attiny85 powered kWh-meter-meter and 1-wire host

At Revelation Space we wanted some insight into our powerusage. After a few prototypes on arduino’s I ended up building a small board powered by an attiny85.

In later revisions I added a Dallas/Maxim 1-Wire bus that will read out DS18B20 temperature sensors, so I could monitor the temperature of the server that was logging the data (it didn’t have any supported sensors).

The schema shows the attiny85, the programming header (non-standard pinout ISP header) and used parts.

The LED will blink for every received pulse, so you can verify this with the original meter. I made the connection to the LDR as a 2-pin jumper-block, so I could easily disconnect the LDR, which is connected to a long cable so it can reach the led on the kWh-meter.

JP1 on the schema is the 1-Wire bus. I ended up using just the 2-pin version. On my board I terminated this on a 8×2 pin female header.

The final board:

The kWh-meter I am monitoring blinks a led a number of times per kW used. In this case it blinks 1000 times per kW, but I have seen other meters that blink 400, 480, 600 and 1000 times. This is usually documented somewhere on the meter itself.

It looks somewhat like this one:

The trick is to attach an LDR in front of the LED on the meter, and then read out the data using an analog port on the attiny. Make sure to blank out any outside light that can hit the LDR using black tape.

The code for the attiny can be found on my Github

The last components are rrdtool to make graphs and a perl script to log data from the (serial) attiny and pipe the data to rrdtool. Rrdtool and perl code can also be found on github.

After a few days of logging you should have a good insight into your power usage:

If your kWh-Meter uses a rotating disc, Juerd has a nice design that works with those meters. It’s based on an arduino and uses 7-segment displays to give a live insight into the used wattage.

USB-ASP Avr programmer from e-Bay, issues with attiny85

I recently bought a USBASP programmer from e-Bay, the one pictured here

It seemed to work just fine when I was programming attiny4313’s or atmega’s, but would fail on verification when programming attiny85’s (and probably other small/slow attiny’s).

Setting the fuses would work, but writing flash would fail consistently.

The trick to resolve this was to update the firmware to the 2011 version of USBASP, which can be found at

To program the USBASP itself, you need a second programmer (I used a usbtiny) and you need to jumper J2 (which are the 2 solderpoints next to the leds).

After upgrading the firmware, use avrdude with the ‘-B 10‘ option to write to slow devices. This option is not needed with faster (1.5Mhz and up) devices.

Fake PS3Break

This week I received a PS3Break device, but it seems it’s a fake version. It works just fine with the 3.41 firmware on the PS3, but doesn’t have an upgrade button to upgrade the payload on the device.

This is not (yet) a big problem, but it would be nice if I could get it to upgrade to a newer payload. It’s label claims it’s version 1.1. Here are some pictures of the device itself, and the PCB inside.

outside casing of fake ps3break
outside casing of fake ps3break

Front side of the PCB
Front side of the PCB

Back of the PCB

I have already tried shorting several sets of pins on the PCB where I guess the programming button might be wired, but so far I have not yet succeeded.

Might be a nice little project to look into on the next hackerspace meeting.

…becomes a mame-cabinet

The scrapmame cabinet is getting to a functional level. Today I was able to play a game of asteroids, pacman and galaga on it.
All the buttons have been wired up to the controller using screw terminals (kroonsteentje) for easy (dis)assembly.

Galaga running on scrapmame

Before getting there I had to do some creative wiring, as it seems I cut some corners with wiring up the controller. A few separate ground-points were connected, however this was causing interesting signals on the analog joystick axes. A lot of trial-and-error later it seems I have clean inputs from all buttons and the digital joystick.

I still need to hook up the second stick (which will hopefully arrive from DX this week), some buttons (insert coin, start game) and a usb keypad (for misc buttons). A set of speakers should also arrive this week, allowing the scrapmame to make a decent amount of noise.

Next week I’ll probably fit in the PC (I’ve been using my laptop for testing so far) and then I’ll start on the launcher/menu interface.

I’ll try to add some more pictures of the inside of the cabinet later.

Some woodworking…

At Revelation Space I have been doing some woodworking for a project I’m working on. I’m normaly the guy with two left hands when it comes to working with heavy machinery and old-skool hardware, but I must say I’m not doing too bad.
It helps that revspace has more then enough of the necessary tools (at home I only have a hammer and a screwdriver :P)

The project is a self-built Arcade cabinet, which at this time has been called ‘ScrapMAME’, as it’s been built completely out of scrap-materials.
Step 1... getting lots of wood

I PSX Controller (pre disassembly)have built the frame and outsides of the cabinet from wood boards that were originally bolted onto the windows in the workshop, together with some wood beams that were still lying around the space.
The buttons and joystick have been ordered from DealExtreme, and the first batch has arrived. Hopefully my second shipment with the remaining parts will arrive soon.
I’m using this arcade stick, and these buttons from DX. The buttons will be wired to the pcb of an old PSX controller, which will be connected to a PSX->USB adapter.

2010/06/26: The buttons and joystick arrived, so I could finally drill the holes.
Unpacked the newly arrived buttons and arcade stick
Holes for the fire-buttons and joysticks

There is still lots to be done… which will be added as soon as I get to it ;P

Replacing/updating my PGP key

I’ve decided that my PGP key could use some updating after 9 years. I have created a new (longer) and better-hashed PGP key. Now using 4096 bits of RSA, and using sha2 from now on 😉

And the new key is:

pub 4096R/C4F69BD2 2009-05-11 [expires: 2019-05-09]
Key fingerprint = 6AFC 0D83 5346 C729 6754 113F D867 4D8F C4F6 9BD2

I have placed the key on my server: and uploaded it to the pgp keyserver(s).

I have also created a transition document, which is signed by my old and new keys, located on If you have signed my old key (0x357D2178) please consider re-signing my new key after verifying the signatures on this document.

If you prefer to meet me face-to-face for a key verification, that shouldn’t be too hard to arrange. I’ll try to keep some keyslips on my at all times.

Update 2009/05/18: The file was signed first with my old key, and the result was signed with my new key. This gave some problems, and the file is now replaced by a new version which is signed by both keys simultaniously. This should verify correctly.

curl -s | gpg –verify
gpg: Signature made Mon 18 May 2009 11:00:17 AM CEST using RSA key ID C4F69BD2
gpg: Good signature from “Mark M. Janssen <>”
gpg: aka “[jpeg image of size 1522]”
gpg: Signature made Mon 18 May 2009 11:00:17 AM CEST using DSA key ID 357D2178
gpg: Good signature from “Mark Janssen (Maniac) <>”
gpg: aka “Mark Janssen <>”
gpg: aka “Mark Janssen <>”
gpg: aka “[jpeg image of size 1522]”

I have also updated the old key to expire on 2009/12/31

Solaris 10 as LDAP client (to OpenLDAP 2.4.x)

During the last couple of weeks I’ve been working on getting a central directory setup for my client, running on OpenLDAP 2.4. Not having worked with LDAP a lot before it proved quite a challenge, especially getting Solaris 10 to work with the LDAP server without any glitches.
In this document I’ll try and describe how this setup was made, because I have been unable to find a single consistent document describing all the intricate details.
At this time I have all my problems fixed (AFAIK), but during the setup phase I experienced various problems:

  • Solaris 10 not seeing any users from LDAP
  • Solaris seeing users, but not letting them log in
  • Log-in working from console, but not ssh
  • Passwordless login (pubkey) not working in SUN-SSH
  • Users being able to hack extra permissions for themselves
  • etc…. etc….etc…

The entire article has been moved to a more permanent location, as a page on this site. You can find it under the ‘Pages’ header on the right. Setting up ldap

HAR2009 Publishes its CfP

HAR2009, the four-yearly Dutch outdoor technology-conference has published the Call for Papers. I’ll include it verbatim here:

From the ancient days long before the first wayback-machine snapshot, hackers have a track record for appropriating technology that was meant for something completely different and putting it to alternative uses. And every four years since 1989, the international hacker community has descended upon The Netherlands in great numbers for a conference that focuses on contemporary and future issues surrounding technology and its social and political consequences. One reason that these conferences have been successful is the wide range of participants: from students, amateurs and aficionados to researchers, scientists and entrepreneurs who are recognized as some of the best in their respective fields.
Continue reading HAR2009 Publishes its CfP

Optimizing memory usage on Virtual Private Servers

A repost from
The smaller VPS’s might be a bit limited in memory if you want to run more advanced web-setups. In this post I’ll try to give some pointers on reducing memory usage, so you can have a more feature-full environment without running out of memory:
Looking at the memory usage in a smaller VPS a few things are obvious:

  • sshd
  • mysql-server
  • apache

These processes are essential to the functioning of a VPS, but they can be tuned a lot. I’ve tuned a 80MB VPS from 4MB free back to 30MB free by making a few small modifications:

  1. Replace openssh-server and clients with dropbear, dropbear only uses 900K vs more than 3MB for OpenSSH
  2. Configure mysql for low-memory systems, by using the my.cnf recommended by vpslink
  3. Run apache’s prefork-mpm with the following settings:
    • StartServers 1
    • MinSpareServers 1
    • MaxSpareServers 5
    • ServerLimit 50
    • MaxClients 50
    • MaxRequestsPerChild 5000
  4. Disable any apache-module that you don’t really need with:
    • a2dismod <module>

Using ‘top’, and sorting on the “RES” column you can easily spot the memory hogs. Try looking for smaller or simpler alternatives for large processes.

Also consider if you need various services, or that you could possibly do without.

Update 2008/11/04: Some more pointers for reducing your memory footprint (from the perspective of a debian etch install):

  • Install ‘dash’ and make that the default shell (saves 2MB per shell)
  • Install ‘runit’ to replace init and ‘runit-run’ to replace sysv-rc
  • Install ‘socklog-run’ to replace sysklogd/syslog
  • Remove the getty’s (as you only login over ssh anyway) from /var/service and /etc/sv, then reload runit.

This resulted in (on an otherwise idle and default debian-etch minimal install) in a memory-usage of 2564kb, with the following processes running:

root 1 0.0 0.0 104 20 ? Ss 10:38 0:00 runit
root 2930 0.0 0.7 2736 564 pts/0 Ss 10:48 0:00 dash
root 3429 0.0 0.0 132 32 ? Ss 10:57 0:00 runsvdir -P /var/service log:
root 3431 0.0 0.0 108 28 ? Ss 10:57 0:00 runsv socklog-unix
log 3432 0.0 0.0 160 76 ? S 10:57 0:00 svlogd main/main main/auth main/cron main/daemon main/debug main/ftp main
root 3433 0.0 0.0 108 32 ? Ss 10:57 0:00 runsv socklog-klog
log 3434 0.0 0.0 128 40 ? S 10:57 0:00 svlogd -tt main/main
nobody 3435 0.0 0.4 2528 332 ? S 10:57 0:00 socklog unix /dev/log
root 3436 0.0 0.3 2528 300 ? S 10:57 0:00 socklog ucspi

You still have all the features, and can still use bash for your interactive shells, but you can have the memory resources free when you need them, and also still have cron, syslog and sysv-rc functionality.
You can limit diskspace usage (not mentioned up to here) by using busybox and friends and uninstalling some packages busybox replaces. However, in my experience, diskspace is less an issue than memory.