Page three - the progression of the SMiRF (Serial Miniature RF  device). The SMiRF started in June of 2004 when the 2.4GHz transceiver module  was successfully used to transmit data from one module to another.

These modules are very flexible and powerful, but the amount of  firmware to get a link up and running is daunting to the first time user. Many  users simply needed to move serial data from point A to B. This could be done,  but it required an separate microcontroller to receive incoming serial  bytes and handle the protocol interactions with this module. Our goal was to  make a 'black-box' product. The user shouldn't care how the data gets there, all  they needed was a way to send serial strings over a wireless connection.

The very first proto was two RF-24Gs in a breadboard with two  16F88s. Once the idea was proven, we used the bare-bones PCB  service from Advanced Circuits. You can see those boards on the left.

There was all sorts of problems with layout, but this was also  the very first time we had soldered such tight pitch components. This  board uses the CP2102 (it was the CP2101 at the time). The first time we attached this  board to USB it was shear terror. The first time the board actually enumerated  and a Virtual Com Port, it was shear euphoria. Once you solder a leadless, tight  pitch IC like this, and get it working, all other assemblies begin to look easy  - at that eventually became a problem! (More below)

Here we have 'v02' in 7-28-04. Notice the footprint for the  RF-24G is flipped around the wrong way. Ugh. You can also see the internals of  the RF-24G module with the shield removed. That looks like a QFN like the CP2102.  That means we could possibly solder that ourselves...

More testing 8-10-04. The board could be populated to become  either a 4-pin serial device or it could populated with a USB connector and  CP2102 to become the 'base' unit. Because the parts were on both sides of the  board, each unit had to be hand soldered. I believe we ended up selling a few  pairs on units with this external module/backpack design.

The next step was obviously to incorporate the Nordic IC and  circuitry onto the same board. This board was our first attempt at soldering the  nRF2401 and surrounding 0603 components. It was poorly laid out (for 2.4GHz  freq!) and used a rather large 16MHz crystal, and QFN packaged 16F88. This  device was to use a Nordic RF circuitry, and PIC, to wirelessly transmit  accelerometer data. We  used this board to prove the nRF circuitry as well  as develop a new product. Notice the lack of antenna - there was supposed to be  a piece of wire-wrap-wire that would be soldered onto the board near C11. Much  to our surprise, the board worked! We were able to actually get this thing to  talk to our original module-based SMiRF. Excellent! The board emitted enough RF  noise that it probably wouldn't pass FCC testing. Range was limited to about a  meter, but it worked!

So we quickly rev'd the board to include a power switch (what  a crazy idea), status LED (handy), PCB trace antenna, and a smaller and much  cheaper 16MHz resonator. The board didn't work. So we trouble shot the board,  checked the connections, re-checked, re-programmed, nothing. So we built a  second board - tediously checking everything. Nothing. This was a painful lesson  on the difference between a resonator and a crystal. The Nordic IC requires  16MHz clock source to operate. The clock needs to be within 20ppm (parts per  million) so that two units can understand each other at 2.45GHz. This is a  pretty tight tolerance, but obviously needed for such high frequency  communication. A resonator is a piece of ceramic cut in such a way that it  shakes or 'resonates' at a given frequency. The process is much cheaper than the  equivalent quartz crystal process. However, where the quartz crystal can be  created with 20ppm tolerance, a resonator can only achieve +/-0.5%. 20ppm is  +/-0.00002%. Ahah! The board is functioning just fine, but the 16MHz resonator  is so sloppy, no two Nordic circuits can talk to each other. AHHHH.

So a quick retro-fit of the original 16MHz quartz crystal and  the board came to life. What a great sanity check that was. So this proto taught  us that the nRF circuitry would work, the PCB antenna worked well enough  (limited but ok range of 10-15 meters), and we would need to find a suitable 16MHz SMD crystal  that was smaller, and slimmer in height.

And here is the eventual super slim design. Date your boards -  we forgot to. This design was fabbed around 9-15-04. This is a rather small  design constituting an accelerometer, controller, battery, and RF link. Sound  familiar? This is basically a foot-pod like the Nike+iPod thing or the  ANT based Garmin Forerunner.

Here was the collection of boards that got trashed because of  the faulty resonator design.

Here are the protos for the SMiRF with etched antenna. We  found that this lump-node antenna didn't help our reception range so we resorted  back to the trace antenna. The boards did work however! Sometimes I really miss  the days of shiny leaded solder...

The 'final' rev. Here you can see the USB populated board with  stub trace antenna using a larger 16F688 PIC (and a small jumper wire fix).  Worked well enough, but the firmware was always a problem. You really couldn't  push the data rate past 9600bps and the system would drop packets if the range  was any more than 20-30 feet. This was a hassle. Around this time we started  experimenting with the Bluetooth module from Mitsumi.

This  Bluetooth module was a pre-packaged solution to our  serial cable replacement problem. It was everything the SMiRF was trying to be,  just with a whole team of Mitsumi engineers behind it. The cost of this module  is much higher than the part costs of the Nordic solution, but when the  development cost was weighed against the low-purchase volumes, we decided to  abandon the fun SMiRF and created the  BlueSMiRF based on this technology. Works  great!

   
• Page 1 - The general collection
   
• Page 2 - The development cycle that started SparkFun Electronics
   
• Page 3 - The epic progression of the BlueSMiRF

This constitutes a small portion of the hard work we put in  here at SparkFun Electronics. Thanks for checking us out! Please use this  information at your own risk.

January 7th 2007