As an exercise in electronics design and soldering, I had the idea of recreating the visual of a mercury or alcohol thermometer using smd components. The circuit features an LM35 temperature sensor amplified by an LM293 op-amp, 30 LEDs powered by 15 LM393 dual comparators LM293 dual op-amps comparing the op-amp output with the values of a voltage ladder made of discrete resistors. All resistors and LEDs are of size 1206.

Front of the LM35Hg circuit board
Back of the LM35Hg circuit board

Being a first attempt, the circuit and process had a few issues:

  1. Comparators like the LM393 are not push-pull like their op-amp counterparts, but instead open drain. This means they can not pull their outputs high, which is an issue if you are trying to drive LEDs on the output.
  2. SMD pads are much easier to delaminate compared to through-hole pads. This is an issue when attempting to remove those 15 LM393s.
  3. Op-amps like the LM293 have an upper voltage range limit of vcc - 1.5V. This means, on a 5V supply, they will stop functioning as comparators at around 3.5V. No LEDs above 22°C will light in the current circuit, no matter the temperature.
  4. 5V usb power supplies are not exactly 5V. A voltage ladder being driven directly off of such a supply may skew the output significantly.
  5. 30°C may be too low for some indoor situations.

From this, the following fixes and improvements will be implemented in a 2.0 revision:

  1. Use LM293 op-amps as comparators. Hysteresis is not a problem as the output does not need to be clean.
  2. Supply the voltage ladder with a regulated source, potentially using a 3.6V zener diode and the free half of the temperature sensor op-amp in a voltage-follower configuration.
  3. Modify the amplification and voltage ladder to match a higher temperature range. I am considering a new combination of E27 resistors resulting in a temperature range of 6 - 35°C

Closeup of the LM35Hg circuit board

With 111 SMD components, the circuit is certainly a way to get familiar with manual SMD soldering, without it being too difficult. Throughout the second attempt, i realized that two terminal SMD components are easily soldered by letting the surface tension of solder on one pad drag the component into place. Afterwards the second terminal can be soldered without issues. The components will be slightly offset, so if centering is essential, both pads must be heated at the same time.

The circuit is tracked in a git repo at github.