A doble-sided keyboard mishap with springs

OH dear.

Pictured above is my IBM Model N2 keyboard, mere moments after getting disassembled in a rather unprofessional manner. It had been acting up, so I figured that I’d take it apart and try to fix it. I should have looked for some instructions first.

I think you should be able to spot what I did to remedy the problem.

The small SMD electrolytic caps had gone bad. The 2.2µF unit got replaced by a 4.7µF one without any obvious problems arising from it. I added some “China goo” to keep them in place afterwards.

Typing off-key (geddit? GEDDIT?)

T-t-t-time laaaaaaapse!

All springs in place!

There's something satisfying about clicking a hundred keys back in place.

Of course, ONE key didn't work as it should.

I’m leaving that for the future, though. When the spring is place, it triggers all the time. It sort of works without the spring, so I’ll fix it when I’ve got the patience.


A double-sided keyboard mishap with coffee


We've all been there.

The curse of using nice keyboards is that when accidents like these come around, you can’t just toss’em away and buy new ones. No, sir. You need to painstakingly take them apart and manually scrub every last drop of dried-up coffee out of there, and just pray to a higher power that the smell will go away.

That doesn't look too bad, does it?


Looks are deceiving.

Coffee everywhere.

 I didn’t bother taking more pictures after that. Lots of hot water cleaned everything right up. Except it still smells like coffee.

Fixed until next time.

FF’s personal headlamp

This may not be what most people think about when they hear the phrase “personal headlamp”. I don’t blame them, as I don’t consider myself to be “most people” very often at all. In particular when it comes to mad, rushed electronics projects involving lead-acid batteries and duct tape.

The back story of this lovely piece of kit is that I, years ago (before I got myself a pocket-sized Dealextreme LED torch) too often found myself walking around in the dark Finnish winter with but a cell phone to light my way. A late night with plenty of coffee, tape, creativity and small lead-acid batteries at hand spawned this:

The first revision. Mid-2008, I think. EXIF data is since lost, but I'm quite certain that the picture is from 2009.

It consisted of a 21W bulb inside of the rotating armature (from one of those silly LED lights with a crank), as well as the three-position switch and 7 Ah 12 V battery that are still there in its current incarnation. It had one level of brightness and it served me well for a good while. However, once the bulb went bust last winter due to an incident involving a big dog, air travel and snow, I wanted more for this year. I wanted an H4 bulb.

H4s are lovely due to their flexibility: They have two filaments, which can easily be hooked up to run in series, as the following “pseudo-schematic” demonstrates:

Schematic over how my H4 bulb is hooked up

In a car, pin 3 is almost always ground. When you switch between low and high beams, you turn on one filament at a time. The main difference between the filaments is that the low beam filament has a bit of metal beside it, which prevents light from getting reflected straight ahead and into oncoming traffic.

In my application, however, I abuse the fact that the filaments share a common ground for the purpose of running them both at once and in series. By applying a voltage across terminals 1 and 2, I’m effectively halving the power dissipation of the bulb with no external circuitry. Since pin 1 is always ground, I can then run the high beam filament by simply switching my voltage source from pin 1 to pin 3.

It isn’t a particularly good solution, as the bulb is vastly less efficient at lower power; I’d get more light out of the 25-ish Watt it consumes by using a circuit to PWM one filament at 50% duty cycle. However, if I was to care about efficiency, I wouldn’t be using an incandescent bulb to begin with.

Here’s how the hook-up looks in practise:

Black is GND, red is VCC, yellow is high power, orange is low power.

With that out of the way, let’s get onto the actual building of the headlamp in its most recent incarnation! Some painstaking abuse of my side-cutters resulted in this:

It looks like the light is supposed to go behind it, doesn't it?

The holder for the bulb is made out of an old computer fan grille.

Some abuse of my fingers then resulted in this:

I don't know what I'd do without my red copper wire.

Low power mode, where both filaments are activated.

High power mode, where only the lower filament is activated.

A "normal" exposure of the light being on, low.

A "normal" exposure of the light being on, high.

The bulb I used for this has an extremely cool colour temperature. The white balance I use for these photos is set for my 600W halogen flood light, which I consider to already be quite cool. This little 60W bulb still trumps it in that regard. I was actually fairly surprised at how white the bulb is, as the battery voltage drops to well below 12 V when using one filament.

So, that’s the story on my headlamp. It’s incredibly inefficient, not particularly durable, very heavy and ugly beyond belief, but it’s saved my butt more times than I can count. While my little LED torch is fine for most things, you can’t just put it down on a table and expect its little focused optics to light up a room properly. It also makes cars turn their high beams off for me at night, and I can even return the favour!

It looks a lot better in real life, as the filaments fade down over about a second after it's turned off.

Super Flower Golden Green Pro – Part 2

A sad sight indeed.

Pictured above is how my poor computer has been running since a couple of months back, when my old Corsair HX520 bit the dust. The ARENA PSU is at least a Chieftec on the inside, at least.

MOLEX connectors are the best.

My graphics card got its power through a fan adapter AND a PCI-E adapter!

Sparing you a build montage, poof!

One could argue the sensibility "two PCI-E connectors on one cable" design.

The cable sleeving is quite nice for routing on this PSU. They’ve used some form of flat sleeve, so the cables are quite easy to route around thin areas within the case.

A small detail, but I find it noteworthy.

Of course, while disassembling my computer, I can’t just replace a part and leave it at that, no Siree! Some improvements to the case:

I never understood the point of internal fan grilles.

I think that looks beautiful, don't you?

The reason for the duct tape is that whoever designed the fan grille for the plastic front of this case didn’t do a very good job. A fourth of the fan is simply covered up by plastic, whilst a fourth of the filter has no fan behind it. The duct tape is installed in a way which seals the whole unused HDD bay off from the rest of the computer, with the exception of the holes that the fan covers. This allows for a considerably larger area of the fan filter to have air flowing through it at a more even spread. This should both allow less dust into the computer, and allow for more throughput at the same time.

Oh, and the fan being placed closer to the graphics card helps remedy this.

There is a small . . . flaw, with the Arctic Cooling Accelero GTX Pro (aside from its name), and that is that that they didn’t quite consider the amount of heat that was generated by the VRM of the GTX260 c216. Without some extra heatsinks stuck on top of the PCB and a fan pointing directly at the card, they will simply overheat. 145°C seems to be as high as they’ll run without causing issues. By moving the fan and creating the “pressure chamber” out of my HDD bay, the FURMARK temperature of phase 3 dropped to below 120°C!

Being made in Switzerland does sound nice

Two out of three, at least.

I can’t say that I’m particularly satisfied with the cooler overall. The fans are rather noisy, and at very low speed, the above happens.

The power supply considerably reduced the power consumption of the computer.

Testing everything out, the Super Flower does deliver! This computer would consume well over 400W when loaded down similarly and running on my ancient HX520. It is quiet in both electrical and mechanical noise, the fan controller follows a nice speed-to-temperature curve, and it even has that RELAY that’ll go click when you turn it on or off. Lovely!

In conclusion, I have a strong feeling of this being a worthwhile purchase, and as of now, I can’t do anything but recommend it. It’s too early to tell if it’s going to last or not, but the signs are very good indeed.

For those who care, I’ll close with the specifications of the computer:

  • CPU: Intel Core 2 Duo E7400 (3.8GHz/1.35V)
  • CPU cooler: Xigmatek HDT-1283 (Noctua NF-S12B rear case, NF-P12 front case)
  • Motherboard: ASUS P5Q Pro
  • GPU: Nvidia Geforce GTX260 c216, 896MB (Slightly over-clocked)
  • GPU: Nvidia Geforce 8400GS
  • RAM: 4GB DDR2
  • HDD: 320GB Samsung Spinpoint F1
  • Sound card: ASUS Xonar DX

A use for sleeve bearing Yate Loon fans

My basement workshop was not designed with heating in mind, the only heaters present being placed ten centimetres from the roof, with the goal of providing heating for the floor above. I’ve previously used a pair of fifteen-year-old, ball bearing Nidec Beta V fans to circulate the air around the room, but no matter what power they’re running at, their old bearings vibrate enough to cause every last radiator in the house to convey their music.

Three broken cheap-o Yate Loons and a 10V cable wart.

The solution I came up with was to use the ever growing collection of worn down, sleeve bearing Yate Loons that I’ve got. They all come from old computer power supplies, where they’ve reached their EOL with that classic “RRR” noise of sleeve bearings used in horizontal fans.However, they will work reasonably well if they’re turned upright (as in the picture above) or flipped around (with the motor facing down), which is how I’ll use them.

Boxers or briefs?

The big, white fans are the old Nidec Beta V units that I used to use for the same purpose.

Slingbox Pro SB200-XXX power supply failure

The 230V connector on this unit was attached to those jumpers at the edge of the board. They require desoldering before the board can be removed.

Ktec AC adaptor, manufactured in 2007 (assumed), model no. KSA0600350W1UV-1

C12, C13 and C14 had failed. C12 and C13 are connected directly in parallel, and were CapXon GF at 16V and 680µF each. C14 was also a CapXon GF, but at 16V/220.

C12 and C13 were successfully collectively replaced by one recycled 16V/1500µF Chemi-Con KZG. C14 was successfully replaced by one 50V/100µF Nichicon PS.

The solder joints for the three-legged component beside C12 were not failed, but very poor. I recommend re-doing them at the same time.