21 November 2018

Sending fragments of data to a microcontroller

Alright. I have (autogenerated) a new problem to solve: I want to be able to send few bytes of information to a microcontroller installed into a box without using a physical input device (a.k.a. "button"). This operation has to happen more than rarely, so I would like to avoid throwing in some hardware that potentially will never be used or would lock me in to an external technology. At one-off production run it is purely a stylistic affair, not a cost-cutting measure.

Specifically, the device is a digital clock and I want to be able to set the time without extra drilling the box and installing a button. So "wirelessly". There are two holes, one for DC input socket and one for the Nixie tube protruding out of the cover. The display acts as a feedback to the user, of course.

My night-long brainstorming came up with these options:
  • Plain old button(s) requires an extra hole.
  • WiFi or Bluetooth are too power hungry, too costly, too technically advanced because they require a software counterpart.
  • A reed switch, but the box is quite thick and it requires a powerful magnet that might not be available when needed.
  • An IR receiver under the Nixie and a TV remote as remote controller that might not be available when needed.
  • A microphone to receive AFSK or similar, but requires a software counterpart.
  • An open capacitor to take advantage of hand proximity effect.
  • A LDR under the Nixie tube and operator's hand to command through shade/light.
  • A tilt sensor, not the cheapest way.

The following table summarises my options:

Type Hole? Extra tech?
Button Y N
WiFi / BT N Y
Reed switch N Y
IR receiver N Y
Microphone N Y
Open capacitor N N
Tilt sensor N N

I like the last two the most. Both LDR and tilt sensor are hand-operated and allow for an easy repurpose too: at powerup they serve as input devices to set the time, during operation they can provide information to control brightness.

I need to verify whether the LDR under the Nixie receives enough light to understand "light vs shade" situations. Otherwise I will order a tilt sensor.

Readers, please leave a comment if you have further interfacing ideas!

11 November 2018

Reviving a ceiling lamp from CFL to LED

We've had at home a ceiling lamp for about 10 years, and for the last 4 years it has misbehaved. It contains two 55W CFL tubes with an external HV power supply by OSRAM (Quicktronic Professional QTP-DL 2x55 GII).

The defect was that it would switch off after about 15 minutes of operation and it would restart only after a rather long while. In any case it was not usable at all.

The original configuration.
I tried replacing the gas-filled elements without success, so the culprit must have been the embedded power supply. Since I am still developing my spider superpowers, working upside down on the ceiling is not my thing so I pulled down the lamp on the table. Still about 15 minutes of operation even without the front glass (in case it was a heat issue).

I took apart the power supply case and looked for the obvious, that obviously wasn't that obvious. I simply reconnected everything with the power supply out of its case and it would run for hours. HA! Maybe a cold solder joint failed with heat cycles over time and moving the circuit around sort-of fixed it.

I took my chances and put everything into the metal case, and still working. So I secured the case back to the lamp and ... 5 seconds of operation!

Enough. Either I got another identical power supply or I had to find something else. My readers probably already guessed which one I picked! "Something else", of course.

The lamp base is metallic and hidden against the ceiling so it can be drilled as needed to accomodate a different setup.

Few searches later I learned that there are LED replacements for neon tubes, and they include the power supply too. I chose two circular replacements whose total equivalent power would get close to the original 110W.

Of course I had a look inside the power supply, that is very simple since it has to light up a series of LEDs:

The LED strip power supply.

And the final look, up on the ceiling before placing the cover glass:

Final config. Notice something "light"?
Now the two light sources are independent, so if one fails the lamp will still work at 50% luminosity. If you look closely at the last picture you will see LEDs are "on", dim, but "on". Even with a single wire attached, the culprit is probably some leak of the long driving wire towards ground. I couldn't get rid of it by swapping the two cables, so I consider it as a night light undocumented feature.

We don't use that lamp very often, so hopefully it will last another 10 years!

06 November 2018

Elka 135, calculator from Bulgaria

Meet the Elka 135 scientific calculator! I bought this marvel in FN 2018 fair. Batteries had leaked inside, so this was the fault to fix (yet again).

There are few pages online that describe this calculator made in Bulgaria. I like the 8-digit LED display. It can be powered from an external 6V DC source through a 3,5mm audio jack (!!).

Using a variable voltage power supply I could confirm that it works down to 4.5V and a small "L" appears in the leftmost position: this means I can use this calculator with USB power, without fully restoring the battery contacts.

Few keys are bouncy resulting in double-triple press, but I am not sure whether they were like that in the original, non-damaged calculator.

More complex operations take some time, and the left position in the display keeps updating until the result is ready.

This device is quite power hungry, because of the LED display, but modern batteries (or USB power) can handle it.