[click to enlarge: 175kb JPG]

Finally, all the boards had to be joined to make up the complete device. Therefore, lots of cables had to be soldered at the correct positions which became pretty messy towards the end. (Hint: Never make these connecting cables too short. Always make them longer than needed.)

Furthermore, a good idea on how to put all these things into a hard case had to be found. I chose a relatively cheap case from Reichelt which was large enough for all the parts. Since it had aluminium front and back plates, I decided to line up the boards using 3mm threaded bars which was a good idea. You can see the result in the image on the left.

The front panel was designed using Eagle. I printed the hole and description layers to be able to drill the front panel holes for the LEDs and keys at the correct positions. The PCB was fixed with several 2mm screws. The 5 digit display is actually not soldered on top of the front panel board but on a separate board behind the front panel and mounted on sockets.

[click to enlarge: 185kb JPG]

On this image one can see the 6 boards lined up. The transformer was fixed at the back plane above the 230V standard female connector (not visible here) and the additional service port (9 pin SUB-D with SPI, Reset, Ground and +5V) was installed. The 5V fixed voltage regulator (7805) was also screwed directly on the back plane so that the heat can spread easily. (Of course it was insulated since the back plate is tied to ground.)

The transparent slides from the PCB creation were used as insulator between the boards (holes for line-up can easily be made using the soldering tab) and tried to make all the cables stay within the limited space provided by the case...

The large flat cable on the left only connects the 5 digit display. Yes, I know, nowadays people are using time multiplex operation but my display does not flicker.

[click to enlarge: 154kb]

Now this is the most critical (and exciting) point: All the boards worked fine on their own in the lab tests -- but would the complete device function as expected? I had already written a first (simple) version of the firmware and downloaded it onto the microcontroller. (Writing the firmware wasn't an easy task either since I had no possibility to test the executed code other than actually putting it into the device. No debugger, no simulator.)

I won't bore you with the details now. If one builds a complex device, the rule of thumb is that it will initially not function correctly unless you are very lucky. (Hint: Always expect it to fail to make your life easier.)

In my case it worked "a bit" and I spent some time to figure out what to do to make it operate properly. The image on the left shows it in frequency counter mode with the first well-working firmware version displaying a frequency of 28kHz which is fed into the device by my small LTC1766-based oscillator.