things have seemed a bit quiet over on the research side of underscores lately – and not for lack of a (growing) backlog of interesting diy projects that we want to release !
introducing Polyphase Portal
its partly because i’ve been putting more time recently into an exciting side-quest: polyphaseportal.xyz – an educational focused online space and media-art collective
together with some friends, organizers and educators we are creating a new online space to host educational programmes such as talks, live-streams, workshops and multi-week courses. taking encouragement from the success of PHASE SPACE in NYC we are aiming to facilitate accessible non-hierarchical learning over open and ethical technology – find out more at polyphaseportal.xyz & join the polyphase mailing list to stay in the loop!
which brings me to the next announcement – im teaching an online course !
you can read more about it and join the class wait-list on the Polyphase Portal here
no prior experience or specialized equipment is needed – this course will focus on the practical side of building modules on breadboards – you will receive a course materials kit in the mail that includes all the parts and components we will use.
the course makes use of this custom vga_breadboard_module pcb to interface between vga signals and the breadboard.
one day i hope to make this circuit and the corresponding course material kits + module designs available in the shop as stand-alone products you can build and learn from independent of this online course – but for now if you are interested you should sign up – it will be fun!
since the underscores soft-launch there has been a few updates to the recurBOY project:
the recurBOY circuit depends on more harder-to-source components than others in the underscores catalog. This is because it was originally designed only to be built together in group soldering sessions – where we would source all the parts for the workshop beforehand.
Now that it is a fully open diy project (with some people sourcing the parts for themselves) I have updated the build guide with more information about selecting these parts
recurBOY uses a 1.8 inch tft display (ST7735S) in a form-factor commonly found on ebay/amazon/aliexpress – unfortunately this is not compatible with the waveshare version – who i have found to be a reliable supplier of display hardware in the past (perhaps a future hardware revision should somehow allow this version of the display to be used also ?)
based on some feedback from users + my own exploration of testing these displays, I found that some of them were not working on the recurBOY r0.6 pcb. However when i tested the same (faulty?) displays with pin-jumpers wired to the pi directly they would always work!?
A laborious process of trial and error discovered that in particular the SDA pin ie SPI data input between screen and rpi was the problem – soldering a floating wire to this was enough to consistently fix this issue across all the “faulty” screens i tested.
I couldnt really see any difference to the signal before and after this “fix” on the scope, but my best guess is that the trace on the pcb is a bit too weak for some screens to read it.
I updated the pcb to make this trace much bolder – bumping circuit from v0_6 to v0_7_1 (v0_7_0 fixed it also but accidentally broke a trace on the rca out – whoops) and now every screen that I have tested works
all of this could have been avoided if we had selected a more reliable part from a reputable supplier – however this could incur also a large increase in price for the project + ( potentially ) more difficulty for diy-er to source in small batches.
for as long as we design with hacky parts occasionally we will need to make hacky updates – and i stand by this decision !
open projects have the flexibility to adapt to their changing surroundings and this is a good example of that!
recurBOY was designed to distill some of the best parts of a r_e_c_u_r video sampler while running on cheaper hardware (a rpi_zero) and keeping the interface simple.
one feature of r_e_c_u_r that i use often and miss with my recurBOY is processing live-video feed (eg from a camera or another video-synth)
processing live input
running input from a piCamera over csi into recurBOY proved to be too much for the pi_zero to handle. To my surprise however it could seem to handle input from a usb-video device like EasyCap capture cards
i cant guarantee it will work with any usb-video-input , but now if recurBOY detects a device on startup you will get access to a third source mode that allows you to preview that input and apply effects to it !
img file & other updates
you can download new image file and see the full list of changes on the v2.1.0 release page but in short, a few of the other ‘quick wins’ i also added in this release:
usb_midi support – including independent shaderParam and effectParam controls
sampler mode now plays images – jpeg/jpg, png & gif (single image only) supported
output sd over hdmi – defaults to 640×480 and seems to handle this without lag
please let me know if you have any issues or find bugs in this release – it has only been tested by me so far (i will keep old image up too if you need the stability)
when looking for things to update i rely on the issue tracker on github – no promises but putting your suggestions here is the best way to have them noticed!
the start-up input current of the buck-boost converter is higher than the standard buck-mode regulator, and this may overload an input power source with a current limit less than 1.5A. Using a delayed turn-on or an undervoltage lockout circuit (described in the NEGATIVE BOOST REGULATOR section) would allow the input voltage to rise to a high enough level before the switcher would be allowed to turn on.
however my various attempts to use delayed turn-ons / larger input capacitors failed to correctly bootstrap the circuit with these regulators.
i found online a description of a problematic similar sounding method to generate +/-12v rails for eurorack that seems to use fake buck-boost regulators. Im not sure if this is in any way related, but still.
im always game to try out cheaper parts from questionable sources – in many cases they can be just as good for a fraction of the price. but in this particular case, given:
that in general you shouldnt cut corners on power-supplies
it failed for me in a way that could hurt people
there is reports of related issues using similar parts online
Im happy to stick with the more expensive mouser sourced regulators !!
this is the first major revision I have done to an underscores circuit that is in production. it is a good opportunity to think about the process as I want all these projects to be living / always improving.
I created a branch in git for this release and committed all the changes over the last few months to this feature branch – this ensures the changes and revisions are available yet also preventing people downloading WIP gerbers from the main page.
when everything is tested and ready to go i can then merge this feature branch into main.
on release i also need to update the project in a few other places:
the shop description
the circuit booklet
the circuit images (in this case decided not to bother with these)
the video walkthroughs – videos.scanlines is fine but youtube will not allow for editing existing videos so i either need to re-upload a new one or find a way to amend the existing one
tl;dr – im thinking of making a revision of sync_ope to replace the ac-ac psu input with a standard switching ac-dc one…
UPDATE: this has been implemented in v1_0_0 – read more about it here
sync_ope is a project iv been collaborating with Gael Jaton on for a few years. initially a concept dreamed up and tested on bread-board, iv worked on a fair few design iterations based on feedback and support from the wider community – we presented a prototype of the concept at FUBAR_2020, and continued improving it via discussions on the scanlines forum.
its been quite slow progress – and like many diy / community driven initiatives (who get passed around and only prioritized in short bursts & on productive weekends) has some built-in identity insecurities:
sync_ope is designed to be diy & beginner friendly – spaced out across all of a 100x100mm pcb- with footprints for even the specific video-ic’s in DIP packaging
the use case however requires already owning analog-glitch devices, and having pre-established capture flows where sync dropouts are a problem
the circuit is strictly analog, and more complicated/advance than many of the other ‘entry level’ underscores projects iv been prioritizing – one LT1251 ic can cost more than a single-board-computer!
Many similar projects in this realm would have fully committed to serving (only) the eurorack video synthesis community – where there’s an active base of designers and users creating complex functions with analog circuitry:
the advantage of this is you don’t need to (re)design common function blocks like power supplies and enclosures – there is even many design best-practices to follow to ensure compatibility
the problem i see with this approach is that it can be off-putting for some people who havnt fully invested in the modular setup – where certain design preferences and restrains do not apply to them. by only serving the established (and well-enough off) we risk alienating those on the fringe.
the beauty of the sync_ope circuit is it can be just as interesting and useful in a mess-of-circuit-bent-mixers&diy-guitar-pedals as in a large modular synth system
technically, the circuit requires a dual power supply to operate (unlike some of my other projects which can be run from a single usb-supply)
this is common and expected in euro-rack-land where all systems carry +/-12v rails. however it is less common for the average ‘non-modular’ hardware setup…
the current design has both a eurorack header– that takes +-12v straight from the rails – and a 12-15 AC barrel-jack input – that uses a Rectifier Diode Bridge to convert alternating current to dual dc.
i thought that this implementation struck a balance between serving those who may want to rack-mount it (atleast, all the interface controls are front facing!) and those who will just wanna plug it into a wall.
however as i prepare to distribute a small number of these circuits internationally im discovering that AC-AC wall-warts can be quite tricky to come by.
a different approach that i have been seeing recently is to start with a single dc supply and use a step-down regulator in inverting buck-boost configuration to generate the negative rail.
since composite video sits around the 1-2v range +-5v is plenty of headroom for this circuit to operate. we could start from any single dc 12v wallwart (or the positive side of a eurorack-powerrail) and create the dual 5v’s we need onboard.
although the (ac-ac) current version of sync_ope is available in store now im already thinking about making another revision to the board to use single dc supply instead.
it could follow a similar configuration to Bastiens psu schematic below – instead using LM2575T-5.0 to create the +-5v rails we want. this change may increase the total BOM cost slightly – but since its already quite an expensive circuit i also think its worth ‘doing it right’
i can already imagine another future revision with smd parts and a more compact ‘rack first’ interface design that would better suit the modular crowd. but for now i will continue to try bridge the gap between this and a more hacky diy approach.
if you understand the ac-ac psu problem im describing and want a sync_ope now anyway then this version is for sale right now in the shop! but if you think you would rather wait some more months and hold out for an ac-dc supplied version then this is also an option for you too !
the beauty of open source is multiple forks of a project can co-exist. time will tell whether this revision will be a direct successor of what we have now or an alternative version
the idea was to try focus on bringing batches of 3 projects to market – avoiding my usual mistake of taking on too many concurrent ideas, but still benefiting from batch efficiency / parallel processing
this would also open up feedback cycles for the general processes before getting too far along in any one direction
as it turned out a number of setbacks delayed the shop launch:
covid related shipping disruptions early this year made distribution from nz near impossible
recurring issues with the sync_ope circuit caused me many revisions and much confusion
i was away from my studio for 3 months over nz winter – traveling in australia and then visiting berlin
more the merrier
although distribution needed to be put on pause, this didnt stop my research and development ! as others projects approached a stable state i decided to add more and more to the launch line-up
the final result is an 8 product launch (although some of the more recent additions still needing proper photoshoots and demo videos)
one main advantage of this is the ability to combine shipping on mulit-orders. it works out to be a similar price to ship 3 products as it is for 1 (this calculation based on weight is built into the shopfront) so with more in stock at once you will have more options to combine and save money !
silent launch ?
im calling it a silent launch because at first there will be very little announcements regarding this. ultimately i plan to try keep everything more or less available (restocking projects in batches of 10) but currently i only have parts for between 3-8 of each project.
while managing such a large inventory i want to avoid being overwhelmed at first – i will stagger the announcements of the shop being open to allow time for feedback from early adopters and any adjustments needed to my workflow – also these initial orders can help fund the 8x10_bulk_part_order for a future full shop restock