data_opera prototyping a dust_data based analogue soundmachine #2
posted by lupo on 15 December 2013
Day No 5 24/11/2013 The walk for a new hairdryer and some other needy stuff to solve the problem with transoprt of air into the box situated in the lab. The Box for the sesnor array in the background mounted analogue synthesizer The box from above This a day of sayings. First what you buy cheap you often need to buy twice. Kilpishalli K-Market shop has literarlly almost everything you need, even a cheaper, better hairdryer with a mode with out any heating involved and additionally a mode for cold shot, compared with the considered a good buy in Helsinki. Kilpishalli the incredible shopµm 3 hours after the I left the staion I'm back, happy after a nice walk and lucky about I got the things I need. Day No 6 25/11/2013 Mounting the new hairdryer with new and testing air flow in complete setup for long duration test. Meanwhile sensor casing and tests with filters and datalogging. This showed, that the air is really clean, very low concentrations within the fine dust particle size. Broadband Filters, 3 to 15 µm work, but Filters 3-5 µm, 5-8 µm deliver very low concentrations and even longer periods without any detection, 8-11 µm and 11-13 µm  work show higher concentrations but as well deliver periods of zero detection. That needs to be handled  in the code for measurement data transfering to voltage output. First tests to drive analogue synthesizer with voltages generated via the arduino works good, but shows the known ripple problem of PWM analog voltage output. Tests with simple lowpass filters show that the audio lowpass filters of the synth work better, as the frequency can be regulated and the design of those is better and meeting the needs for sound manipulation. The need for 30s measurement cycles can be handled with slew controler and controled slew limiter. Day No 7 26/11/2013 Works with the code takes a while, especially Marcus puts a lot of time in debugging the code, further sensor housing and optimizing for the problem of partikels remaining on the lens is second priority of today. In the evening further tests with the system running, but still without air from outside, as it is very windy and lots of snow is in the air. All that snowing that goes on cleans the air even more, so this is a day to stay in the lab and tests are run with indoor air, which is very low polluted too. As the Ardunio does not deliver enough power to drive at least 4 or 6 parameters needs darlington arrays, these where built on breadboard for test purpose. Tests showed unexpected behaviour, producing negative voltages, more strange ths did not happen with basic sketches used for frist tests, like sweep generator. So this must be the code, that leads to that. Day No 8 27/11/2013 Markus Decker takes a look at the code and solved problem, which was that levels and level calculation was done in dezimal instead of hex.  Now it works perfect. So no more obstacles to test the system first time in a sounding setup. Voltages where used to unharmonize a harmonic setup of three voices via filters (filterfrquenca and resonance), modulating pulsewith and fine-tuning of two of the three voices and a voltage controlled variable LFO that modulates mor filter parameters and FM of one oszillator. It sounds nice but the signal change according to the dust concentration is still a problem, as the concentration is changin rapidly but remains low. So darlington where driven with 6 V instead of 5 what led to better results. The nice experience is, that it even works with low concentrations, so higher concentrations easily found in urban spaces will give much more range to play with in terms of sound shaping. First recordings will take place tomorrow and I'm lucky that as it sounds now, I can finally get rid of the beard! The whole setup during tests Day No 9 28/11/2013 Now as it looks like main things work the next step was to set up at least two sensor-cases each with different filters in the box and to connect the outputs via the darlington arrays to the analogue synthesizer and use at least four parameters straight forward and to use in additon at least two 2nd order parameters to shape the sound according too dust concentration and ratio, both measured in each sensor house in 30 second detection cycles. Due to the long detection cycles some slew limiting seemed to be necessary to smooth the changes between the current and the older measurement data. Further some cables had to be soldered, the sensor housings had to be re-arranged and as I found out not only the incoming air-flow of each sensor house needs to be filtered, but as well the out fans too. Done all this and some tuning of the non influenced synth - four voices involved (two quantized at musical / chord laws and two without quantization) I started to work with the incoming data. Two voices pitch and some other parameter are controled directly by the incoming voltages 0.1 - 6.0 V and in addition two VC LFO's get controlled by the incoming data of particle concentration. These LFO's then control CV controlable parameters of Filters, Pulsewith of the Rectangle Voices and FM of one voice. As a further sourde of 2nd order data results of the analog comparator, which compare the dust concentrations of the two different particle sizes (12-15 µm and 5-µm). This leads to results which definitely react on changes in the air quality, but this is not yet meeting yet my audible / musical expectations, while I'm still happy the system works. Day No 10 29/11/2013 Some search for math implementation into Arduino is in the first moment a backdraw, as I'd need faster AT-Mega boards and even that seems to be unlikely to work, as there would be needed to use the whole AVR Lib which exceeds the memory of the boards. But finally as it is mainly needed to transpose the linear data into logarithmic data logarithmic amplifiers based on op-amps will be the solution, unortunately these are not available here, so I stop further detail development of the system and will use more thinking on a new layout of the setup, as I realized, that decentralized measuring via split single sensor stations connected to the central soundgenerator - the analogue synthesizer is the solution to some problems not easy to be solved in an sensor array put into one box. Maybe this will make it possible to get rid of the fans or at least get rid of the need to have in and out fans for each sensor. Further I see one more benefit of this, it makes it easier to make interactive or at least answering setups for exhibitions, as visitors if they move through room(s) opening doors etc. produce disturbance or even turbulence within the airbody and therefore transport particles to teh sensors and finally they bring particles in as they enter the space and as everyone of us is a carrier of dust. So far so good. Here a short excerpt from test recordings: [soundcloud url="https://api.soundcloud.com/tracks/122470279" width="100%" height="166" iframe="true" /]