Coby CX-61 FM Radio Dissection

This radio cost a little under $9 at a shop in the Mall del Sol in Guayaquil. It has a small LCD display (four-digit seven-segment with a decimal point, a colon, and three or four other indicators) with an optional backlight (two orange LEDs), six buttons, and a volume potentiometer with thumbwheel that doubles as a power switch. It uses the cable on the earphones (included in the price) as its antenna. It runs off two AAA batteries, and I didn't notice any trouble running it for three hours continuously on the bus ride from Guayaquil to Cuenca.

The whole thing is about three inches long and an inch in diameter. It normally hangs from the earphones, which fit around your neck like a necklace; the case is smoothly shaped and bright white, with a transparent yellow window (smoothly continuing the curve) protecting the LCD.

Within

When I open it up, I see three electrolytic capacitors (220uF, 100uF, 100uF), two 1/8W resistors, the potentiometer thumbwheel, and the headphone jack, along with a little more detail on the buttons.

I got a big surprise when I looked at the other side of the board. I figured that, given the price and the PLL design, the whole thing would be contained inside a single chip, except maybe for a couple of capacitors and stuff. But there are a large number of components, nearly all surface-mount, except for two RF inductors. There are also three discrete transistors (Q2, Q3, Q4, but Q1 is on the other circuit board described later), surface-mount capacitors C6 through C25, surface-mount resistors R5 through R17, a big sort of gull-wing diode, and a 16-pin gull-wing IC, marked CD9088CB FA61GL024J. There's apparently supposed to be a third inductor, but it's been replaced with a resistor on the other side of the board. Plus the potentiometer/switch and the six buttons, and 9 wires running to the display.

But there's more. On the back of the screen, under a piece of black tape, are more surface-mount components (including Q1, R2-4, C1-5, and C26) and a telltale blob of black epoxy, presumably concealing a chip designed specifically for this radio.

Total component count is about 63. For under $9!

It's worth listing the surprising things it can't do:

• tune AM (presumably this would have required some more analog parts, perhaps another antenna)
• remember preset stations (even with the batteries in!)
• turn its backlight off after some time (the light from the backlight decays after you press the backlight button in what I daresay is an RC fashion with a time constant of three seconds or so)
• when scanning for stations, reset to the lowest frequency when it reaches the highest frequency
• reliably progress to the next station when you press the "scan" button — sometimes it finds the same station, perhaps 200kHz higher, then drifts back down.

These suggest that its control circuits are extremely simple. The shiny white and gold case, the metal fitting it hangs from, and the headphones presumably cost far more than Coby were willing to spend on "intelligence" in the device itself; I think I've seen FM radios with two buttons for sale for $3.25 here.

Implications

This is interesting because it demonstrates that it's possible to build and sell an RF communication device for under US$10 at retail, and it has a couple of interesting attributes not found in the other devices I dissected during this project:

• It dispenses with a cumbersome separate antenna by using the earphone cable as an antenna. FM audio is near the bottom of the frequency range where this is feasible.
• The buttons in the Brick Game, the crappy calculator, the other Brick Game, and the nice calculator work by pressing bits of conductive rubber or carbon onto printed contact areas on a circuit board. They may be sturdy (I have my doubts), but they give terrible tactile feedback. By using discrete switches instead, this radio achieves much better tactile feedback. I've seen switches with a similar feeling made out of small bits of metal soldered to a printed circuit board as well, but I don't have photos.
• It has a dramatically larger number of discrete components than any of the other devices, demonstrating that devices with a retail price under $10 need not necessarily be low-component-count devices such as the non-RF devices I examined.

I speculate that it's possible to go a lot cheaper on the components if you only want digital communication, not analog, you get to pick your frequency band (e.g. 2.4GHz instead of 100MHz), and you can live with low bit rates; people can type about 110 bps at most, and talk or read at a little under 300 bps (in ASCII), so 5 or 10 kbps (and thus an effective bandwidth of 5 or 10 kHz) should be sufficient for many human-communication purposes.