Recently we already discussed a part of this : when one hast two T sensors, of which the second is within the sensor enclosure.
In such a case the ‘internal’ sensor is ‘too hot’, because of the electronics dissipation, and the very reactive folks here already published an update to somehow cancel that second wrong reading.
But I found that in my specific kit (others may not be like that) even the external T sensor is really biased by the thermal dissipation from the inner electronics.
It is very significant indeed, as soon as one compares with a basic thermometer : a matter of more than 1°C.
This comes from the fact the T sensor, even inside its nice grey external tube, happens to be in the ‘output flow’ from the particle sensor and electronics.
Indeed the particle sensor sucks air through the other tube (the black one on the side), and this sucked air has no way to get out other than through the grey T tube, as everything else is water-tight.
So, all the air heated by the electronics flows around the T sensor. Result : ~+1°C.
What I did is the following :
create some venting holes in the enclosure, through which the heated air will get out. I did it on the transparent cover, aware that this will cancel the water-tightness*, see picture 1
close the air leak through the grey T sensor tub : this I did very simply with some basic cotton, as you can see on picture 2.
The result is striking : the measured temperature now is 1°C lower, and perfectly in line with both my basic thermometer and the city T measurement close to here.
picture 1:
picture 2:
(*) you may want to drill somewhere else, and/or to protect the new holes from water -for me this way was OK because I have a small casing around the detector, open but with a roof, so no water
The increased temperature due to dissipation though might protect from condensation during the winter. Is it that important to have a lower reading by risking oxydation?
@air-ais
Well, one could say that If the choice is between risking oxidation or publishing wrong results, it’s everyone’s decision.
But for me I choose accuracy.
In the previous state of my device I was at risk to be just ridiculed by colleagues (‘simply from your T measurement, your device looks pathetic’), and I sincerely consider this is harming the whole sensor.community.
Some herearound are expecting more ambitious devices with redundant sensors, etc. -things that become costlier, and harder to afford.
I for one aim for correct measures within the existing hardware, and that one correction was easy…
Although in this case It was not by design, it is not unusual to have both internal and external temperature sensors in any type of device. That way you would be able to identify early problems such as overheating.
Sorry man my bad, I had in mind the other thread about the two readings and then read quickly this one so I thought you did that to get the same reading from both temperature sensors.
That could be a design issue indeed. I always wondered whether that hole was adequate, since sensirion schematics suggest larger openings. However could it be something on the installation that is preventing airflow which might also mess up with your pm readings?
Wise remark. I bought the so-called ‘pre-soldered kit’, which came with the main circuit already installed within the enclosure. Then once finalized I just fixed the complete kit inside a wooden case that is fully open on the front size.
So, I presume everything flows around it, but if for instance there is an output hole in the back I’ll have missed it, and now it is fixed on the case. I have to check this. But it would be the only possible issue in my mounting, I think…
