The calibration process is only for the dust sensor part of the
DCES-DTRHF, the other sensors being already calibrated.
Calibration is not an easy process, for the dust sensor and since this
sensor is a data centre environmental sensor it is designed to be very
good at detecting any dust anomalies even if not calibrated*
Even if you correctly calibrated the dust sensor, you should expect
some variations because optical particle scattering is impacted by
relative humidity*, therefore if you need really precise measurements
at high frequency, you need to go for a commercial model with a regular
professional calibration service.
As said earlier, this sensor is aimed at data centre environmental
Often in project on the Internet, Sharp GP2Y1010AU0F readings are
transformed to measurements following the specification sheet graph,
which in practice can be far from calibrated measurements.
Here is the curve from the spec
The only good news is that it states that measured voltage is a linear
function of dust density.
First of all, to achieve correct calibration you either need to put the
sensor in an environment containing a precise amount of dust particles
of a given size...
Or more easily you can go for relative calibration and calibrate the
dust sensor against a calibrated commercial sensor.
I put the DCES-DTRHF sensor and the DC1100 in a cardboard box with a
hole, started the measurements and collected the data from the 2 sensors
and sprayed some propylene glycol at different points in time in larger
and larger quantities. Propylene glycol degrades exponentially and allow
to plot a nice calibration curve and no residue remains in the
The remaining part is to add a small software layer that applies
calibration factors on the UART collected dust readings:
Calibrated_measurement = A * UART_reading + B
Here is a curve showing the calibration result of dces-dtrhf-ser1ch-v1
against a Dylos 1100
This graphs shows >5um particles density per m³ for the 2 sensors over
time. The 3 dust peaks respectively correspond to 2, 3 and lastly 4
sprays of propylene glycol in the cardboard box containing the two
The results are very close but dces-dtrhf-ser1ch-v1 is performing 20
times more measurements than the Dylos sensor.