Holiday Trinkets

I've been messing around with ring oscillators and decided to make a little holiday-themed trinket to give out. It was a good opportunity to get back into PCB layout and to learn how to use LTSPICE to simulate the circuit.

I originally wanted white LEDs, but with their high voltage drop and the 3V of the coin cell, they would never light up. In the end I had to use red LEDs with their Vf of ~1.8V. (This is when a green soldermask is more fitting for the occasion and better complements the red. However, the purple still looks amazing.)

Snowflakes!

Backside!

Of course, since I didn't think there could be an error in layout after checking that the schematic was right, there was an error. An error I found a few days after sending off the design.

Turns out the pin out for the default 2N3904 in KiCAD differs from the SMD MMBT3904 (which I really should have checked the datasheet for). The base and collector pins have swapped pin numbers, so when I solder up these circuits, I will have to flip the transistors upside down. And I thought I would have learned from the nixie tube clock. Oh well.

Ray Tracing and College

Working on a ray tracer has been quite fun, especially when debugging it when the image doesn't come out right.

I don't think that box is supposed to be cut off.
This was a bug with transforming the bounding box with a rotation.

It's mainly based off of Peter Shirley's raytracer in the Ray Tracing in One Weekend series of books, just implemented in Java. It's certainly not fast, but that's not really a big issue since I'm treating this renderer mostly as toy to learn more about the fundamentals of ray tracing.

Teapot! This uses a rather inaccurate model of glossy surfaces
but looks nice.

Reverse Polarity

A quick update during AP testing. I was messing around with a gear motor I got a while back (and used before for the barn door tracker)...

...and destroyed the encoder. In preparation for another project, I was working on creating a controller to regulate the speed of the motor and how much the shaft rotates. (Closed-loop speed and position control)

Dead.

Despite remembering each other time I used the encoder to check the pin out on the datasheet, I decided it was fairly safe to assume that the red wire was for VCC, the two wires in the middle for the encoder outputs, and the last wire for GND.

Big mistake. Turns out the red wire was GND and what I thought was GND was VCC. Add that to no reverse polarity protection, and you get a dead encoder. Bad on my part for not checking the datasheet again, but why is the GND wire colored red!

At least I got the motor/encoder combo used for only $20, and I can still use the nice Globe motor and gearbox. I looked up how much the encoder was worth new, and it was ~$50! Those 500 PPR and signal filtering don't come cheap. Oh well.

BatteryBoard: Osu! Keyboard

Today I made a quick project: a Osu! keyboard for a friend. I used an Arduino Pro Micro and Red Cherry MX switches.

Volume Control

After a quick test with the new IEMs, I decided to glue the shells together permanently with epoxy. The final result is rather nice looking:

Ooh.

Triple BA Driver

After testing out the single driver IEMs, I tried out the triple driver unit with crossover, the GK-31732, as recommended by the DIY IEM subreddit and more specifically this post. When the author said the drivers would be a tight fit, he wasn't kidding. After trying my best, the SE535 shells barely "close." I couldn't get the back end to close up completely (the back lip of the shells wouldn't overlap, <1 mm), so for now I just taped the shells together.

Soldered the driver directly to the input.

Taped together.

Compared to the single drivers, this unit has far more bass (as I expected from the relatively massive woofer unit). At first I thought it was too bassy, but after realizing that my HTC One M8's "Boom Sound" feature added a peak to the bass and treble and disabling it, they sounded much better.

Also, it turns out that increasing output impedance of a source is bad, especially for balanced armature drivers (which I did by adding the 18 ohm resistors to the singe driver units). I'm going to need to tackle this volume/noise issue another way. There are adapters that add impedance to headphones, but I'm not sure if they are doing anything special other than adding resistance in series. I've seen some people using ifi's iEMatch, which is supposed to help reduce output impedance and reduce gain, which decreases volume and noise.

I'm just going to need better electrical isolation and an audio output with a lower noise floor and lower output impedance. I'll probably remove the 18 ohm resistors too (but this means they can only be used with devices that have lower noise floors).

Building IEMs

My dad came back from China yesterday with parts to make an IEM. He brought back putty, SE535 shells, ED-29689 (well the China OEM edition) balanced drivers, MMCX connectors, litz wire, one pair of filters, and a headphone cable.

Earbud Testing

A few weeks ago I was interested in getting a pair of IEMs, or in ear monitors, for listening to music on the go. I have a set of over ear headphones and a basic two speaker setup on my computer, but neither are really portable (over ear headphones are a bit bulky for mobile use). I do have the VE Monks, a pair of earbuds that sound quite good for their cheap price, but they lack any isolation.

Water Droplet Photography Continued

After yesterday's successful attempts at taking photos of droplets, I improved the set up slightly and took some more photos.

Wood scaffolding.

Water Droplet Photography

Like astrophotography, high speed imaging is another intersection of technology and photography. I wanted to try out some basic high speed photography, and last night I took photos of water drops falling into a bowl of water.