When thinking about where to depart from for our workshop at the University of Lisbon in autumn 2023, and during the initial conversation with Elena, I returned to this “study” (prototype, trial, unfinished piece) from 2019. Not by coincidence was it developed during another workshop, Sensory Cartographies, that seemed related to our endeavour for Lisbon. I am now thinking of building a new version, composing a new configuration, focusing (perhaps) entirely on the sound bit, and developing a new antlers based pair of helmets. The process will be documented here.
{hhr 240102}
First design of an antler made of a number of segments in OpenSCAD. The segments can then be separately printed, and one has to find a way to attach them. For now, I'm printing some inlays for the piezos, and connecting pieces which are glued, then screwed.
{hhr 240127}
First test with the TalentCell LiFoPO power bank LF4100. Runs a Pi 3B from its 5V USB output, and a small amp off the 12V output. Occasional undercurrent warnings in the Pi, but might be due to the too thin cable; perhaps better to use the 12V output with a DC Buck converter? My old RavPower power bank is seen in the top left. It has a nominally larger capacity, but in reality it weighs a fraction of the TalentCell. It's main issue was that the USB output had too little current.
{hhr 240127}
The first OpenSCAD design was done quickly, and with no regard to minimising the angle of the segment cuts. Surprisingly, the 3D printer had no trouble printing at more than 45° angle.
{hhr 240127}
Since the beginning of the year, a lot of the “development” of the piece happens in my head, owing to a busy agenda in which the project is only one part, and thus thinking “between” spaces, between activities. That poses a challenge for documenting the process, because it is so much easier to just upload the occasional photos from the ongoing building process. A way out may be the piece-wise adding of thoughts, things that can be formulated in ten or fifteen minutes, from time to time, without extensive editing.
Having built the first horn based on a slightly flawed 3D model (requiring glue between segments and placing the piezos in an angle that makes it difficult to fasten and wire them, difficulty in fixing the screws between segments), printed with remainders of duo-coloured filament, a was able to create a second model that is much better constructed. Each segment is now fastened to the next using three screws, and there are conducts that allow to fasten the screws from the sides. The new horn is very stable, and I began testing the piezos. As of this writing, I attached the three larger piezos towards the ear side. Their sound is a bit disappointing at first, since they are not are attached to any conductive surfaces, by themselves they are very narrow banded in the high frequencies. But I don't want to give up on them and flip them for small coil-based speakers or transducers, I still like their flatness and their brass appearance. Perhaps in a future version, they can be attached to vibrating surfaces, thin ceramic, glass or metal inlays. For now, I want to continue and see if I can make an interesting piece that relies on this faint high frequency sound. The paired and iron-wire-wrapped copper leads to the piezos look great, the only worry is that their stiffness can exert pressure on the piezos themselves when mounting, risking to break off the piezo; and it's still unclear if it will be feasible to thread them from the outer segments all the way to the horn's base.
I'm thinking that there will be two more leads through the horn; one for an LDR sensor, and one for the microphone. I found a hopefully flat enough microphone cable that will arrive soon. The LDR could be used to guide the balance between the piezos. I'm now thinking that since the sound is so faint, I will pair them only in two parallely wired groups—an inner and an outer group, requiring thus one mechanical relay per horn, or three relays for the entire person. I think the microphones should have a kind of parabolic dish attached to them, focusing their otherwise omnidirectional pickup pattern. Perhaps one of the “double horn" on one side could be spiralled back towards the person's face, thus allowing them to speak or whisper.
I am also thinking about the simularr interval. It appears to me that this piece serves a similar purpose as the partially assembled rogues last time. In the sense that they could become research devices, and I don't mind leaving their final shape and composition open for now. If I manage to print all parts for two persons, we could engage in a few duo situations of experiments and conversations. I'm also seeing the amount of work that Emma (Luke) put into the first iteration of this piece, there are a lot of small details on the helmets that show her skills with fabric and wearables, skills which I lack myself. So it remains a challenge to assemble the whole new piece this time, from creating the “rucksack” of the electronics, to the connectors, to the chin strap etc.
Perhaps I want to go back to dream diaries, something I liked to to in Almat. I had a very interesting dream last night, but I didn't prepare a scratch book to record it, and now I am too busy to take the time to reconstruct its details. Perhaps they could become part of this piece?
{hhr 240206}
View of first segment from perspective of the ear, showing two brass discs of the piezos.
{hhr 240127}
A recall (from memory), how the first version worked. The technical components page lists the setup. Unclear which microphones were used, perhaps some AKG small condenser—die OM1 was not in my possession at the time. It says the USB stick audio device with plug-power microphone was too noisy, but again this might have been an issue of the microphone; would the Edutige ETM-001 be “ok”? It has around 18 dB worse noise floor than the Line Audio OM1. There are these 48V-to-plug-power adapters (e.g. Røde VXLR+) that could be used to test the ETM-001, which is still 30% of the cost of an OM1.
The RAVPower was definitely a weak link with its 12W supply, the Pi 3B crashed a lot, though the UMC22 is only rated 2.5W draw. Perhaps the HDMI screen added to it. What I didn't try is inserting a DC booster to stabilise the 5V (or go slightly above).
If we go for a 4x4 channel configuration, there are two possibilities: Make each 'antler' independent, i.e. one Pi Zero 2W with cheap USB interface, ETM-001, one channel amp. Or make it centralised, one Pi Zero 2W or 3B with Roland Rubix44, 4x ETM-001 with P48 adapters or 4x OM1, one Auna C300.4. While the asynchronous version is attractive, it may also be the technically more demanding; unclear if the mic-pre of the 1x2 USB devices are good enough, then eight devices on the Wifi etc. Another question is whether to stick with piezos or use the Dynavox PS-138 instead. They might be too heavy (times eight or times sixteen).
{hhr 240104}
Microphone choice. Both the Line Audio OM1 (phantom powered) and the Edutige ETM-001 (plug-in powered) are quite small, the ETM-001 indeed is very small. However, as I'm planning to use three or four per node, an audio interface with multiple mic inputs is required, and to my knowing this requires one with professional XLR inputs. For the ETM-001, one would have to insert a 48V-to-plugin-power adapter, like the VXLR+ depicted.
{hhr 240127}
Subsequent test with the power bank and all three loads connected: Pi 3B on the 5V output, Rubix44 on the 9V output (attention has to be given to the reverse polarity that the Roland uses), with one OM1 connected and phantom powered, and the Auna amp on the 12V output. Directly connecting the line signal creates a horrible noise, but inserting the Aukey anti-hum transformer works as always. Today I don't see undercurrent warnings on the Pi, which may be due to a better USB cable, or due to the fact that I chose a different Pi (with a different software image).
{hhr 240127}
Growing an antler. Left-most at the workshop in Lisboa, trying to create a sequence of piezos directly sustained by a reinforced wire. Next printing and mounting one PLA segment after another. The colours are random in that I was using remaining filaments, it just turns out to allude to a rainbow unicorn.
{hhr 240127}
I tried to use the OM1. Even though it is small (75 by 20 mm), the XLR cable plug adds significant space. I designed a holder in OpenSCAD with a slit for the cable. I want to reduce the length in a future version. Perhaps one could take out the circuit board from the OM1's casing, or change the regular XLR plug for a smaller one.
{hhr 240127}
Quick thought registration. How much current do the mechanical relays draw? An Enantiomorph (with), were the sound are spaced with relays along the antlers, for instance reflecting on the co-rhythm of the two wearers (IMU ?). Do we then need to multiply the discs as we move away to account for volume loss?
{hhr 240104}
