The ChronodeVFD is a personal project I’ve been working on for a couple of months. It’s a wristwatch built around the IVL2-7/5 VFD display tube. I originally purchased a few of these tubes to build a standard desk clock, but after playing around with them, I realized I could probably build a wristwatch too. The tube has a number of features which make it more suited than most Soviet-surplus VFDs for this purpose.
- nominal 60mA filament current @ 2.4V, but still works with ~35mA @ 1.2V.
- It’s small — only 1.25 x 2.25″
- It’s flat, as opposed to the round tubes like the IV-18, which would be much clunkier in a watch design.
- can operate from a relatively low grid voltage of 12-13V (up to 24V)
- pulls only about 2.5mA/segment from the grid rail @ 12.5V. (“8″ = 20mA)
One other feature that I like about this device is that unlike nearly every other VFD tube, the IVL2-7/5 has no opaque or diffuse backing behind the digits. It’s completely transparent front to back, which means that if you put it on top of a circuit board, you can (with a bit of backlighting) see the PCB below.
The circuit is pretty straightforward. The core is an Atmel ATMega88 AVR, and the real-time clock is a Maxim DS3231. The DS3231 is nice because it’s an all-in-one solution: it has a 32kHz temperature-compensated crystal and capacitors built in, so the only external component required is the backup battery. However, the DS3231 is pretty expensive. I happen to have a tube of them left over from another project, so I decided to use it in this design. There are other, much less expensive RTCs available which would work quite as well (BYO crystal, though).
The VFD display is driven by a Maxim MAX6920 — a 12-bit shift register with high-voltage (up to 76V) outputs. The 12-bit field is designed specifically for driving 4-digit clock displays (4 digit channels and 8 segment channels). It’s easy to use and very reliable and compact. It really is the only chip of its kind, so it’s a tad expensive. It’s also possible to drive the VFDs using a bunch of discrete components, but that was impractical here due to space constraints. If you’re interested in that, you should check out Riad Wahby’s excellent inGrid clock build, which describes his discrete HV driver circuit in some detail.
The circuit is powered from 3 voltage rails. The first is the battery voltage itself, which is used to drive the display filament (switched by the micro with a low-side MMBTA42 NPN). The battery voltage also feeds into the 5V boost converter (MCP1640 SOT23-6), which drives the #2 rail. This 2nd rail powers the AVR, DS3231, and MAX6920 logic, as well as acting as the input voltage to the second boost converter (NCP1403 SOT23-5), which produces the 13.5V VFD grid voltage (the third rail). The NCP1403 is enabled/disabled by the microcontroller, so it’s only in operation when the display is lit, which helps limit current draw from the battery.
I included jumpers (0805 pads) on the board so that I could switch the filament power from the battery to the 5V rail, in series with a current limiting resistor. In this configuration, the filament stays at a steady brightness for the life of the battery, and the display is brighter. However, the load on the 5V rail is higher, and considerable power is wasted as heat through this resistor, so the battery life is significantly shortened.
Also included on the board are a number of sensors: one analog and two digital. The analog sensor is a phototransistor used to detect the ambient light level (Q2). The digital sensors include a BMP180 barometric/temperature sensor and a MMA8653 accelerometer, to detect movement. Neither of these are populated in the photos above, though I may decide to add them in the future. Both digital sensors share the I2C bus with the DS3231, and an I2C breakout is provided on the bottom-left edge of the board, for troubleshooting or expansion purposes.
Other headers on the board include a 6-pin FTDI serial block, breakouts for two AVR ADC inputs (one is used by Q2), and test points for the battery voltage and boost converters. The AVR ISP 6-pin header is on the underside of the board beneath the display, to allow programming after assembly is complete.
The cage is built from extruded brass tubing — sometimes called ‘telescoping’ tubing, because each size nests inside the next size up, with a clearance of less than 5 mils. This makes it well-suited to the type construction used here: thinner straight pieces fitted into slightly larger elbows. This brass tube is commonly available from hobby and train stores (and Amazon), and it’s used often for scratch-building model engines and locomotives. It’s available in every size from ID=1/32″ to ID 1/2″.
I made the elbow pieces using 1/8″ OD stock in an inexpensive tube bender. As purchased, this tubing arrives in a hardened state, so when it’s used in the tube bender, the wall tends to collapse a little bit — note the dimples on the outer surface of the elbow pieces. If I wanted to maintain the integrity of the tube during a bend, I’d have annealed the tubing first. However, I needed a tighter turn radius than you’d get with annealed tubing, so I didn’t bother. Structurally, it’s still strong enough for what I need it to do. If I wanted to reinforce the corners, I could always place a solder fillet on the inside of the bend.
The roll cage is soldered to the PCB at four points — plated through-holes slightly larger than the OD of the straight brass tubing. This allowed me to plan the placement of the cage during PCB layout, and also allowed me to use one of my spare PCBs as a precise fixturing jig while I put together the various pieces of the cage.
To make the elbows I placed a full (or near full) length piece of stock in the tube bender, bent it to 90 degrees, and then trimmed off the bulk of the excess length with a hacksaw. I then used various needle files to dress it down to the proper size, and clean up the rough edges. There’s enough slack length in the design that didn’t have to precisely match all of the elbows to the same size, though I did try to get them close. Likewise, the straight pieces were cut to approximate size using the hacksaw, and the ends dressed appropriately.
The watch band for this project is a leather cuff I bought on Amazon. I like it because it looks good, has multiple mounting points (if you’re creative), and it’s comfortable to wear. The three straps on the cuff are handy because I can adjust for the taper of my forearm and it still fits well.
I went through several iterations and experiments before I finally settled on a method to mount the PCB to the cuff that provided enough flexibility, while at the same time ensuring a secure fit. The method I settled on was what I call an underslung loop (photo above).
At the top and bottom edges of the PCB, on the underside, there are two 1.25″ strips of bare copper (courtesy of a polygon on the bottom stopmask layer). Soldered to this is a piece of the thinner brass tube stock. Looped around these are the two strap loops, made of 12AWG solid conductor household wire. Both of these loops fit onto the center strap at the smaller openings on the sides of the cuff. The top loop is a bit longer than the bottom loop so that the watch has a slight cant towards the user when worn. When the cuff is removed, the loops hinge out a bit so that the whole thing lays flat. Removing the PCB from the cuff is just a matter of pulling out the center strap — reattaching means rethreading the whole thing, but this is pretty quick and easy too.
Tools and Materials
The following mechanical materials were used to build this project. The board and schematic files and electrical BOM will be released in the coming weeks.
- C260 Brass tubing: 3/32″ OD and 1/8″ OD.
- Brownbeans 3-strap leather cuff.
- 12-gauge solid copper household wire
- pipe solder flux
- pipe solder (you can also use 60/40 electronics solder in a pinch)
- drill press vise — I used my trusty Craftsman vise, but any heavy drill press or machining vise with flat jaws and a v-channel (to seat the tubing) should work well.
- small needle files — at least one flat and one round, no wider than 3/16″
- flat diamond files (for squaring off tube ends)
- hacksaw (preferably a smaller, hobby-sized hacksaw)
- 300 and 600 grit sandpaper, to remove tabs from edge of PCB and clean up brass tube ends.
I’m a freelance electrical engineer based in New York City. I work with startups and agencies in every stage from concept to prototype to manufacturing, and I’m particularly interested in the wearables and internet-of-things space. If you’re interested in working together, get in touch: email, Twitter or LinkedIn.
Date: 21 October, 2014
Categories: EE - project & practice