Governance by those who do the work.

Sunday, June 21, 2015


Last month saw some setbacks.

While measuring a voltage, my probe slipped and shorted the 20 volt supply to a signal pin, burning out the STM32F3Discovery board and many of the electronic components.  The damaged components are now replaced; the electronics are working as before.

The analog-to-digital converters on the STM32F303 have multiple inputs.  While checking voltages on the repaired board, I noticed that for each ADC, the first input to be converted was converted accurately, but the others were not.  Increasing the converter sampling-time from ADC_SAMPLETIME_1CYCLES_5 to ADC_SAMPLETIME_19CYCLES_5 results in all readings being reasonably accurate.  The noise on converted signals is also reduced.

The other setback is more serious.

Heat flow measurements are of heat flow from both sides of the plate. To infer the undesired heat flow from measurements of total heat flow requires that the heat flow from the front of the plate be precisely known.  But the point of this experiment is to measure the heat flow from the front.

An insulating cover for the front of the plate turned out to interact strongly with the convection from the short sides.  The only option remaining is to analyze and estimate the heat flow through the back of the plate.

I am making progress in calculating this, but the estimates are significantly smaller than measured by the experimental apparatus.  I am rebuilding the insulation to match the estimate model.