[Nederlandse versie]

<Ob>

This musical robot belongs to the category of our automated classical music instruments: the oboe. The approach here was quite experimental and was an attempt to realistically automate an existing unmodified instrument, and thus it does in fact make use of a fine classical oboe. The instrument used is a Brussels made concert instrument by F.Debert, probably to be dated first half of the 20th century.

Electromechanical control of the levers did not confront us with any real problems apart from the quite delicate and differentiated mechanics. Silent operation of these have been our main concern. We simplified the fingerings such that we could suffice with less than 16 solenoids: six closing the open holes, and the strict minimum of seven for the essential levers. Some of the levers (such as the three octaving levers as well as the levers to facilitate trills) are essential for players, but have much less importance in an automated instrument where the attack of the tone is guaranteed by the nature of the sound mechanism and where resonance on partials can freely be used. Hence our simplified fingering table: the yellow lines indicate the pads that are solenoid controlled in our final robot project.

On the left side of the table the notes/pitches are given as midi note numbers. Levers that are in no instance colored red, were left un-automated in our robot. For some levers, we did not implement a solenoid to action them the way human fingers would do, but rather choose to automate the closing pad directly. We are developing special fingering tables for quartertone and other microtonal applications. Alternative fingerings in order to obtain different sound colors are implemented as well. The standard fingerings are indicated as hexadecimal numbers in the right column on the fingerchart.

The double reed however, became the main problem. The first experiments conducted us to the design and building of double reeds made from piezoelectric material glued to brass plates. We got a few prototypes build along this line, up and working and indeed the concept is workable. The main problem here was the very low obtainable sound pressure, even when driving the piezo-material well above its rated maximum voltage (35 V). The second series of experiments was carried out using a double faced piece of piezoceramic bonded to a central brass plate and placed just touching to an absolutely flat thick brass plate with a central perforation of 4.2 mm. This mechanism gives a strong buzz but unfortunately, sound production is very frequency dependent as well as dependent on applied air pressure (after placing the assembly in a closed container). A secondary problem in this approach was the noise generated by the compressor. We used a small DC motor driven vacuum cleaner type compressor capable of producing the required pressure of about 15 to 30 mBar. We dropped this prototype and the compressor later was brought to good usage in our <Melauton> robot... Therefore a thirth series of experiments was carried out using tweeter motor driver made for driving an exponential horn. Instead of coupling the driver to an exponential horn, we designed an acoustic impedance converter modeled after a real reed in a human mouth cavity. This piece had to be fabricated on the lathe and is drawn here: The 12 mm long cylindrical part on the left side of the piece fits inside the outlet of the RCA pressure driver, without touching the titanium dome inside however. The other side fits nicely into the oboe, replacing the reed. With this mechanism, the realism of the produced sound becomes highly dependent on the waveform applied to the driver. Something trapezoidal seems to work best. However, in order to come close to original oboe sound, articulation is very essential: frequency modulation, phase modulation of at least the first two partials above the fundamental as well as some amplitude modulation (envelope shaping). More experiments with the exact shape of the acoustic impedance transformer will be performed in the near future. The circuit for driving this motor was derived from the circuits designed earlier for robots such as <Korn>, <Bono> and <Aeio>: It uses the same PC-board and the same 16-bit PIC microcontroller. The firmware however, is quite different. For coupling of the circuit to the motor driver, we use a classical audio output transformer. The two resistors and the capacitor shown in the circuit diagram form a simple formant filter tuned together with the inductance of the transformer, to the required strongest formant frequency for oboe sound. Some experimentation with capacitor values was required here.

As an extra feature, we suspended the entire automated oboe construction in a cradle. Thus the instrument has freedom to move in different inclinations. The axis of suspension is provided with a dented wheel driven by a chain and a Crouzet DC motor with reduction gears. This way, any inclination can be held and controlled. The movement possibility was added since it mimics a bit the behavior of a human oboist. In the software we use for controlling the automaton, we are implementing rules such that the robot derives its gestural behavior from the music it gets to play. This feature however will be implemented in our GMT language and software environment, rather than in the firmware of the machine itself. The circuit secures that the instrument is not allowed to turn fully around, since that would ruin the robot. Movement is limited to an angle of ca. 90 degrees. In the first version we used mercury switches as sensors for the position, but these suffered a lot from double triggering. For version 2, we first tried using a new design using proximity sensors (Pepperl+Fuchs). Obviously the firmware had to be completely redesigned, but the results with these sensors where still unsatisfactory. Finally we decided to call in an analog tilt sensor by Penny & Giles (Type STT290/60/P2) allowing us the read the position of the instrument at all times using an analog input port on the PIC controller. For the motor control we made use of a Trident 4-quadrant DC motor controller:

In order to facilitate interactive applications, we provided <Ob>> also with a midi output port from which the position of the oboe in its cradle can be read in at any time.
Finally, some light effects were added to the design as well. These are also controlled by the midihub board, since this board had still plenty of outputs available.

Only when integrated in the context of our Logos Robot Orchestra with its wealth of varied sensor systems allowing full interactivity, this automate becomes a true robot. That's after all were its destination is to be sought.

Midi channel: fixed to 14 (counting 0-15).

Note Off: Implemented for notes 58 to 96. Starts a short decay (release) on the playing note. The duration of the release is determined by the value of the controller 19. For legato playing, note-off's can even be dropped, since the instrument is monophonic. Note-off with release is not implemented here, so only controller 19 can be used yo this purpose.

Note On: Implemented for notes 58 to 96. Velo-byte is used for the level of the sustained sound portion. To steer the attack force, controller 17 must be used. Note that for this robot, the usual general volume controller (#7) is not implemented. The lights are also mapped on notes, but make use of the 0-5 range, outside the normal range of the oboe. For those lites the velo byte is implemented to make automatic flashing possible. The velocity value determines the speed of the flashing. With value = 127, the lites will be permanent on. With zero value, they will be off.

Key pressure (160 + channel, msb, lsb) is implemented to allow for alternative fingerings. msb contains the solenoid activation bits for the lowest 7 solenoids, lsb the remaining 6 bits/ solenoids on the high side. The mapping follows the wiring details given in the maintenance section at the very bottom of this webpage. The note aftertouch command should be issued immediately after a note on. It than will override the normal fingering. Key pressure commands can be sent during the playing of a note, thus rendering sound timbre effects possible as well as improved fingering for quartertone music. After a note off command, aftertouch will have no other effect than soundless key clicks. In fact, a noteoff will always release the finger-solenoids. Although only implemented for debug purposes, it might be good to know that midi note 0 will always release all valves. As such it is equivalent to sending 174, 0, 0.

Controller 17 is used to control the maximum sound level during the attack period.

Controller 18 is used to control the duration of the note attack. The interdependencies of these controllers together with the velo byte is shown in the graph below:

Controller 19: Release time. If legato playing is in use (meaning we do not send note-off's) the setting for this controller will be overridden. If you use sequencer software that does not allow you to not send note off's, legato playing can be obtained by setting this controller to values between 105 and 112.

Controller 20: Tuning. The basic tuning of the instrument can be shifted a quartertone upwards. At setting 5 the tuning corresponds to A=440 Hz. (Precision better than 1 cent).

Controller 22: Inclination controller. Value 64: downward vertical, 63-0= move backward, 65-127=move forward. The full traject is ca. 90 degrees.

Controller 66: Robot on/off switch. On power down (ctrl.66 set to 0) the motor will bring the instrument back to a full verical position

Controller 123: switches the sounding note off, releases all keys, does not affect motor position, dims all the lights.

Pitch bend: The <Ob> robot can be used in any tuning system. Using the Midi Pitch Bend command, any playing note can be raised of lowered by a quartertone. Hence the full range (0-127) corresponds a semitone. The tuning returns to normal with the pitch bend set halfway (64). In the drawing below we give the coding example for a fragment of a quartertone scale:

Most good sequencer software (such as Cakewalk or Sonar) use the signed 14 bit format. Note that one unit of the msb corresponds exactly to a 0.78 cent interval. To convert fractional midi to the msb only pitchbend to apply follow following procedure: if the fractional part is <= 0.5 then msb= 63 + (FRAC(note) * 128), if the fractional part is larger than 0.5, we should switch on the note + 1 and lower the pitch with msb= (1-FRAC(note)) * 128. Note off should reset the pitch bend for the playing note!

Program change: not implemented.

Lights: The lights are mapped on very low midi-notes as follows:

• note 0: white LED's (four) on oboe sensor
• note 1: nc
• note 2: yellow LED strip (mounted on the large power supply, backside)
• note 3: yellow LED strip (mounted on the large power supply, frontal)
• note 4: white LED sportlite, directed upwards
• note 5: yellow LED strip (mounted inside the frontal bend in the stainless steel frame)

Midi output port: sends the actual position of the oboe as a 7 bit number, using the <Ob> midi channel. The format is that of a note pressure command. The pressure value corresponding to the angle of the instrument: 0= extreme inwards, 64= vertical, 127= extreme frontal. The data is only sent on changes of actual position. The midi output connector is the DIN outlet on the extreme right side of the hub board.

Technical specifications:

• size: 400 x 460 x 1000 mm. Flightcase:
• weight: 14 kg. Weight with flightcase:
• transportation: needs a flightcase. Can be taken as checked luggage in airplanes.
• power: 230 V ac / 65 W (Fused: 3.15A, fast)
• Tuning: based on A = 440 Hz (within 1 cent). The tuning can be adjusted.
• Ambitus: 58-96 [notes 97-126 for experimental uses only]
• control: MIDI-input, 5 MIDI-Thru. (UDP/IP Ethernet port to be implemented later)
• Insurance value: 13.650 Euro

Design and construction: dr.Godfried-Willem Raes

Collaborators on the construction of this robot:

• Johannes Taelman (dsPIC coding)
• Kristof Lauwers (GMT implementation)

Music composed for <Ob>:

• Godfried-Willem Raes "Primes" (with <Qt>,<Heli> and player piano. MP3-download
• Godfried-Willem Raes "Slones" (2012)
• Godfried-Willem Raes "Fall'95" (2017), integrated in his 3th Symphony.
• Daniel Pastene "Terminator in Afghanistan" (2010)
• Yvan Vandersanden

Pictures taken during the construction, maintenance and repair in our workshop:

Back to composers guide to the M&M robot orchestra.

Back to Main Logos page:index.html

To Godfried-Willem Raes personal home page...

To Instrument catalogue

Nederlands:

<Ob>

Voor deze robot stond ons als doel voor ogen zo getrouw mogelijk de werking van de traditionele hobo te mimeren. Het entoeziasme van mijn medewerkers bij Logos voor zo'n projekt was nogal wisselend: van uitermate negatief en afwijzend tot gematigd hoopvol. De suksesvolle automatisering van zo'n traditioneel geladen instrument leek bij voorbaat tot mislukken gedoemd. We namen de uitdaging niettemin aan en het lag, gezien het uitgangspunt, daarbij dan ook voor de hand te vertrekken van een bestaand instrument in goede staat. Daartoe konden we beschikken over een orkest hobo gebouwd door F.Debert in Brussel, vermoedelijk in de eerste helft van de 20e eeuw. Ooit eens gekocht op de Gentse prondelmarkt.Afgezien van de relatief ingewikkelde en weinig repetitieve mechanika, stelde de bediening van de kleppen op zich geen nieuwe problemen. De krachten nodig voor het bedienen van de vele klepjes zijn immers betrekkelijk klein. Niettemin zagen we af van het automatiseren van elke voorhanden klep op het instrument, omdat vele ervan slechts nodig zijn voor menselijke spelers. Zo zijn de drie hulpkleppen voor de oktaven alleen nodig bij menselijke bespeling om de gewenste toon snel te laten aanspreken. In ons ontwerp is dit evenwel gegarandeerd op grond van de wijze van aansturing zodat we aan een automatisering van deze kleppen gerust konden verzaken. Hetzelfde kan worden gezegd van de trillerkleppen. Waar mogelijk maakten we voorts gebruik van de mogelijke maar door spelers eerder vermeden vorkgrepen. Zo slaagden we erin de gehele applikatuur te automatiseren met niet meer dan 6 elektromagneten voor de open toongaten en 7 elektromagneten voor de bediening van de klepjes. De vereenvoudigde vingerzettingstabel kwam er dan uit te zien alsvolgt: de gele lijntjes geven aan welke kleppen via elektromagneten worden bediend.

Mikrotonaal spel, inklusief kwarttonen zijn mogelijk door toepassing van non-standaard grepentabellen. Alternatieve grepen kunnen trouwens ook worden toegepast om klankkleurvariaties te bewerkstelligen. De hele opbouw van de elektromechanika voor het vingerwerk was uitermate tijdrovend, maar toch niet bijzonder moeilijk. Het vereist wel heel precies las-, tap-, boor-, zaag- en snijwerk.

De toegepaste standaardvingerzettingen zijn als hexadecimale getallen weergeven in de rechterkolom van de vingerzettingentabel. Dit is de wijze waarop de grepen-opzoektabellen in de mikroprocessor worden geimplementeerd.

Anders daareentegen met het klankopwekkingsmechanisme: het dubbelriet. Onze eerste experimenten gingen onmiddellijk uit naar de konstruktie van dubbelrieten opgebouwd uit twee plaatjes piezokeramisch materiaal, gekleefd op messing blaadjes. Het grote mechanisch probleem daarbij is dat het piezokeramisch materiaal zich niet laat buigen (het is erg breekbaar) terwijl het riet, wanneer we de opbouw van een traditioneel hobo riet volgen, in elk geval moet worden rondgezet rond het kleine inblaaspijpje dat in het instrument wordt geplaatst in een cilindrische kurken houder. Na ettelijke pogingen slaagden we erin een riet op deze wijze op te bouwen. Aangesloten op onze frekwentiegenerator -beide riethelften parallel geschakeld en dus elektrisch in faze en mechanisch in tegenfaze- bleek het nog te trillen ook! Alleen, de geluidsopbrengst bleek ruimschoots onbevredigend. We haalden hooguit een -voor een normale hobo overigens zo goed als onmogelijk - pianissimo. Bovendien produceerde het riet ook bij bepaalde stuurfrekwenties multiphonics die te wijten waren aan de eigenresonanties van de messingplaatjes. De toon sloeg ook regelmatig over naar het suboktaaf. Opgemerkt moet worden, dat we bij deze experimenten geen wind toevoerden aan het riet, vanuit de ervaring opgedaan met eerdere koperblaasinstrumenten waarbij bleek dat de luchtverplaatsing door het instrument eigenlijk niet wezenlijk is voor de klankproduktie van het instrument. Ook het aanleggen van een sterk verhoogde stuurspanning (de normale toelaatbare maximum spanning wordt in het datasheet opgegeven als 35V) bracht geen soelaas. De dissipatie neemt toe maar niet de geluidsproduktie.
Een tweede reeks experimenten voerden we uit onder gebruikmaking van dubbelzijdige piezoschijfjes. De amplitude van de trillingen is een veelvoud van die in het eerste experiment. Dit schijfje werd passend en met een geringe aandrukkracht aangebracht tegen een vlak geslepen dikke plaat messing die centraal werd geperforeerd met een gat van 4.2mm, overeenkomstig de diameter van de hobo bij de inlaat van het rietje. Via een opgesoldeerd buisje (7mm buitenmaat) kan het mechanisme op de plaats van het riet op de hobo worden aangebracht. Voor windtoevoer moet het geheel uiteraard in een potje worden geplaatst met een toevoer voor de lucht en (luchtdicht afgesloten) voor de elektrische aansluiting. Dit mechanisme levert voor bepaalde tonen een vrij goede klank maar blijkt in extreme mate frekwentieafhankelijk te zijn. Ook is het aanspreken van de tonen erg afhankelijk van de aangelegde winddruk. Het geluidsvolume -althans voor de goed aansprekende tonen- is heel wat beter dan in het eerste experiment, maar toch nog te gering in vergelijking met een realistische hobo.
Een derde reeks experimenten werd opgezet onder gebruikmaking van een tweeter motor driver van RCA. Als koppelstuk maakten we daarvoor een akoestische impedantietransformator gemodelleerd naar een menselijke mondholte met riet. Dit stuk, vervaardigd uit massief messing, werd gedraaid op de draaibank. Dit is de maatschets: De linkerkant is passend gedraaid om 12mm in de RCA motor driver te worden aangebracht, zonder de inwendige titanium dome te raken. De rechterkant werd afgedraaid om in de hobo te passen op de plaats van het riet. De betrouwbaarheid van de klankproduktie is nu uiteraard perfekt, maar om de klank echt realistisch te maken blijven heel wat modulatieparameters essentieel. De golfvorm van de aansturing moet trapezoidaal zijn en frekwentiemodulatie (vibrato) lijkt aangewezen. Bovendien draagt faze-modulatie van de eerste twee partials boven de grondtoon wezenlijk bij tot het realisme van de klank. Ook enige mate van amplitude modulatie (omhullende) blijkt noodzakelijk. Wat dit betreft sluiten de resultaten goed aan bij de bevindingen opgedaan bij de bouw van <Korn>, onze automatische kornet.
Een bijzondere eigenschap van <Ob> is dat hij ook kan bewegen. Daartoe monteerden we de gehele draagstruktuur voor de hobo zelf op een 16 mm as voorzien van een tandwiel waarop een ketting loopt. Deze ketting wordt aangedreven door een DC motor met vertragingskast. De tandwielen en de ketting zelf recycleerden we uit een oude fotokopieermachine. De motor dienden we evenwel nieuw te kopen omdat die doorlopend van draairichting moet kunnen wisselen, iets wat met gelijkstroommotoren kan worden bereikt door eenvoudigweg de poling om te wisselen. Hierdoor kan <Ob> niet alleen schommelen, maar ook op eender welke hoekpositie binnen een bereik van ca. 90 graden vastgezet worden. Iets van de beweeglijkheid van de menselijke spelers werd dus ook hier geimplementeerd. Opdat er geen destruktieve ongelukken zouden kunnen gebeuren, werd het schommelmechanisme aanvankelijk uitgerust met vier kwiksensoren die het doordraaien van de motor moeten verhinderen en het bovendien mogelijk maken het instrument zuiver vertikaal te positioneren. Bij een latere upgrade vervingen we deze kwikschakelaars door een Penny&Giles hoeksensor wat een veel betrouwbaarder resultaat opleverde.
Tot slot bouwden we ook in <Ob> enkele bestuurbare lichteffekten in, gemapt op note on/off midi kommandos.
De volledige midi-implementatie is opgenomen in de Engelse tekst hierboven. <Ob> luistert uitsluitend naar midi kommandos via kanaal 14.

De elektronische schakeling bestaat uit enkele afzonderlijk funktionele boards:

1. TC/UPD en/of midihub board;

Dit board bestuurt de vizuele komponenten van deze robot: de slingerbeweging en de positie van het instrument evenals de diverse ingebouwde lichteffekten.

Het schakelschema voor de motor links op de tekening was onze eerste versie. Als beschermingsschakeling werkte die echter al te goed, d.w.z. wanneer een veiligheidsschakelaar werd bekrachtigd, werd de motor stroomloos. Het probleem dan was echter dat hij met geen (elektronische) middelen nog in beweging te krijgen was... Vandaar dat we in de tweede versie opteerden voor een automatische omschakeling van de draairichting. Daarbij werd beroep gedaan op een impulsrelais. Omdat de mikrokontroller nu evenwel geen informatie heeft over de eigenlijke draairichting, voorzagen we in twee analoge outputsignalen die kunnen worden ingelezen. Wanneer beide signalen positief en hoog zijn, is de motor onbekrachtigd. Het ene signaal van beide dat laag is, geeft de draairichting aan. Beide signalen kunnen nooit samen op massanivo zijn, tenzij bij uitval van de motorvoedingsspanning zelf. De kondensator en de diode werden toegevoegd om de stoorspanningen van de motorkollektor en van het PWM signaal uit te filteren. De spanningsdeler staat er borg voor dat de PIC ingangen nooit meer dan 5V te verwerken kunnen krijgen. Hoewel deze schakeling toch bijna twee jaar heeft gewerkt traden er toch geregeld problemen op met de kwikschakelaars die soms dubbele pulsen gaven waardoor het impulsrelais kon gaan oscilleren... Een manuele nood ingreep bleek telkens weer noodzakelijk om verder onheil te voorkomen... De derde en voorlopig definitieve versie (Ob versie 2.0) werd opgebouwd onder gebruikmaking van een Trident 4-kwadrant DC motor-kontroller: Dat was niet zo eenvoudig als het lijkt omdat alle stuurspanningen bipolair en zwevend moeten zijn voor dit type kontroller, wat de interfacing naar mikrokontrollers erg moeilijk maakt. Eigenaardig genoeg zijn er erg weinig goede 4-kwadrant kontrollers voor kleine gelijkstroommotoren op de markt (die van Maxxon even buiten beschouwing gelaten want die zijn onnoemelijk duur...).

Het volledige schema van dit mikroprocessor board is weergegeven aan het eind van deze webpagina. Het UDP/IP ethernet board is volop in ontwikkeling. Als extra, werd op dit board ook een midi output voorzien die toelaat de juiste positie van de hobo in te lezen. De meest rechtste DIN konnektor op het hub board wordt hiervoor gebruikt en kan dus niet als midi-thru worden ingezet.

2. Kleppenbesturingsboard

De PIC op dit board, een 18F2525, staat in voor het juist toepassen van de vingerzettingstabel op de gevraagde te spelen tonen. Het PC board is hetzelfde als wat we ontwikkelden voor heel wat slagwerk robots zoals <Dripper>, <Casta>, Troms>, <Psch>, <Snar> e.a. De zestien beschikbare mosfet driver uitgangen volstaan voor het aktioneren van alle nodige kleppen: dertien. De firmware is uiteraard totaal verschillend. De Midi input (in TTL formaat) wordt toegevoerd vanuit het midi-hub board. De op de print voorziene 74HCT14 chip is niet gemonteerd. Details van het schakelschema zijn te vinden onderaan deze web-pagina.

3. Rietbesturingboard:

Hiervoor konden we hetzelfde printontwerp gebruiken dat we eerder ontwikkelden voor <Aeio> en dat gesteund was op prototypes ontwikkeld en gebouwd voor robots zoals <Korn>, <Hurdy>, <Autosax> en <Bono>. Ook hier werd eenzelfde PIC mikrokontroller toegepast: de Microchip dsPIC30F3010.

Pas wanneer deze automaat wordt geaktiveerd in de kontekst van het M&M orkest, waarbinnen hij vanuit een veelheid aan diverse sensoren en via doelspecifieke software kan worden aangestuurd, wordt hij een heuse robot. Daarin ligt ook zijn bestemming.

Construction & Research Diary:

• 05.1992: Purchase of the oboe on the local Ghent flea market.
  
• 07.09.2008: first practical experiments with the construction of an electronically driven double reed. We made the first prototype reed from brass with two sides and layers of piezo electric ceramic.
• 08.09.2008: Experiments with double faced piezodiscs striking a flat perforated plate.
• 09.09.2008: Construction on the lathe of an acoustical impedance transformer for a motor driver replacing the reed.
• 10.09.2008: Research on fingering tables in order to minimize the complexity of the mechanics.
• 11.09.2008: Measurement of required forces for activation of the valves.
• 17.09.2008: Start design of the instrument holding chassis.
• 20.09.2008: Construction of the oboe holding parts and clamps. The backbone could be made of a rectangular stainless steel profile 30x50x2, as long or even somewhat shorter than the oboe itself.
• 21.09.2008: Considering to give this robot at least one degree of freedom to move. A pendulum kind of movement seems appropriate here, since it resembles human players' behavior. Once again, a 3-wheel base construction appears to be the optimum choice.
• 22.09.2008: The thread on the compressor driver appears to be very obscure. Outer diameter measures 34.5mm and the speed appears to be 1.5mm. However, M35 x 1.5 nuts do not fit well. It cannot be BSW or UNF neither. Maybe PG35 (unfindable in any of our catalogues) or some obscure American ISO thread???
• 24.09.2008: Machining of carrier for the lowest three solenoids (notes 58,59,60). Delivery of classical prepared oboe reeds.
• 25.09.2008: 10 solenoids mounted in a first approach. It seems that 13 solenoids is the minimum required to fully automated the oboe fingerings.
• 26.09.2008: All 13 essential solenoids mounted on welded-on stand off's.
• 27.09.2008: Calculation of required motor specs. for the pendulum like movement. A geared DC motor seems most appropriate (Crouzet type). Speed should be in the order of 15rpm and power 10 to 20Watts, if we want the oboe to be capable of moving over a 90 degrees angle in 1 second of time. We go for a DC motor here, because we need an easy way to toggle the direction of rotation. Single phase AC induction motors in this application would have been a lot cheaper and easier to find, but their direction of rotation cannot be changed. 3-phase AC motors, the most silent and flexible, would be overkill here...
• 28.09.2008: Wheel selection: 200 mm diameter, axis 20 mm seems appropriate for the two back wheels. Front wheel ought to be pivoting. Vertical carriers: AISI304, 25x25x700, thickness 2 mm. Thickness of the round balancing stave: 16mm, ends turned of to 8mm for tight fit in bronze bearings.
• 29.09.2008: Design of the horizontal base plate and structure for the front wheel, used for mounting the PC-boards and the power supplies.
• 30.09.2008: Welding works on the wheel base. Positioning and selection of the required transformers and power supply components. Positioning of the Crouzet motor.
• 01.10.2008: Wiring of the power supply board (8 V ac, 15 V dc, 24 V dc). Order of alternative toroidal transformer, since the type we initially fitted was 2x15 V. We need 2x12 V at ca. 15 W. Start placement of components on the midihub board. Power supply board tested and mounted in place. Mains filter, fuse box, switch and IEC inlet combination wired and mounted in the chassis.
• 02.10.2008: New toroidal transformer mounted. Bracket for the motor adjusted and mounted on the frame with two M5x40 bolts. Toothed wheel mounted on motor spindle. Power supply components wired. Power entry block with fuse and switch fixed in place with Parabond 600 Hitack.
• 03.10.2008: Motor circuit changed. Comments on the circuit added in the main text. Johannes Taelman informed about this new robot project.
• 04.10.2008: Search for a suitable 10-20Watt audio transformer with 1+1 : 1 ratio and impedance in the order of 8 Ohms. Frequency response is not very critical, since we start only at about 230Hz. Such part seems to be no longer on the market. We have to dig up something from our collection of vacuum tube era components. As an alternative, of course, we could always wind a transformer ourselves... Tedious but not particularly difficult.
• 05.10.2008: Small 16-output pulse board soldered and mounted on the Ob chassis. All mosfets are IRL640. This board does not have a programming connector, thus the clamp will be required for uploading the firmware. Midihub board populated with extra parts for the analog inputs on the breadboard section. Board mounted in place on the Ob chassis.
• 06.10.2008: All power supply related wiring, included the motor drive wiring, temporarily finished. The mercury switch we had in mind for use as a vertical position sensor appears to be a bad idea. It requires 180 degrees to fully switch...
• 07.10.2008: Design and construction of a new position sensing system using four separate mercury switches. For the sensing of the vertical position, we use two mercury switches connected in series. When both switches are closed, the vertical position is reached. The remaining two switches are parallel connected and are used to steer the motor polarity reversal relay. Wiring of the sensor board completed. Note that the use of mercury switches nowadays is illegal in commercial products. The picture shows the sensor assembly. The two mercury switches above are for the vertical position sensing, the ones below for excursion limit determination. The four LED's are just a visual extra feature. The sensor assembly is mounted on the oboe carrier profile with a single M4 bolt. The hole in the carrier is threaded. A 10 mm stand off should be used. Once ready and the exact positions of the mercury switches are determined, they should be fixed in place with some dots of MS-polymer or hot glue.
• 08.10.2008: Wiring of the solenoids and the 16 pole Weidmueller connector to the 16-output processor board. Oboe position fixed, eccentric on the axis. Carrier spot-welded on axis. Wiring of the sound pressure driver and mounting of the audio transformer in the chassis. All electronic hardware, with the exception of the visual light related features, is fundamentally ready and operational now. <Ob> is ready to accept the firmware for its PIC's.
• 09.10.2008: LED-strip lights (Yellow) mounted and wired to the midihub board. Construction and assembly of the pads for the open holes. Hexadecimal fingering tables (default lookup table) worked out for programming in the PIC controller. PIC-specs written out for programming the firmware.
• 10.10.2008: Waiting for Johannes...
• 01.11.2008: GMT implementation worked out. Still waiting for Johannes...
• 05.11.2008: No, <Ob> was not named after Barack Obama
• 11.11.2008: dsPIC programmed with version 1.0 firmware. The generated waveform should be a sawtooth wave. Midihub PIC and valve PIC as yet to be done.
• 22.12.2008: tests on sound production. First results are promising but the sound volume is now too soft. We either have to raise the voltage, or look out for another transformer.
• 24.12.2008: Further research on the sound production. Experiments with capacitors on the primary of the transformer.
• 25.12.2008: Price quote request sent to Sawter Transformers (UK) for custom made transformers to use for <Ob> as well as for other similar robots such as <Autosax>, <Korn> etc...
• 28.12.2008: Bug removed in fingering table. Fingering code for key pressure implementation added in GMT, including a new function to read and convert the fingering data chart into two midi compatible 7-bit numbers. Further experimental work on sound production. A single 2.2uF cap across the primary of the transformer seems to give the best oboe sound. It also increases sound pressure with at least 12dB. We removed the diodes across the windings, but this made no audible difference.
• 29.12.2008: Programming session with Johannes Taelman on the firmware for the valve control PIC. This is now working according to our lookup table. Alternative fingerings are possible using the key pressure command. This command also works when no notes are being played.
• 30.12.2008: Test session. The solenoids on some of the valves require a somewhat higher force.
• 31.12.2008: Test code in GMT improved. Tests for pure legato added.
• 02.01.2009: Start alignment works on the valves.
• 04.01.2009: Fine alignment of the three upper valves finished.
• 05.01.2009: Further testing and alignment. The notes 67,68,69,70,71,72,72,73,74 already work well.
• 06.01.2009: Lower valves aligned. Tuning checked: the tuning controller (nr. 20) must be set to value = 5 in order to tune the instrument to exactly A=440Hz.
• 07.01.2009: As a test, we orchestrated W.A.Mozart's oboe concerto (KV314), for <Ob> and other robots in our orchestra.
• 08.01.2009: Start recoding of our 'Primes for Cheops' (1988) for the new <Ob> and the robot orchestra.
• 09.01.2009: Expression controls added to the <Ob> part in Primes.
• 10.01.2009: News from Sowter Transformers is on the way...
• 20.01.2009: The Sowter Transformers build to our specs will cost ca. 180Euro a piece.
• 25.01.2009: Experiments and measurements performed on the existing driver: raising the power supply voltage to 25V brings the output level to realistic values for an oboe. Current draw is then 600mA, so primary power comes to 15W. The transformer gets pretty hot now. However, when we measure the voltage over the secondary (and the compression driver) we read about 1.3V rms. This would correspond to an output power of only 210mW... Hard to believe.
• 26.01.2009: Extra power supply module 9V/1.6A added in series with the existing 12V.
• 27.01.2009: Power meter circuit designed such that users can monitor safe volume levels on a small moving coil A-meter. We had a miniature in stock, Ri=464 Ohms, 100mW full scale (or, 215uA).
• 28.01.2009: Eagle transformer 40W - 100V line ordered from Farnell for tests and experiments. If we wire this component as: we can obtain a much higher sound volume from the oboe. The problem is that we do not have the physical space for placing it onto the robot...
• 30.01.2009: Experiment with the new Eagle transformer. This requires a 150nF over the primary in order to obtain a good oboe sound. The net result with the same supply voltage of 21V is not a bit louder than with the previous transformer and a 2.2uF capacitor, but now we could raise the supply voltage up to 48V without danger to burn out the transformer. We connected the driver (8 Ohms nominal) to the 15 Ohm winding because of the fact that the driver is coupled to the instrument with a capillary rising its impedance considerably. So this connection is neither a mismatch nor a mistake. The new 40W transformer was mounted sideways with silicone mounting kit to the chassis. The new voltage transformation ratio is 4:1 With the previous transformer it was 2.2 : 1 but it had a serious impedance mismatch with the load, the impedance of the secondary being only 3 Ohms. It should be noted that in the new circuit the 150nF capacitor forms a resonator with the primary of the transformer (inductance 159mH) tuned to 97kHz, way above audio, but very effective apparently to suppress PWM induced HF components in the transformer. This is presumably the explanation for the empirical fact that the sound is about four times louder with this capacitor than without it.
• 31.01.2009: Supply voltage for the audio driver raised to 37V. XP-Power module 24V-15W switched in series with the 12 V power supply. This raises the maximum obtainable sound pressure level to about 6dB above what a normal orchestral oboe will sound. Now the electric maximum power for the driver is 6W.
• 01.02.2009: Further impedance transformer research.
• 16.02.2009: <Ob> demonstration for the composition students from the Ghent conservatory.
• 10.02.2009: Waiting for firmware for PIC1: motor movement and light functions...
• 31.03.2009: Even Sebastian Bradt -after objecting deeply against the <Ob> project- now wrote an important part for <Ob> in his James Joyce rendering and orchestration...
• 02.05.2009: Programming session on the gesture control with Johannes Taelman. Implementation of the lights on de midihub PIC board.
• 15.05.2009: Gestrobo study for <Ob> coded by Kristof Lauwers.
• 16.05.2009: <Ob> demonstration in Technopolis (Robocup).
• 01.06.2009: First attempt to write firmware for the midihub pic and the motor control. Serious problems discovered with the polarity reversing bi-stable relay: double triggering does occur, with dangerous results for the instrument... Also, reversal of polarity on the motor leads to frequent glitches and resets of the PIC controller. We may have to redesign this circuit as well as the sensor mechanism. Lights on note 124,125,126 implemented and working. Frontal light 126 still has a bug. The white light on the oboe seems out of control. It seems to react to changes in controller 21 (motor speed pwm).
• 18.06.2009: Four quadrant linear output motor driver replacing the relay circuit designed under study. <Ob> premieres Daniel Pastene's solo piece in the Conservatory's concert hall.
• 01.02.2010: back cradle made for ease of transportation.
• 11.02.2010: Bug discovered in the ds-Pic code: note generation is not limited to midi note 96. The problem is that this now causes interference with the implemented lites on the midihub board.
• 12.02.2010: <Ob> is off to Hasselt for the Boink! festival there.
• 15.02.2010: We must have forgotten to document the wiring of the power supplies... reverse engineering required since we need a fully floating 24V supply for the new motor drive circuitry.
• 09-10.10.2010: <Ob> plays on the opening of the STAM museum on the Bijloke campus: 12700 visitors...
• 15.10.2010: Start design for new firmware for the hub and motorcontroller using the Proton+ compiler under Microchip MPLAB. This also entails the mounting of a better sensor mechanism.
• 16.10.2010: Testing of behaviour of Pepperl+Fuchs sensors. We found them not ideally suitable for the application. The sensing distance is too small (2mm) and the lack of analog output makes is very hard to design the steering software properly.
• 18.10.2010: Penny & Giles analogue tilt sensor ordered: type nr.STT 280/60/P2. We can read the analog voltage using a single 10 bit analog input on the PIC controller. Further design of the firmware with the motor control.
• 19.10.2010: New sensor tested and implemented in the firmware. New midihub board assembled.
• 21.10.2010: Further research on hardware and firmware to get a smooth motor movement. The Trident controller is tricky to use, as it insists on getting fully floating differential control voltages. Thus we had to modify the steering circuit to: This circuit generates a floating bipolar voltage between the X17-2 and X17-3 terminals. The 410 Ohm resistors are in fact each two 820 Ohm resistors connected in parallel.
• 22.10.2010: New board implanted in the robot and tested with firmware version 2.00. Movement works now although carefull adjustments of the parameters using the trimpots on the Trident motorcontroller is very important to avoid overshoot and oscillation problems.
• 23.10.2010: Firmware versions 2..1, 2.2, 2.3, 2.4, 2.5, 2.6... gradual improvements.
• 24.102010: Movement range still too small. Attempt circuit modification with a 270nF cap. between the mosfet drains. This is indeed better, regulation is not perfect yet. Finalisation of mounting of motor controller on the underside of the robot. A very small and short screwdriver will be required for trimming of the motor current on the Trident controller. Wiring cleaned up and finalised. The definitive circuit for the floating motor drive now looks like We are now at firmware version 2.7 for PIC1 on the midihub board. White LED spotlite added, mapped on note 4. Now, waiting for user feedback and comment...
• 25.10.2010: Firmware versions 2.8, 2.9, 3.0, 3.1...
• 26.10.2010: Version for the hub firmware is now at 3.11. It introduces an extra controller (CC23) to be used to set the movement speed and, more particularly to stop possible oscillations in the movement. After completion of a movement, the controller should be set to value 28 to stop any oscillations around the point reached.
• 14.10.2010: Rehearsal of 'Zwiep & Zwaai', a demo piece for the movements written for Dominica Eyckmans.
• 10.04.2011: Upgrade of dsPIC coding. Now range in the dsPIC is effectively limited to 58-96, instead of the former 58-127, causing interference with the lights.
• 19.11.2011: Glissando playing now fully implemented under GMT.
• 19.09.2012: Wheels replaced with sturdy ones from cast aluminium and strong red polyurethane tyres.
• 17.03.2013: Start construction of a sturdy flightcase for <Ob>.
• 18.03.2013: Further work on the <Ob> flightcase. It will leave the back wheels of the robot protruding such that they can be used for transportation with the case.
• 19.03.2013: Flightcase for Ob nearly finished. Still need to mount handles.
• 20.04.2013: <Ob> survived the trip to Glasgow very well! We expect him back in Ghent on monday 22nd of april.
• 23.04.2013: <Ob> is up and working fine again.
• 19.10.2015: Close check-up on <Ob>. A bolt on the chassis apparently got loose during the transport to Dok19 yesterday.
• 28.10.2016: <Ob> on the road to Brugge for 'Iedereen Klassiek'.
• 30.10.2016: <Ob> set up again in the orchestra and found fully o.k.
• 26.07.2017: The motor mechanism found blocked. Repair session.
• 02.08.2017: <Ob> on the road to Liepaja (Latvija). It survived the trip very well. We are expecting it back on august 10th.
• 09.08.2017: <Ob> returned safely from its trip to Liepaja.
• 05.11.2020: Live webcast on <Ob> via Facebook in corona lock-down times. We presented Daniel Pastene's composition 'Terminator in Afghanistan' from 2010.
• 02.09.2021: <Ob> is present in the exhibition of robots at the SMAK museum. It will play there until the end of september.
• 27.09.2021: After playing uninterrupted for 25 days, <Ob> returned to the robot orchestra in perfect condition.
  
• 22.07.2023: Na terugkeer van de tangoproduktie in het Baudelopark, bleek een aansluitdraad van de bovenste Laukhuff magneet afgebroken te zijn. Oorzaak: onvoorzichtig overschuiven van de transportkist. Herstelling uitgevoerd en vastgezet met epoxyhars. De spoel vervangen kan niet meer, want Laukhuff is inmiddels ter ziele gegaan. De klepjes en zowat het hele mechanisme van de hobo blijken wat zwart te zijn uitgeslagen. Is het verzilverd? Kunnen we dit verhelpen zonder de hele robot te moeten uiteenhalen? Het wordt hoe dan ook een werkje waarvoor heel wat geduld nodig is... Voor de klank maakt het evenwel niks uit.
• 14.09.2023: Vertrek <Ob> en 49 andere robots naar de Deutsche Oper in Berlijn.
• 18.09.2023: Lossen en opstellen robots in de Tischlerei. <Ob> in goede staat bevonden.
• 25.09.2023: Aankomst Godfried-Willem Raes in Berlijn. Grondige inspektie van alle opgestelde robots.
• 27-28-29.09.2023: Produktie 'Zeroth Law' voor Deutsche Oper Berlin i.s.m. Gamut Inc.
• 30.09.3023: Laden vrachtwagen LuxTax voor terugtransport vanuit Berlijn naar Gent.
• 02.10.2023: Lossen vrachtwagens. De kist met <Ob> bleek omgevallen te zijn op de laadklep en <Ob> is erg zwaar beschadigd. Het chassis is helemaal verbogen de schakelingen hangen los en de konnektors zijn afgebroken. Het draaimechanisme is helemaal verbogen... Ook de hobo zelf was helemaal van zijn drager losgerukt. Het moet wel een heel erg harde klap zijn geweest. Foto's van de schade gemaakt.
• 07.10.2023: Hele dag gewerkt om <Ob> te herstellen. Spelen doet hij echter nog lang niet...
• 08.10.2023: De 5V voeding blijkt niet meer te werken... Kortsluiting of spoorbreuk op een van de PCB's? Bleek een kortsluiting te zijn veroorzaakt door verstrengeling van de aansluitdraden aan de konnektors. Alles nagezien en opnieuw goed vastgeschroefd en <Ob> bleek weer te werken. Enkele vilten polstertjes waarmee de kleppen worden ingedrukt bleken ook afgevallen. We vervingen ze door nieuwe exemplaren. <Ob> is weer up-and-working.

(Terug) naar logos-projekten: projects.html

Terug naar Logos' index-pagina: index.html

Naar Godfried-Willem Raes personal homepage...

Naar katalogus instrumenten gebouwd door Godfried-Willem Raes

M&M orkest

Last update: 2023-10-08 by Godfried-Willem Raes

The following information is not intended for the general public nor for composers wanting to make use of our <Ob> robot, but is essential for maintenance and servicing of the robot by our collaborators. It also might be usefull for people that want to undertake similar projects. Feedback is mostly welcomed.

Technical drawings, specs and data sheets:

Power supplies:

• +5 V DC - 250 mA(Logic and microcontrollers), Erea 8V transformer coupled to the on board regulator situated on the midihub board.
• +12 V DC - 5.1 A (Solenoids & sound driver) Linear power supply module Power One type HN12-5.1A. Adjustable from 10.3 V to 13.5 V. Used for the solenoids, the lights as well as for the sound driver.
• +24 V DC - 15W XP power module type ECL15US24-E, used in series with the above power supply.
• +15 V DC - 670 mA(Mosfet drive for sound driver) Integrated SMPS module. XP Power, model ECL10US15-E
• +24 V DC - 625 mA(Motor drive). Derived from a toroidal transformer, rated 15 VA. (must be kept floating!)
• 
  

Wiring & circuit details midihub board:

Circuit details solenoid driver board:

dsPIC board - compressor driver:

Wiring for the dsPIC-board: This is the same board as also used for <Aeio>, <Hurdy> and <Bono>.

Motor Specs:

Crouzet 80807001, 24 V, rap. :125, motor 3000 rpm. Geared 21 rpm. Farnell order nr. 308-9514. Current consumption without load @ 24 V: 300 mA. Lowest voltage required for rotation: 2.6 V. Price (01.10.2008: 217.90 Euro).
Toggle relay: ESDI type nr. NF-8250, 12V coil, contacts 16A 250V ac. Purchased at RadioHome Electronics, Ghent. [ removed 18.10.2010]

Tilt sensor : Penny & Giles (STT280.60.P2) Data sheet. (PDF) [cost: ca. 250 Euro]

Mechanical construction drawings and welding plan:

• frontal projection:
• 
• Ground plane projection, wheelbase:
• 
• All chassis parts made of AISI 304 stainless steel. Wheel axis as well as the rotation axis with the chain wheel are made of hardened steel ST54. Welding was performed using the manual TIG process with intermittent cooling using compressed air.

RCA-driver:

Tweeter horn driver TW15W 8 Ohm. Cross over frequency by design: 800Hz. Peak allowable voltage: 10.9V, to be derated with 12dB/octave for the frequencies below 800Hz. Price in 1982: 2.360 Bef. (ca. 60€)

Yellow LED strips: LM-FB26Y-12, Farnell order code: 122-8819

Oboe details:

Builder: F.Debert, Bruxelles. Model C60. We have no indication as to the year of construction, but since the tuning conforms to A=440, we suspect it was made not much before 1939.

Download high resolution pictures of this robot:

• <Ob> picture 1
• <Ob> picture 2

Download high resolution robody pictures of this robot:

• <Ob> robody 1
• <Ob> robody 2

References:

Bartolozzi, Bruno, 'New Sounds for Woodwind' , Oxford University Press, London 1982 (1967), ISBN 019 318611-x

Benade, Arthur H., "Fundamentals of musical acoustics", Dover Publications Inc., New York 1990 (1976). ISBN 0-486-26484-x

Goossens, Leon & Roxburg, Edwin "Hobo", ed.A.J.G.Strengholt's Boeken, Naarden, 1977 (1980), ISBN 90 6010 504-4 (This book contains the fingering tables for the oboe).

Raes, Godfried-Willem, "Expression control in musical automates", 1977/2023,

Smith, Bob H., "An Investigation of the Air Chamber of Horn Type Loudspeakers", in: The Journal of the Acoustical Society of America 25, 305-312 (1953); https://doi.org/10.1121/1.1907038

Weijers, Rembert, "Rieten ontwerpen en maken", Muziekuitgeverij van Teeseling, Nijmegen, 1978.