A slime mold has been growing in my portfolio πβπ«
Wanna play with it?
Check it out at:
π inigoliz.com
Posts by IΓ±igo Lara
This seems like a fairly permissive software license π
Check out this cool project I did over the weekend turning a CRT TV into an oscilloscope for music visualization!
πΊπ΅π§ͺ
bsky.app/profile/inig...
I'm a bit tired so I think it might be time to wrap up this hack weekend!
It has been fun π
I'll leave with a series of pictures showcasing the cool effect that happens when I turn off the TV π
To properly calibrate the system I need:
- An amplifier that keeps the DC bias of my signals.
- Two amplifiers, one for each axis, to correct the squeezed signals.
I'll stop by the electronic workshop this week to get new materials, since I don't have the parts at home.
I also noticed that the drawing was always appearing in the center, even when it was supposed to be on a corner.
I noticed that the DC bias of the signal was being removed by the D-class audio amplifier. That caused the picture to bounce back to the center.
However, the system needed some calibration:
If I'd play a music track supposed to draw a square Lissajous pattern, the Y axis would look squeezed in comparison to the X axis π
I finally played some music and... The dot was moving!
Nothing too beautiful, but it was working.
Now it was finally time to draw some music!
I got a headphone jack and cut the cable, getting access to the L, R and ground wires. I connected them to the audio amplifier.
The deflection coils of the CRT are inductive loads.
In an inductive load, the impedance is proportional to the frequency of the driving signal.
As a consequence, decreasing the frequency of the sine wave in the signal generator increases the size of the oval being drawn.
That's cool!
And there it was!
It's not a perfect line because I guess that the amplifier is introducing a small phase shift. But close enough!
I got my signal generator and made a since wave at 500 Hz. Not much thinking behind the parameter choice.
I fed the signal into the L and R inputs of a D-class amplifier that I had around, and connected the R output to X and the L output to Y.
I thought I could start by drawing a diagonal line on the screen.
If you'd played with Lissajous patterns before, you'll know that a line can be drawn by feeding the same sinusoidal signal to X and Y.
That seemed easy enough to start with.
Alright, where did I leave it yesterday?
I had gotten access to the X and Y deflection coils. With that, I can start drawing pictures to the screen.
Interestingly enough, if I zoom into the line being drawn, I can see the colored phosphorous strips that yield each color.
Likewise, if I unplug the X axis, I get a vertical line.
And if I unplug both the X and Y axes, I get a dot in the center of the screen π
So far, I have identified the controls for the RGB channels.
Next, I want to identify the deflection coils: one for the X axis, one for the Y axis.
The process is simple: unplug cables, check what changes, try again. Finally I got them.
If I unplug the Y axis, I get a horizontal line:
The Cathode Ray Tube emits electrons using thermionic emission.
It works kind of like an old-school light bulb π‘:
A filament heats up and emits an electron cloud which is then accelerated by a high potential.
Actually, the glow of the filament can be seen from the end of the tube π
The TV is a Philips color CRT from 1980 made in England.
Since it's a color TV, it has three coils, one per color.
It was fun to discover three potentiometers that allow adjusting the intensity of each color channel.
By tuning them, I can get light of different colors π
I discharged the CRT tube. It was scary, since I had seen so many warning about doing it.
With a bit of care, it was fine. I'll better get used to it since I'll have to do it several times as I reverse engineer.
The spark was pretty big though.
First look inside: dusty mess.
I found it on the street. I wonder what stories of the past this piece of hardware has seen.
When I record the screen with my phone, I get this funny MoirΓ© effect that shows the refreshing rateπ
It's hack-weekend π
I'm going to turn this vintage CRT TV πΊ into an XY oscilloscope to display retro futuristic images with sound.
Tag along as I post the process!
This video about the universe, its future, and its dynamics at different timescales was wonderful to watch:
π₯ youtu.be/uD4izuDMUQA?...
Google giving reassurance about ML
Thanks, Google, for the reassurance π
I guess it's kind of true?
π ai.google.dev/edge/litert/...
Open neuromorphic
Feels good to get promoted at Open Neuromorphic π
Wait, you don't know what I'm talking about?
β‘οΈ We built a compiler that maps arithmetic operations into spiking neural networks for executing the computation.
It computes without all the heavy Von-Neumann stack!
neucom-aps.github.io/axon-sdk/
π± So cuuuuuuuuuteeee π»
I finally understood intuitively Floating Point representation π₯³ (FP16, FP32, FP64, defined by IEEE-754).
The dynamic ranges, the varying precision, all make more sense with a mental picture.
Good resources:
π newsletter.maartengrootendorst.com/p/a-visual-g...
π flop.evanau.dev/arbitrary-fl...
I'm radicalizing against techno-solutionismβ
- Water faucet? -> πͺ
- Replacing it with an infrared-LED proximity sensor that is hidden below the mirror so that the toilet looks cool? -> Ohhhh yeah π
The result? Having to look below to see where that goddamnit sensor is.
Let's do tech right!
This weekend, Turing patterns and Lilo πΈ
Playing with my new toy:
A finger-sized Digital Logic Analyzer.
It's muselab's one, and it uses Pulseview's open-source visualization software.
π 8 channels at 24 MHz for 15 β¬. Not bad!