Thinking of a new gadget...

[Bazbro_imp]

Wow! What a great progress path you've followed, Csory.

This is going to be one great item.

Looking forward to getting mine.

[GoldOxide_imp]

Slick! I really like the summary dashboard - everything at a glance. Well done on the optimizations.

[TwoBrothersBusa_imp]

The new changes look fantastic. Can't wait to see the "new" finished product!

[Csory_imp]

New boards are in one day earlier. Could not resist - put together the first one (no display yet, only electronics.) So far, so good.



Here you can see how I could save some thickness by putting the connector "into" the notch



Housings will have to be re-printed since the opening for the ambient temperature sensor is a tiny bit larger than it should be.

[TwoBrothersBusa_imp]

This is just an amazing project to watch unfold. Thank you so much for sharing.

[GoldOxide_imp]

Really of thing of art, Csory.

[Cormanus]

This is just an amazing project to watch unfold. Thank you so much for sharing.

Well said, TwoBrothers.

[Bazbro_imp]

Hear hear, TwoBrothers.

[Csory_imp]

Today's update

Mated the first of the new electronics with the display. Later on, it will be a no-brainer, but proceeded with caution with the new design. First, powered up without the display and checked the power supply circuit:



It was 3.294V at 12V input, well within limits for the nominal 3.3V. Next, soldered on the backlight and LCD modules - as you can see here, not much space left in between:



Yaay! Everything went smoothly - display lit up, and the programmer was able to "talk" to the unit. Right now it is just a small (and useless) paperweight - first will have to be initialized and calibrated. There is a special program for that, used only during this step of manufacturing. So loaded this calibration routine, and started with initializing the default values:



Next comes the issue with timekeeping. To save weight and size, the BikeGadget uses an internal oscillator built into the processor. This oscillator provides an 8MHz clock frequency, but it is a much less precise source than external quartz crystals. One method to improve precision is calibration. There is a divider built into the chip, and by adjusting the value of this divider the frequency can be tuned. The program emits an 500Hz signal (or what it believes is 500Hz), then the divider is adjusted one step at a time until the emitted signal's frequency gets as close to 500Hz as it can be. Here I could adjust it to 499.984Hz which is pretty close Originally, it was 473Hz - that would have made the ride timer about 5% late.


Now that we have the unit initialized and the clock source calibrated, time to calibrate the voltage measurement circuits. The same logic applies to all three inputs (battery voltage and the two temperature sensor inputs). There is a separate "magic number" for each input used as a multiplier which can be adjusted using the rotary button. By measuring the actual input voltage and adjusting this magic number until the BikeGadget displays the same (or very close) voltage we can have the inputs calibrated permanently. This multiplier is then stored in the EEPROM (non-volatile memory) of the unit, and can not be changed afterwards. (There is a slight variation between components from unit to unit, hence the need for individual calibration. But since these components are soldered in, there is no need or way to change their values - therefore, no need to ever change this ratio...)

In this example, the input voltage was 12.1567V:



By adjusting the coefficient to 16.081, the BikeGadget measured 12.157V - as close as it gets... Note: in normal operation these voltages are displayed using 1-decimal precision, but internally are represented with the full 10-bit resolution.



Next step will be soldering the main cable and casting the electronics into epoxy (using the printed housing as a mould). More about these in the next update

[Cormanus]

Brilliant.