I now have a prototype in use at the bedside for beta-testing. The components are not all inside the clock itself – there is a box below with the main circuit on a breadboard, which is connected to the clock case with a ribbon cable – this allows me to make adjustments as I go without having to make a PCB.
The clock isn’t working perfectly, but it is waking me up in the morning. A few things I have noticed:
The clock occasionally “crashes” when adjusting the alarm time – some programming glitches remain to be worked out
The display is way too bright, even with the tinted glass. Adjustable or automatic dimming will definitely need to be an option.
The doorbell mechanism for the alarm is way to metallic sounding
The interval for the alarm is set so that the alarm goes off once every 30 seconds. With the doorbell mechanism I am currently using, this is really annoying. This will definitely need to be an adjustable setting.
If you’ve stumbled upon this, please take a moment to fill out this survey about Alarm Clocks – and add your contact info if you are interested in participating in development through future surveys, focus-groups, beta-testing, etc…
What makes it possible for an amateur like me to take on a project like this is the overwhelming amount of support that comes from a worldwide community of developers, often without their knowledge. For example, the circuit design and firmware for the icetube clock are freely available to download and re-use under an open-source licence. The clock design is similar in some ways to my own. The schematics and code don’t have everything I’ll need in the final design, but I have learned a great deal and saved hours, days, maybe weeks worth of time in procurement and experimentation, and this gave me a great tutorial for developing my own design.
I was able to light up the IV-18 tube and have it display digits, without completely understanding how it all works. I guess this is the real beauty of open-source. You don’t have to know everything, you don’t have to do everything. You find something and build on it, and bring your own contribution.
For this prototype I am using an obsolete HV518P high-voltage driver IC, which takes signals from the microcontroller and outputs high-voltage directly to the display segments in the VFD tube. I chose this one mainly because it comes in a DIP package, so I can fit it into a breadboard.
The most difficult part of getting all this to work was wiring from the driver to the VFD tube. The schematics for the icetube were for the popular MAX6921 driver, which I will likely use in the final design, but the wiring for the HV518 seems to be a bit different – so I had to cross-reference the icetube schematic and the data sheet for the HV518 to know which pins went where, which didn’t seem to work so well. It can be a bit like a rubix cube – even if it’s off by just one bit, it can take hundreds of re-tries to get it all in order. In the end I got it working by sheer brute force and determination.
I bought a bunch of these VFDs last year for a bargain. I may use these for the first version of the clock. The have a nice large display with an extra 2 digits, an alarm indicator, and indicators for the day of the week.
Over past year I have had a few meetings with Brent Rouke about the case for the alarm clock. By the summer we had developed a prototype for the case, and in the fall I had some custom glass cut to match the face. Here is the result:
One of the challenges of working with a project like this that uses “exotic displays” such as Vacuum Fluorescent or Nixie Tubes is that these displays are very fickle with the way they are driven. VFDs typically need about 20-80 volts to light up the anode/grid segments of the display, and because they are vacuum tubes, they also need a filament to be heated up with about 1-5 volts.
There are a variety of ways to get the higher voltages. If the tubes you are using are not too picky you might get away with 12-24 volts, but if you want to support a variety of displays, and if you want to be able to drive the displays with multiplexing (i.e. flashing the segments+digits one at a time, in rapid succession), then you will need up yo 80v.
I have very little knowledge of analog circuitry and switch-mode power supplies, so I wanted to change that, and researched various ways to get the voltages I would need, and settled on a switch-mode PSU based on the MAX1771 step-up converter. After some trial and error with procuring the right components, I settled on a design based on this switch-mode power supply. This circuit was designed for nixies, which need up to 250 volts. It’s overkill for my current requirements but I want to support a variety of displays, and may use nixies at some point.
It took me a long time to get this working and I smoked a few mosfets and 1771s along the way, so it was pure joy when I saw the voltmeter jump up to 50v. After playing with some resistor values I soon had it cranking out 250v from a 5v supply. I added a trimpot to make it adjustable from about 25-100v.
I then got out one of the russian IV-18 tubes from my collection and hooked it into a breadboard, and ran 5v into a trimpot and then to the filament, and ran the HV power to the anodes/grid. After some trial and error I finally got it to light up. It seems to like about 60v.
The IV-18 tubes more than are worthy of a close-up. I am quite enamoured with them.
If you’ve been to youtube lately (or… ever), you have probably noticed evidence of the over-population of Windows Movie Maker users, eager to show off their madskills at generating white text on a blue background with their favorite to 90’s track pumping. Advanced users will also add real, full-color photographs with transitions such as wipe and crossfades.
For people who are playing with video editing software for the first time, this is great. They can throw something together and upload it to YouTube in a few minutes, and gain some sense of accomplishment without having to do anything overly technical.
For people who are on YouTube looking for an actual video (which I assume is all or most users), this is just awful. I’ll go out on a limb and assume that most people searching for something on YouTube are not overly critical, they will tolerate mediocre-quality video and amateurish production – that is the whole appeal, but it is fair to assume that these people are, at the very least, looking for video. Not slideshows.
And you can’t blame the MovieMakers. Making slideshows is easy, and fun. But what I cannot fathom is how YouTube has fostered the growth of this truly unwelcome phenomena on their service. If I am searching for an eagle attacking a wolf, I would expect, at the very least, to see an eagle attacking a wolf, not a handfull of photographs someone found of the same, with a cheezy hip-hop midi in the background.
Considering all of the energy Google has devoted to identifying copyright-protected content in YouTube, you would think they could use some of that same mojo for at least identifying what content is actually a video. How many “videos” for example, are there out there, that are just a still image with a full-length copyright song? Come on!
The technology required to identify slideshows and still images is pretty basic. This could be implemented where the user uploads a video. They could tick off a radio button that identifies the content as “full motion video” or “slideshow”. Viewers could also flag content as a slideshow when the creator does not. And furthermore, the video itself could be analyzed, either when it is uploaded, or on a random / periodic basis, to determine if there is motion from one frame to the next. Really, there is no excuse for them to leave all these slideshows “in the soup” of a site intended for video.
All it takes is a little “Mark as slideshow” link, like this:
