouts:
Just to be clear, this is pin 1:
I made a little breadboard adaptor for it…
Ending-up with something like this:
After triple-checking the connections, I fired-up Arduino IDE. After many failed attempts, I was still not able to bootload the damn things. Attempts at querying with avrdude always returned the following error:
avrdude: Device signature = 0x000000
avrdude: Yikes! Invalid device signature.
Double check connections and try again, or use -F to override
A web trawl suggested that this was due to the crystal not oscillating. Switching-out the crystal and trying instead a 16Mhz resonator returned the same error.
As it turned out, the default avrdude transfer speed is too fast for the virgin chips, therefore it’s necessary to adjust the baudrate (-B option in avrdude) to allow the initial communication. I didn’t realise this until after I solved it, and this was the problem that soaked-up most of my troubleshooting time. Bah!
Later, I realised that I could have done this much more quickly with Linux, and without the need for that massive Atmel install. Here are the required commands (Ubuntu).
Now fire-up Arduino IDE. No serial port is recognized on the AVRISPMKii, therefore trying to upload a sketch will return a ‘no Com port‘ error. It took me a few minutes to figure-out that you must instead ‘Upload using Programmer’ i.e press shift when sending the sketch (or just cntrl-Shift-U). Also, it’s important to ensure you are uploading to the correct board type in the Arduino IDE (Arduino UNO vs Sanguin 644).
I attached a MIDI out port and an LCD, loaded-up a simple MIDI sketch to test it.
It worked first time. A great start!
Next steps will be to wire-up some pots and buttons, and start writing the code.
2013 started brilliantly, let there be no doubt. By April, my newly found obsession with DIY synthesisers reached a peak, and a small pile of worthy project PCBs was collected. Then work took over. For a year.
There were times during that winter of DIY-lessness that I looked forlornly, hopelessly, at the cardboard boxes of fresh components, and the half-built units soaking-up dust on the shelf.
Whilst unable to actually build, dreams of what was possible prevented me from going mad, and I spent my spare moments planning, plotting, scheming for the day I would wield the soldering iron again.
Now that that day has come I can reveal the outcomes of my lengthy pause for cogitation:
The original plan for the 4x4Pole Mission was to house the 4 voices together with a scavenged Fatar keybed. Many panel designs were drawn-up, with this being the final iteration in that direction (click to view).
The problems with this setup soon became obvious:
In the end, I went for a compromise solution – but one I am very pleased with:
The redesign went much faster now that I had a clear idea of the endpoints. The alumium faceplates were quickly designed in FPD and sent to be cut by SchaefferAG.
I was very pleased with the results – with one minor niggle: I only used 2mm Aluminium for the Rack faceplate, meaning that there is some flex. This is fixable, but I should have used 3mm aluminium, similar to the keyboard panel, which has no flex at all. Oh well, you live and you learn.
Must say though, the anodized panels turned-out way sexy-looking. Victory candy included!
No excuses now….and no time to waste….
In my last post – a full 18 months ago! – I ravaged my poor old Evolution UC16 MIDI controller to prepare it for the installation of panel-mounted pots. In the intervening period the assaulted PCB eyelets gave-up hope and fell off, leaving me with a considerable amount of microsurgery to get the board working again.
Not pretty, but perfectly functional. Liberal amounts of hot glue and some PCB varnish should give it a few years of extended life.
Once that was done it was a matter of mounting the boards and wiring-up. Time-consuming but largely trouble-free.
So, the keyboard components have been tested and all work together. MIDI is internally wired from UC16 > MKE > MIDpal > out. Initial tests show this to be a stunning combo. The MIDIpal rules!
All I need now are the side and back panels. I know what I want, I just can’t afford the additional cost right now (about 70 euro). However, this is no problem because the 4x4pole is quickly coming together. As of last night, all boards were working as expected, the filters have been tuned, and the sound of awesomeness is in the air.
Just the wiring to go…..
in 3 or 4 days I should be there.
It’s finally here: NLB’s special edition modded Shruthi-1 SMR4 mkII.
Here is the SVG file for the case (right-click to download).
Many thanks to Mutable Instruments and the community.
To satisfy a couple of requests, I made this little vid to demonstrate non-destructive removal of the spring from standard gamepad joysticks (Xbox 360, PS3, Logitech, etc.).
If you don’t have a nephew’s games closet to raid, you can always buy the joysticks. This is my source. Tell Uwe I sent you.
Here’s where I show how to wire-them to the CV inputs of a Shruthi-1.
Back and building with vengeance…
This is the 1U rack version of the RY30. Release year: 1992 (manufactured from 1992 to 1995).
I’ve been eyeing them for quite a while and always thought I should pounce if one became locally available. Then two came along at once, and at 100euro per unit I had to get both.
Even without any expansion cards, for a 20-year old module it’s sounding great, with a small set of high-quality drum samples on-board. There’s a limited but interesting drum-focused synth engine. Pitched synth sounds are also facilitated, with basic but functional parameters to edit. No crazy pads here, but definitely usable for bass work – there’s a satisfying low-end.
The 64 factory kits are OK, if a little too generic. Although there are not many basic samples, the preset sounds demonstrate what is possible. My interest is creating custom kits. There are 128 slots for user-created sounds using the on-board and imported RAM waveforms. More than enough.
Each drum sound can be made-up of two ‘Voices’. Each Voice in turn contains 2 ‘Elements’, which consist of one drum sample and and it’s synth engine.
The synth engine is simplicity itself. There’s a very basic pitch envelope generator, an amp envelope, resonant low-pass and non-resonant high-pass filters ( (12 and 24dB), and LFO for pitch, cutoff and amp modulation.
The MIDI delay effect allows the first note to be switched-off, leaving only the repeats. Looks interesting, especially for glitchy/granular effects. Finally, there’s good control over velocity sensitivity, where volume, pitch, cuttoff and decay can be set to different values.
CC assignments are basic – 6 fixed parameters can be assigned for real-time control at kit level. The CC assignment can be switched off per sound, so there is some flexibility.
No internal FX, however some samples have reverb built-in, and this can be gated. Nothing fancy, but the sounds do seem to fit together, making it sound nicely cohesive out-of-the-box. The compactness of the 1u RM50 is also strongly appealing, though it is bloody heavy.
Deep editing with only the front-panel buttons may seem a challenge, but I’ve found it surprisingly intuitive and straightforward.
There’s also a cross-platform patch editor. Simple, but does the job – except no MIDI input under Windows.
Thankfully there are Headphone outputs. Until I get a proper soundcard I can’t record any clips, but soon….
Other nice things:
+ Each drum sound has two separately-editable voices or ‘Elements’ that can be mixed using the balance control which, mapped to CCs provide for morphing kits!
+ Handles heavy MIDI streams.
+ Deep velocity sensing options for dynamic kits.
+ Reversible samples.
+ Audio Trigger inputs = modular fun!
– CC control at kit level only, and only 5 fixed parameters. But selectable per voice.
– No real-time pitch change except via EG, which is rather basic.
– No sample start adjustments, however there’s a unique delay control per Element which looks good for glitchy sounds.
– No LFO to pan. Voice outputs are mono anyway.
– No noise samples, and a shortage of Hihats, however…
Lucky for me, one unit unit came with additional wave sample ram already installed, which allows me to upload 512k of my own samples. Not too shabby.
There is also a third-party expansion module available, but I think I’ll have enough in one.
I’ve already uploaded some Goldbaby samples via the MIDI port using Elektron’s C6 sysex manager. Easy.
Overall, I am very pleased with these acquisitions. I was going to build both the Sonic Potions LXR and the TR8060, but I simply don’t have enough time recently, and it’s beyond time I stopped farting around and got serious about recording for a change.
– Switch off RM50
– Press and hold PLAY + MACRO + SOUND while pressing the Power On button
– Hold play + utility buttons when powering-on
3V lithium backup battery (CR2450)
– 16-bit AWM2 (48 kHz sampling frequency) with digital filter 22-bit linear D/A converter
– Layering 2 elements/voices, 2 voices/notes
– Polyphony 16
– Voices 500 preset + 500 variation + 100 user; optional Wave Card: 32 user (x3); optional Data Card: 500 variation + 100 user
-Rhythm Kits Internal: 64 preset + 64 user; optional Data Card: 64 user
– Operation Modes Multi Play mode, Multi Edit mode, Voice Edit mode, Utility mode
– Trigger inputs!
– MIDI Parameters Program change mode selection; Program change table; Control change settings; Control change assignment; Remote mode selection
– Displays Multi-function 48-character LCD; Edit LED (red); MIDI received LED (red)
– Expansion Slots External wave card slot (x3); External data card slot; Internal expansion memory board slot
– Connectors Headphones, Line out L/MONO & R, Individual line out (x6), MIDI IN, OUT, THRU
– Power 120/220 V, 14
– (W x H x D) 480mm x 44mm x 347mm (D); 18 7/8” x 1 3/4” x 13 5/8”
Release year: 1992 (manufactured from 1992 to 1995)
Finally, I get some time to myself. But not much, so let’s crack-on.
First I connected-up the PSU to the mains. Now, if you are going to try this yourself I shouldn’t have to tell you to be careful, but I will anyway. BE CAREFUL. I’m not responsible if you fry yourself.
Anyway, I took no chances for first power-up. I stood well back and used a stick to switch it on, and I’m not ashamed of it😛
But it works. So, out comes the multimeter to measure the veracity of the advertised voltages.
All looking good. Nice and stable output, no vibration and barely audible noise. The trimpot allows quite accurate tuning, however, it affects all three busses at once, and there is slight variation between them. Not a big deal?
Next was to grab some stripboard, a couple of large capacitors and suitable headers. It turned out that I ordered the wrong IDC connectors. So I have female connectors for the busboard, and male cable clamps. But this is no problem, as they can easily undergo a sex change with the help of some spare pin headers.
The connectors are wired as follows:
Initially I won’t be using the +5 bus, so I’ll only need the 10-pin headers. Stripboard makes it easy to follow the busses, and the whole thing took just a few minutes. This is just a first run. I’ll probably reinforce the tracks later. But as it is there seem to be no problems.
The capacitors are probably not even needed, but it’s no harm to ensure that the supply is filtered. Some details here.
After connecting the PSU, I measured the voltages at the pins and it seems that we’re in business!
Let’s have a dance to celebrate:
Cost of each busboard comes in well under 5euro. Now I just need a finished module to power😛
While I’m waiting for some parts and test cuttings for the 4×4, it’s progress has been overtaken by this modular monkey business. I’m in no rush.
So, after assembling the Baugruppenträger and RAST, as expected, I required mounting rails to complete the job as first envisioned. Since I had a voucher for MusicStore, it was used buy the only rails they had – 12HE Adam Hall thingies at a fiver per pop. I got 4 to keep me stocked for future requirements.
Silly me hadn’t reckoned on them being made from a thick steel. As I don’t have the means to cut them, another much simpler and less expensive option popped-out: I could just glue or screw some 2cm wooden strips and mount the Baugruppenträgers directly onto these. Or, as I quickly realised, I could even just mount the vector rails directly this way and not require those ugly rack ears at all.
OK then – to hell with the Baugruppenträger!
However, it’s 4 vector rails can still be used to complete two 84HP racks, thus filling the RAST. So, for future reference, I will only need the Vector rails and the Thread strips. I could even do away with these altogether, but since I have them I’ll use them.
In depleting my voucher I couldn’t resist trying a Monotron delay, especially because it has exposed CV points on the board and would thus could be re-housed into a 4HE module. It’s nice but surprisingly noisy. That’s not necessarily a bad thing🙂
As I’m quickly learning, starting a DIY modular is not for the faint-hearted. Of course it would be easy to cave-in, buy a basic setup and go from there. But that’s not in keeping with the skinflint ethos.
It’s when looking at how to power my modular that I fully realised how archaic and uncoordinated the Doepfer Eurorack format really is. Options for powering the thing are surprisingly limited, and connector conventions are known to vary between manufacturers. Just to get up-and running with the Doepfer PSU + Busboards + 5v converter comes to a total of 145euro. These are crazy prices. I went looking for reasonable alternatives.
The first question was whether I would build my own PSU based on the many specs found at Muffwiggler’s. After some consideration I decided not to go this route. Being still the newbie in this area, I decided against, if only because there is a real risk of frying myself and my surroundings if something goes wrong. So I’m better-off with a pre-fab PSU option.
Following-up a suggestion made on the Mutable forums (thanks yet again Frank!) I went for this one. For 24 euro it provides three power buses – 5v and +/-12v – and with a whopping 5A on the 5v bus.
The unit is heavy and seems quite robust, with clear connector options – if somewhat exposed. Handy enclosure to protect little fingers. There’s a prominent trimpot on the fron for adjusting the +5v bus. Not sure yet how noisy it is when plugged-in.
There is also a 5v =/- 15v version which will probably find it’s way here at some stage. But let’s keep focus for now.
I’m thinking to build the busboards myself using perfboard and some nice big capacitors, and based on schematics I found at Muffs.
Another option might be to just use a ribbon cable and connect the first modules in series until I can come to a more permanent solution. Much reading to do. Oh for some free time!
I have to face facts at the beginning that this modular won’t be a static device – it will likely grow and evolve as the addiction takes over. So I’ve got to be prepared to make big changes along the way, and try to anticipate my future requirements.
If I can get this PSU working as intended, I’ll be well-covered for power options.
Well the Tempest turned-out to be a damp squib for me. Or rather DSI’s inability to implement features that were hotly discussed over a year ago. Maybe they will eventually pull the rabbit out.
Even so, there seem to be a lot of Tempesters who are perfectly happy with the current OS, and that has to be respected. As with everything else in life, people will have a different approach to, and expectations from, such an instrument. Everyone should come to their own conclusions.
For me, the hardware interface, the sound and the overall concept were incredibly appealing. The forum was abuzz with excited discussions about the untapped potential, and we looked to be in line for a classic that would finally put the Roland box voodoo to bed. A powerhouse on paper, a legend in the making.
Now don’t get me wrong – in some respects it’s nearly there. I had fun designing sounds on this machine and I learned a lot along the way. The glass is definitely more than half-full, and I certainly do not regret the purchase.
But the sequencer is still too rudimentary for my needs. I didn’t see anything in the recent betas that give me confidence in user-requested improvements beyond the necessary bug fixes. DSI’s prolonged silences do not help. To me, most of the requests seem reasonable, and there appears to be broad agreement on the most important. The persistent bugs, though not serious, reveal a lack of attention.
Nobody is asking for a Cirklon-killer – only that the machine is given a sequencer worthy of it’s potential. And it seems so tantalisingly close.
At current pace, by the time the OS is ready, the second-hand market will already be mature, new competing products will have been released, and DSI will have lost sales. Brand damage has already occurred. I am amazed that the can’t see this! I would have said ‘bemused’, but it’s actually quite sad. I really want DSI and their products to succeed – the innovation is wonderful – but there seems to be no ambition to finish the job.
I can only speculate that the feasibility of new features is dependent on internal hardware limitations such as the the 4mb limit (honestly, what were they thinking!!), rather than the coder’s inability. He seems talented, if disorganized and overworked and, it would appear, fighting some serious battles to squeeze better performance out of this box.
Nevertheless, my biggest disappointed has been their habit of raising expectations and not following through. I honestly didn’t expect that, and I fell for it wholeheartedly.
They are also not giving the game away regarding what’s possible or likely to reach implementation. Engaging in further feature discussions seems pointless, as none have materialised. Beta testing has been the simple repetitive process of pointing-out the same few bugs in every release.
Admittedly my passion boiled-over after seeing the latest delay (of several months) was due to focus again shifted to the Prophet 12. OK guys, but at least tell us! It would be ridiculous to expect new product development to stop, but it is also reasonable to expect that existing products should receive the necessary attention.
In short, I am extremely disappointed with DSI’s behaviour. I would not have accepted this from another company, but I was rooting for DSI to succeed, and I trusted them. More fool me.
As a sweetener it was revealed that Tempest and P12 share a common codebase, and DSI have said that features will be rolled-back into Tempest. This sounds great, but makes me wonder, since the Oscillators are completely different, and P12 has no sequencer – the element that needs most attention. Time will tell.
I do earnestly wish DSI success with P12. It looks like it will be a winner and they have clearly worked very hard on it. They must be congratulated for the risks they are taking to innovate exciting new products. In the grand scheme of things, complaining about a few missing features seems churlish.
But when it comes to Tempest, I’ve simply lost faith. It had to go.
With the proceeds I have fostered a beautifully-built MIDIbox Seq4. Now THIS is a sequencer. It completely obliterates DSI’s effort. Open-Source with frequent updates still after 5 years of development. All that user input has been eagerly gathered and distilled into a great interface and a smooth, almost effortless workflow. And zero fuss from the unpaid developer.
There is more than enough cash remaining to start a nice modular and have a big party. I’ve already sourced the PCBs to make my own drum modules, which will be triggered by Seq4’s 8 analog outs. I can simultaneously sequence FM percussion provided by FS1r. What I lose in compactness I more than compensate for in flexibility and power.
My first Eurorack module is ready to be assembled. Little Dieter is due to be born any day now…..
I’m sticking with open-source hardware from now on. No more ‘black boxes’. It’s AD 2013.
There will be no pining for Tempest.
Here are some Kits and sounds – the ones I managed to backup before the new owner ripped the Tempest from my arms. FWIW, he’s absolutely delighted with it.
I dedicate this one to DSI😛
(One of my favourite basslines and easily possible on Tempest)
‘What do African rhythms, spallation neutron source (SNS) accelerators in nuclear physics, string theory (stringology) in computer science, and an ancient algorithm described by Euclid have in common? The short answer is: patterns distributed as evenly as possible. For the long answer please read on…‘
The build is proceeding nicely…
Firstly, the boards have been populated with resistors and most capacitors. The digital control boards are complete save for the OLEDs, buttons and pots, which will be panel-mounted. Tragically, and despite my heroic efforts at military planning/sourcing, I didn’t order enough 100n capacitors for the filter boards. I now diligently plan my next components order.
Quite some time has already been spent working on the case design and front panel. Using Inkscape, I’ve already ordered and received some test cuttings in 3mm acrylic glass. I’ve firmed-up on some nice ideas that I will expand upon in another post. Still deciding whether to invest in wooden cheeks, but then again, they can come later.
I’ve flip-flopped daily over whether to implement the Shruthi XT programmer, or stick with generic midi control. I’ve now firmly decided to not use the Programmer this time – though I did grab 2 PCBs in the latest bulk order, and I’ve already gathered the parts. The reasons are:
So, MIDI it is, and by chance I have an old beat-up Evolution UC-16 which I will be cannibalizing.
Quite easy to take apart….
I thought there might be an extra MIDI port hiding in there, but alas…
Now, these pots are a bit worn so they need replacing. I suspected they were 10k linear pots. A quick test confirmed this. Problem is that these pots are so welded to the PCB that the cleanest route is to just cut them off…
So now I can add my own pots and arrange then as I wish on the front panel. By also adding the keypad I will have access to 25 banks of 16 MIDI CCs, each knob being freely assignable to any CC on any channel. That’s a lot of control!
Further, this can take MIDI input directly over USB, so I can sequence the whole caboodle from a DAW if I wish. But where’s the fun in that?
And why stop there? I’m also integrating my as-yet unbuilt MIDIpal to make this a super-charged master keyboard. Better than the plastic shit available these days, and fully customised. Cheaper too.
MIDIpal has 8 CV inputs = Joystick pron! hehe
The bonus is that, with 4 Shruthis I will also have a routable 4-in 4-out MIDI interface.
I’ve had some time to return to and investigate the wheel connectors on the MKE.
Remember that, when I tested the Novation wheels the range was minuscule. It turns out that the wheel receivers are calibrated to Doepfer’s own standard. It also accounts for the narrow range wheels have normally, due to the spring on the pitch wheel and the large diameters.
A closer look at the Novation wheels revealed 2 x 5K linear pots like so:
The Doepfers are 10K so I decided to use a trusty joystick, this time with the spring intact, to see if it could be used as a mini pitch-wheel.
The result was disappointing. At rest pitch was always at maximum, and moving the stick to one side again swept through the range an a narrow band. No, this won’t do.
To see if I could exactly define this range, I attached a standard 10k linear pot. Again, same result. Hmmm.
So, how could I limit the range of a pot?
Thankfully the answer came quickly and it’s such a simple fix it’s almost embarrassing…
…how about if I just twist the Novation pots in their housing? This way I could tune the range by hand. So, after cleaning the solder off the pins, I plugged the Novation wheels back-in, loosened the pots and re-orientated them.
First, the potentiometer tabs needed to be clipped-off. (I now know what they’re for – holding the pot in place!)
Then plugged-in (after removing the solder I left on the pins😛
And, by Jove, it works like a charm!
The main caveat is when re-orienting the pitch-wheel pot – you need to account for the spring, so best to have MIDIox or other such monitor open to see when you’re at zero before tightening the nut. The pots are d-shaft, so in this case I could easily find zero.
Here they are with their pots newly-oriented and wheels re-fitted.
Everything is now up-and-running with the MKE. Superb!
Buoyed by this success, I’ve decided to adopt a modular approach to the 4×4 pole enclusure, wherein the synth section will be detachable from the Keyboard.
Because, who knows, I might want to build a 4xLadder Filter at some stage…