Need to Blog Again

I need to start blogging again.

But more than that, I’ve been thinking lately about the use of the internet to communicate. I haven’t really been reading my friend’s blogs. Dissapointed to learn that one of my friends blogs at had gone offline. Hopefully it isn’t entirely defunct. People running their own blogs seems to make for a healthier internet than people using one of some small number of monopoly services as a client. It seems, when people want a forum to contact their friends, they all pile onto facebook or twitter.

The advantages: Everyone’s on there.’As I’ve learned in the past, it’s hard to have an active internet social life on an empty PhPBB forum. If no one is on an IRC channel but bots, then you can’t very easily have a conversation.

The disadvantages: Facebook isn’t a platform that you own or control. You’re putting your personal data and a record of your personal sentiments, as well as detailed information on who you’re friend’s with, how often you communicate, etc in the hands of this company. A company which, these days, seems hell bent on manipulating their users.

The belligerence of the tech monopolies has spawned various movements to produce alternatives. I’m not familiar with most of them, but may investigate later.

I’m going to start looking into doing more with my servers. It seems like in the early internet everyone had their own space/forum/what-have-you going. Popular webcomics all had their own forums. Every hobby on earth had it’s own forum (I remember one that I liked a lot – I think it was run by SEDs or the Mars society.) People owning and operating their own servers and forums seems like a desirable end goal (to me at least, to “people”, who knows?) Maybe if I figure out the ins and outs of some of this I can help by posting tutorials. “How to Internet: 101”.

I’ve tried before to create a dovecot/postfix e-mail server. Past attempts produced ill-configured junk and corrupted package lists. Maybe I’ll try again this weekend. This is apparently one of the more complicated and fiddly parts of setting up server services.

In Ye Olden Days there were mailing lists: Why don’t people use wide-distribution e-mail the way they use a centralized forum like Facebook? There wouldn’t be a forum to censor or shut down, just everyone’s e-mail server forwarding the messages. There would have to be a way to separate “serious/business e-mail” from “friends-and-family spam”, from “everyone else’s spam”. People probably don’t want their inboxes filling up with junk. Maybe a better set of sorting filters could take care of those sorts of problems?

Why do people use Facebook instead of PhPBB style forums?

Why do people use centralized chat services like discord, instead of everyone running their own IRC server?

Why do people use YouTube instead of hosting their own videos on their own servers?

(Monetization – YouTube makes it easy to collect ad revenue. That is, if they like you. If they don’t they’ll abuse you and play head-games with your listeners.)

(I expect there to be reasons in each case: Maybe pointing the way to the development (or re-development, as the case may be) of a less centralized internet.)

This weekend in electronics

All the stuff finally came in to try making my ADC board. Here it is:

The board

The board

The mess

The mess


This board is intended to be a general purpose lab DAQ board. Depending on how you set the resistors in the jumper slots, you can sense current, voltage (up to +/- 10V), and set the gain into the ADC and analog output. You can even route the analog output to something else. The way it is designed, it is +/-600V common mode tolerant and has a 130kHz bandwidth. (The main bottleneck being the high voltage differential chip)


Other features include:

  • The world’s worst SMD solder job.
  • Mysterious voltage divider behavior.
  • A broken signal generator, so I had to test it with batteries.

:-P. I can fix that. It seems to work though, which is pretty good for my second PCB project.


Design Files:

Kicad Files


Sequoia National Forest

If you are ever in California, there are a few things you really need to see. When my parents visited in April, we went up to Sequoia national forest. It really is a wonder of the world!

(I only recently got around to posting these, because disorganization:)

20180421_155941 20180421_155942 20180421_160146 20180421_160730 20180421_160748 20180421_160749 20180421_160751

That breakout board, assembled

I managed to assemble one of the breakout boards that came in the mail from OSH-park today. (I screwed up the other two … but 1/3 appears to be assembled and working correctly.)

Here you can see the board attached to an STLink_v2 programmer, with pin PB0 connected to an LED.


Thai Cave Rescue

I hear now that the Thai navy seals have rescued 8 of the children trapped in the cave. Here’s hoping that the remaining children and their coach can be rescued. Good luck.


Update on 11 July: They’re all out and safe! Hooray.

CNN’s story on the matter.

My First Circuitboard with KiCAD

I made a simple breakout board for the STM32F732 microcontroller. I wanted the pins to be arranged in something like an organized fashion. The pin groups are arranged by name and function (PA0-PA15, PB0-PB15, PC0-PC15, and various control pins.) The STLink programmer should be able to plug right into the top-right four pins to program the controller. The controller should be easily powered by supplying 3.3V and ground to the VCC and GND pins on the top right.

This resulted in a bit of complexity in the routing (which you have to do by hand in KiCAD – no autorouter). I figured out a technique to make it all fit on the board. OSH Park came back with a price of $38 for a minimum run of 3 of these circuitboards. In two or three weeks, I’ll see how well this breakout board works.

circuitboard STM_simplebreakout STM_simplebreakout2

This is part of a proof-of-work that I understand PCB technology, microcontrollers, and PCB CAD software. I’m not an electrical engineer by education, but I’ve picked up quite a bit of it in my work. The KiCAD project files can be downloaded from my site:

PS – yes, I know the ground plane is broken up, and held together by traces. This is my first attempt – I may be able to clean it up a bit in post.

Solder wick

How much solder would a solder wick wick if a solder wick could wick solder?

I still have no idea how to get these things to work …. 😛

LA County Airshow, 24 Mar 2018

Weekend Project: Raspberry Pi Cluster

This device will not self destruct in 10 seconds (fingers crossed). So what is it?

It’s a super-computer! Well, okay, not entirely. It *could* be a cluster supercomputer, if these raspberry pi’s came with more ram. Each board actually has enough processing power to make a decent cluster. However, each board only has 1GB worth of RAM, and there’s no easy way to upgrade this – the architecture of these single board computers isn’t very extendable in that respect.

If you could get a raspberry pi with expandable memory slots, or if each came with, say 8GB of RAM, you could chain them to form a very nice supercomputer cluster.

So why do I want it? (Because I’m a neeeerd.) It is set up like a supercomputer. Right now I have the one head node (hedgehog), and two computing nodes (hedgehog_n1 and hedgehog_n2). I intend to play around a bit with the high performance computing software libraries (MPI, petsc, and so on) at use at work on a “supercomputer” that I own and control. I also want to further self-educate on how to set up, administer, and write software for supercomputers.

Further evidence of my nerd credentials: Stare into the abyss that is my desktop. Yes – that is an image of several VNC sessions with the nodes of my device, running inside a linux virtual machine, running inside my main windows PC.

Computers all the way down!

Computers all the way down!

The total cost of this project, btw, is ~$150. If I want to start throwing together a real supercomputer, I imagine the software and architecture will be the same, I’d just need to upgrade the hardware. (There are nice higher power small-form-factor fanless computers available from IBM, but they’re more expensive.) I figure I’ll shake the software products out first, then scale up.

Electronics Lab Pics

Finally found where I had stashed these photos:
October 2017: I’m reassembling my electronics lab in the new house I’m renting.
Here is that 3d printed dowel-truss shelf that I made. I wrote some python scripts that create STL files for dowel-truss brackets. You specify the nodes, it gives you what to print, then you cut the dowell rods and assemble.

Electronics Lab

Electronics Lab

Here is the older, messier version: 2016