23 April 2016

Trying to fix a Copal 602 flip palette

While everything seems to (still) work, inside the Copal 602 clock I found a loose palette. It has one hook missing, and it should be possible to rebuild it with a tiny cut of hard plastic of the proper thickness. I still have to figure out how to keep it in place while the superglue does it job.

But, what if it doesn't work? (Or I get it totally wrong.) I think modern technology can come to the rescue: laser cut. Probably the hardest part to find is the right plastic (let alone the color), then with a 1200 dpi scan of an original palette and a digital ruler on the picture, a digital model can be drawn. Finally it is a matter of few seconds under the laser.

Here is the 1200 dpi scan if anyone needs it.

20 April 2016

Copal 602 flip clock

3/4 side view.
Before LEDs, how were digital clocks made? Not with Nixies, that (in my opinion) were mainly used in instruments where the information to be displayed was more randomic than sequential. There existed flip clocks! It looks like they were invented in Italy.

But at a flea market I found a Japanese production, the Copal 602. It was being sold as an ornament at the cost of few espresso coffees (or one basic fast-food menu, burger + chips + drink): well worth a try, and in any case a curious ornament or case for other projects.

With my surprise the clock does work even though it appears to be running a bit fast, like 40-50 seconds gain per hour. It is very silent, much more silent than a classic clock, and it even has a neon bulb that gently lights up the time (and it works too!).

Copal 602 front view.
Time to have a look inside, since I have no real idea how it works.

14 April 2016

Lucky battery pack

Inside A1060
I've had a battery pack from a MAC PowerBook G4 sitting around for a while. The A1060 pack is rated at 10.8V and probably 4400 mAh. It has charge/discharge protection circuitry and an embedded 5-LED charge status on-demand display. When I took apart the PowerBook I kept the battery connector (board), and that was a smart move.

Needing some 18650 cells, I decided to rescue them from the A1060. I was lucky because the pack was apparently made with servicing in mind, and since the glue has dried over time I could pry it open very easily. This fact immediately changed my plan: check the pack before taking out individual cells!

Now it was much easier to determine the polarity of the battery connector, and therefore of the matching board. In less than ten minutes I was recharging the battery pack at circa 11V 2A input.

Fast forward few hours, the pack stopped charging and the onboard LEDs report 5/5 charge status. I am now monitoring the self-discharge while I plan an endurance test.

Full pack and the matching board inside G4 12"

04 April 2016

Nice-looking waterproof matrix keypad: how?

Say that, for some aesthetic reason, you need to embed a keypad inside the case of a project. Or because you want to build something splash/dust/dirt-resistant. But you need some human interaction, in the form of buttons to press.

Such an application could be a kitchen timer.

How to protect keys against splash/dust/dirt? Keep them inside the box, but then mechanical actions will not work. I came up with these possible alternatives to a membrane matrix keyboard:
  • capacitive sensors
  • hall effect (need a "magnetic" pen/tool)
  • photoresistors (only if the case is transparent)
Either I used wrong keywords, or nobody has documented online a matrix keypad not made out of mechanical switches.

Something must be wrong in my ideas. So far I came up with these cons:
  • capacitive sensors are unreliable, sensitive to static/noise, ...
  • hall is power hungry in large deployments, at 5 mA standby for each sensor (but a single row/column could be scanned)
  • LDRs need transparent case, a ligthed environment and are somewhat slow to react, in the order of 10 milliseconds
Any further thoughts?

I am about to buy a bunch of LDRs and Hall sensors to test it out...

02 April 2016

Arduino 1.6.8 IDE problems on Windows XP

In the lab I still use a 2004 laptop that runs Windows XP. It is compact and does everything I need.

Unfortunately the last update to Arduino IDE 1.6.8 caused a couple of issues:
  1. "ld returned 5 exit status" when compiling, making it impossible to get working bytecode
  2. 100% CPU usage by services.exe and java.exe (or javaw.exe) even when the IDE is idle
The first issue can be solved by replacing the stock ld.exe with the one from old IDEs, like 1.0.5.

But I found no solution (and nobody complaining either) to the second issue, so I am reverting back to 1.6.5 that, as far as I remember, had no problems. Otherwise I will go backwards with releases until I reach a stable point again.

Edit: yes, 1.6.5r5 does work indeed. It is a Windows XP SP3 with a Intel centrino CPU.

16 March 2016

biNixie Clock schematic diagram

For those interested in reproducing my biNixie Clock, here is the (most likely) schematic diagram:

I have realized I did not take notes while building it, probably because it is so straightforward (and a copy of other designs).

You need to pay attention to the current limiting resistor (one per Nixie tube), which depends on the tube model and your anode voltage ("HV" in the diagram).

Put everything on sockets so it can be easily replaced or reused.

At current prices (3/2016) the whole thing should cost less than 30 Euro if you use common Nixies like B5092/ZM1020.

The firmware for Arduino and compatible boards (ATmega368) is here https://github.com/jh1995/biNixieClock.

Please share pictures of your biNixie clock as well as firmware improvements!

15 March 2016

Four digit display with HP 5082-7300

Arduino Nano counting...
My interest for old display technologies got a bit of satisfaction when I lighted up a bunch of HP 5082-7300 chips. These are 8 pin displays with an integrated BCD decoder (0-9, dash, blank) plus RHS decimal point that emulate a dot matrix configuration. The required power supply is 5 V and today they can be easily driven with the microcontroller of your choice. The integrated logic includes a latch, so you can connect more displays in parallel and multiplex data through the [Latch Enable] control.

One remark not clear from the datasheet: the decimal point is latched as well as ABCD inputs.

My first application is a simple 0000 to 9999 endless counter. It is a good way to test all dots work, measure current consumption and do other optimizations.

As expected, these little displays produce a good amount of heat. According to the datasheet, at 5 V the current drawn is well above 100 mA, so that's more than 0.5 W of heat to dissipate. To minimize the risk of failure due to heat I mounted my 5082-7300 on a "socket", so I can add a small fan on the side if I want to.

Then I tried to reduce the voltage supply, both to the Arduino and to the display string. The good news is that there is no change in dot luminosity down to 4.1 V and the average current goes down to 60-70 mA per display: that's about 0.25-0.3 W dissipation per display. Cool, isn't it?

Click to zoom and see the logic through the red plastic. Last digit was changing from 6 to 7.
Voltage supply can be further decreased but then pixel luminosity is not uniform anymore.

So, in the final application I can run the whole setup Arduino+HP 5082-7300 at 4.2 V or so. Given the amount of power needed by the display rack itself it makes no sense to optimize the firmware for a lower consumption or use a stand-alone microcontroller in place of an Arduino.

Now it is time to fill the perfboard with a Nano, an RTC and some form of alarm function. And write some code, of course.

I have some HP 5082-7300 spares in case you land on this post looking for this specific part.