Topic: Test/Evaluation Barani and Co

[tobyportugal]

To finish with the sensors with or without filters, before I left for Portugal (02/11), I put a Pro3 in a closed garage with the two SHT35s (with and without filters).
So no daylight, no wind.
You can see that the SHT35 with filter (blue line) is systematically warmer than the SHT35 without filter (red line), despite the very low amplitude.

My personal opinion, simply the sensors are not aligned, in adiabatic conditions, and I assume it is so, the presence or absence of the filter cannot trigger any difference, it is a simple misalignment dictated by the tolerances of the sensor.

For this reason, personally, I do not take into account differences less than 0.2 °C even if my sensors have been aligned in the range -10 / +40, evaluating such subtle differences requires a certification performed at a metrological institute, which will also issue a measurement uncertainty that will need to be considered

M.

I could have agreed with you, but you can see that from 6pm onwards the sensor with the filter follows the descent, but with a delay.
This contradicts your analysis, if I offset by -0.1°, I'll have a negative drift when the 2 temperatures are identical on this graph.
Here's another example from 03/11, clearly showing the delay of the sensor with filter. (same conditions)

So they are not adiabatic conditions, therefore, in my opinion the comparison is useless
the sensors have a tolerance, excellent in the sht-45, even if we focus, wrongly, on the typical one while it is the average between the typical and the maximum declared to keep in mind
The tolerance is not declared in positive or negative, as it happens for other companies, Sensirion limits itself to giving an estimate of it, but we must take into account that it is not necessarily always positive, or always negative, therefore the uncertainty doubles.

The precision of the sensor should also be considered, which I don't think I've ever seen mentioned in the Sensirion datasheets, this does not nullify its quality but shows the limits of tests of this type.

M.

Maybe it's my translation, but Sensirion declares positive and negative tolerance.
But I don't think we'll ever agree on filters and no filters.   

[mauro63]

To finish with the sensors with or without filters, before I left for Portugal (02/11), I put a Pro3 in a closed garage with the two SHT35s (with and without filters).
So no daylight, no wind.
You can see that the SHT35 with filter (blue line) is systematically warmer than the SHT35 without filter (red line), despite the very low amplitude.

My personal opinion, simply the sensors are not aligned, in adiabatic conditions, and I assume it is so, the presence or absence of the filter cannot trigger any difference, it is a simple misalignment dictated by the tolerances of the sensor.

For this reason, personally, I do not take into account differences less than 0.2 °C even if my sensors have been aligned in the range -10 / +40, evaluating such subtle differences requires a certification performed at a metrological institute, which will also issue a measurement uncertainty that will need to be considered

M.

I could have agreed with you, but you can see that from 6pm onwards the sensor with the filter follows the descent, but with a delay.
This contradicts your analysis, if I offset by -0.1°, I'll have a negative drift when the 2 temperatures are identical on this graph.
Here's another example from 03/11, clearly showing the delay of the sensor with filter. (same conditions)

So they are not adiabatic conditions, therefore, in my opinion the comparison is useless
the sensors have a tolerance, excellent in the sht-45, even if we focus, wrongly, on the typical one while it is the average between the typical and the maximum declared to keep in mind
The tolerance is not declared in positive or negative, as it happens for other companies, Sensirion limits itself to giving an estimate of it, but we must take into account that it is not necessarily always positive, or always negative, therefore the uncertainty doubles.

The precision of the sensor should also be considered, which I don't think I've ever seen mentioned in the Sensirion datasheets, this does not nullify its quality but shows the limits of tests of this type.

M.

Maybe it's my translation, but Sensirion declares positive and negative tolerance.
But I don't think we'll ever agree on filters and no filters.   

Yes, the Sensirion datasheet shows the accuracy tolerance over a wide temperature range that includes negative values.

 [ You are not allowed to view attachments ]

The problem is that that tolerance is the hypothetical error that varies as the range varies but this is not enough to establish whether "that" sensor has an accuracy, for example at 20 degrees of +0.1 or -0.1.

The accuracy is not linear, the error is not linear, the measurement uncertainty must be doubled because we do not know, on that specific value, whether the tolerance error is positive or negative compared to the reference, or rather, to the real value.

Regarding filters, the question is not whether to agree or not, a filter is like a dress for a human being, after all the human body is a very sophisticated contact thermometer, not adopting filters certainly allows the sensor the maximum possible reactivity allowing it to reach the native time constant, but it can be fine for research purposes, at a meteorological level the multi-parameter sensors must be equipped with a filter, obviously its quality and its thermal and hygroscopic characteristics influence the final data.

M.

[tobyportugal]

When a behaviour is repetitive, there is a rational explanation.
I did this type of test for weeks 2 years ago.
But on my return to Belgium on the 23rd, I'm going to remove the filter from the second sensor and you'll be able to see that this behaviour will have disappeared,
there will still be some fluctuations (logical) but not repetitive and above all no longer a delay.

[mauro63]

When a behaviour is repetitive, there is a rational explanation.
I did this type of test for weeks 2 years ago.
But on my return to Belgium on the 23rd, I'm going to remove the filter from the second sensor and you'll be able to see that this behaviour will have disappeared,
there will still be some fluctuations (logical) but not repetitive and above all no longer a delay.

Of course, I repeat, a filter necessarily alters the sensor's time constant, increasing it.
But if you could test the comparison in real adiabatic conditions, which are not those of a closed room, still subject to variations, but for example a climate chamber or a controlled temperature system, see the professional leavening retarder that I use in my pizzeria also for these purposes, you would notice that, after a certain period, which is the time constant plus 4 additional time units, the sensors would give identical data, provided that their accuracy and precision are the same.
The presence of the filter, once the systems have thermalized, has no influence on the data detected by the sensor, it is the entire system that is in equilibrium, a temperature sensor never measures the measurand, but the temperature of itself.

M.

[tobyportugal]

When a behaviour is repetitive, there is a rational explanation.
I did this type of test for weeks 2 years ago.
But on my return to Belgium on the 23rd, I'm going to remove the filter from the second sensor and you'll be able to see that this behaviour will have disappeared,
there will still be some fluctuations (logical) but not repetitive and above all no longer a delay.

Of course, I repeat, a filter necessarily alters the sensor's time constant, increasing it.
But if you could test the comparison in real adiabatic conditions, which are not those of a closed room, still subject to variations, but for example a climate chamber or a controlled temperature system, see the professional leavening retarder that I use in my pizzeria also for these purposes, you would notice that, after a certain period, which is the time constant plus 4 additional time units, the sensors would give identical data, provided that their accuracy and precision are the same.
The presence of the filter, once the systems have thermalized, has no influence on the data detected by the sensor, it is the entire system that is in equilibrium, a temperature sensor never measures the measurand, but the temperature of itself.

M.

We're not talking about the same thing, and it's clear that SHTxx will balance out after a while in a closed, constant environment, with or without a filtre at €100 or €1. But what's the point?
If you want to evaluate or try to evaluate a sensor x or y without having the variables (wind, radiation, light) you need an environment that gives a non-stable temperature with a very small amplitude. In this case it is the sensor that you are studying and not a shelter or the influence of the wind, etc...

[mauro63]

When a behaviour is repetitive, there is a rational explanation.
I did this type of test for weeks 2 years ago.
But on my return to Belgium on the 23rd, I'm going to remove the filter from the second sensor and you'll be able to see that this behaviour will have disappeared,
there will still be some fluctuations (logical) but not repetitive and above all no longer a delay.

Of course, I repeat, a filter necessarily alters the sensor's time constant, increasing it.
But if you could test the comparison in real adiabatic conditions, which are not those of a closed room, still subject to variations, but for example a climate chamber or a controlled temperature system, see the professional leavening retarder that I use in my pizzeria also for these purposes, you would notice that, after a certain period, which is the time constant plus 4 additional time units, the sensors would give identical data, provided that their accuracy and precision are the same.
The presence of the filter, once the systems have thermalized, has no influence on the data detected by the sensor, it is the entire system that is in equilibrium, a temperature sensor never measures the measurand, but the temperature of itself.

M.

We're not talking about the same thing, and it's clear that SHTxx will balance out after a while in a closed, constant environment, with or without a filtre at €100 or €1. But what's the point?
If you want to evaluate or try to evaluate a sensor x or y without having the variables (wind, radiation, light) you need an environment that gives a non-stable temperature with a very small amplitude. In this case it is the sensor that you are studying and not a shelter or the influence of the wind, etc...

Ok, we have a different vision of how to evaluate a sensor but that's okay.
In the metrology lab that I had the honor of visiting the methodology in the thermometry sector was very different but that's how it is.

Good continuation for the tests

M.

[tobyportugal]

When a behaviour is repetitive, there is a rational explanation.
I did this type of test for weeks 2 years ago.
But on my return to Belgium on the 23rd, I'm going to remove the filter from the second sensor and you'll be able to see that this behaviour will have disappeared,
there will still be some fluctuations (logical) but not repetitive and above all no longer a delay.

Of course, I repeat, a filter necessarily alters the sensor's time constant, increasing it.
But if you could test the comparison in real adiabatic conditions, which are not those of a closed room, still subject to variations, but for example a climate chamber or a controlled temperature system, see the professional leavening retarder that I use in my pizzeria also for these purposes, you would notice that, after a certain period, which is the time constant plus 4 additional time units, the sensors would give identical data, provided that their accuracy and precision are the same.
The presence of the filter, once the systems have thermalized, has no influence on the data detected by the sensor, it is the entire system that is in equilibrium, a temperature sensor never measures the measurand, but the temperature of itself.

M.

We're not talking about the same thing, and it's clear that SHTxx will balance out after a while in a closed, constant environment, with or without a filtre at €100 or €1. But what's the point?
If you want to evaluate or try to evaluate a sensor x or y without having the variables (wind, radiation, light) you need an environment that gives a non-stable temperature with a very small amplitude. In this case it is the sensor that you are studying and not a shelter or the influence of the wind, etc...

Ok, we have a different vision of how to evaluate a sensor but that's okay.
In the metrology lab that I had the honor of visiting the methodology in the thermometry sector was very different but that's how it is.

Good continuation for the tests

M.

??
Neither you, nor I, nor anyone else here has a lab.
If you place sensors in a hermetically sealed box, i.e. at a constant temperature, you can simply validate or invalidate a sensor for this single temperature and not for a higher or lower temperature.
The measurement principle of a SHT is still analogue and electrical, with all the associated risks.
For example, until recently SHTs had digital compensation as a function of voltage.
The calibration that Sensirion claims is a myth, Sensirion calibrates the analogue to digital conversion, but never the analogue measurement.

[tobyportugal]

I'm not sticking to my guns, HA rises despite low RH

 [ You are not allowed to view attachments ]

[mauro63]

When a behaviour is repetitive, there is a rational explanation.
I did this type of test for weeks 2 years ago.
But on my return to Belgium on the 23rd, I'm going to remove the filter from the second sensor and you'll be able to see that this behaviour will have disappeared,
there will still be some fluctuations (logical) but not repetitive and above all no longer a delay.

Of course, I repeat, a filter necessarily alters the sensor's time constant, increasing it.
But if you could test the comparison in real adiabatic conditions, which are not those of a closed room, still subject to variations, but for example a climate chamber or a controlled temperature system, see the professional leavening retarder that I use in my pizzeria also for these purposes, you would notice that, after a certain period, which is the time constant plus 4 additional time units, the sensors would give identical data, provided that their accuracy and precision are the same.
The presence of the filter, once the systems have thermalized, has no influence on the data detected by the sensor, it is the entire system that is in equilibrium, a temperature sensor never measures the measurand, but the temperature of itself.

M.

We're not talking about the same thing, and it's clear that SHTxx will balance out after a while in a closed, constant environment, with or without a filtre at €100 or €1. But what's the point?
If you want to evaluate or try to evaluate a sensor x or y without having the variables (wind, radiation, light) you need an environment that gives a non-stable temperature with a very small amplitude. In this case it is the sensor that you are studying and not a shelter or the influence of the wind, etc...

Ok, we have a different vision of how to evaluate a sensor but that's okay.
In the metrology lab that I had the honor of visiting the methodology in the thermometry sector was very different but that's how it is.

Good continuation for the tests

M.

??
Neither you, nor I, nor anyone else here has a lab.
If you place sensors in a hermetically sealed box, i.e. at a constant temperature, you can simply validate or invalidate a sensor for this single temperature and not for a higher or lower temperature.
The measurement principle of a SHT is still analogue and electrical, with all the associated risks.
For example, until recently SHTs had digital compensation as a function of voltage.
The calibration that Sensirion claims is a myth, Sensirion calibrates the analogue to digital conversion, but never the analogue measurement.

No, in fact, the airtight and dark box technique allows verification on a single point
Usually, when I add a sensor, I first do a check in my pizza retarder, in which I can work in a range from -10 to +40, with 1 degree steps, but I usually do it in 5 degree steps automatically with 30 minute programming for each variation
I do not aspire to measure the accuracy, even though in mine the original temperature probe has been replaced with a 4-terminal pt100 in class A, but I can have a fairly reliable overall vision of alignment and precision

the machine is like this

 [ You are not allowed to view attachments ]

the probes inside, the temperature sensor on the left is not these, but a pt100

 [ You are not allowed to view attachments ]

M.

[tobyportugal]

  pluvio test

 [ You are not allowed to view attachments ]
 [ You are not allowed to view attachments ]

[tobyportugal]

Correct electrical installation after several showers over 3 days. The electrical assembly works and counts the tilts correctly.
Lambrecht dismantled for shipment to Belgium for precision testing of rainfall amounts and rates.

 [ You are not allowed to view attachments ]
 [ You are not allowed to view attachments ]

[tobyportugal]

For those interested in technology.
 [ You are not allowed to view attachments ]
 [ You are not allowed to view attachments ]
 [ You are not allowed to view attachments ]
 [ You are not allowed to view attachments ]

[tobyportugal]

SHT45 mounted on the same date in a FARS Barani and a Pro3.
Although in the FARS the probe benefits from natural ventilation, you can clearly see the SHT problem in the FARS.
If you process the negative photos a little, you can see the direction of the forced ventilation.
I can't imagine the result of a SHT in a TS100 without some kind of assembly to minimise the problem.
I'll try to take some photos with a magnifying glass to get a good look at the ‘pollution’ of the membrane.
Note: the 2 SHTs were to the south.

 [ You are not allowed to view attachments ]
 [ You are not allowed to view attachments ]

[tobyportugal]

People aren't really interested, but technology is my thing.
Reminder: same duration of use

Pro 3:
https://u.pcloud.link/publink/show?code=XZTMnV5Zf2TemQP5jq45XeScz1b0xREWhiuy

FARS:
https://u.pcloud.link/publink/show?code=XZqMnV5ZGP2jFpBLyUB1Wd0rq5KLLyEfmoWX

[tobyportugal]

I don't want to die stupid, so I'm thinking and testing.
With an alignment, you simply move the problem from one to the other.
Pro3 = no filter
Pro Ventiled = with filter
Room without any thermal influence (heating, light, ventilation, etc...) the temperature simply follows the ambient fluctuation linked solely to the insulation of the room.

 [ You are not allowed to view attachments ]

[tobyportugal]

In Portugal, there's always something to study, even at night!
Why it's important to monitor humidity.

 [ You are not allowed to view attachments ]

[tobyportugal]

There's been an internet blackout in Belgium since 7.20 this morning! (old cable in the street that doesn't like water) so no more MeteoRain online.
In Portugal, I opted for a temporary solution.
I hadn't received my new SHTs, so I put on SHT35s without filters until the end of January when I returned to Portugal.
The pro3 and the new FARS have been in service since midday. The FARS gives me a good impression. It seems that the new tube solves some of the overheating. To be confirmed.
I've placed my Helix and MeteoRain with the 2 shelters, and they're very close. I'm thinking of abandoning the WS67 because the radiation is underestimated and the light wind inaccurate.
2 essential parameters, but I need to redo my process and link 2 data sources in a coherent way. That's where the work lies.

 [ You are not allowed to view attachments ]
 [ You are not allowed to view attachments ]

[Jasper3012]

Can you give a clear picture of what shields you currently have and at what location they are installed? You seem to be moving around and setting up your stations constantly, at least I'm getting that impression?

[tobyportugal]

Can you give a clear picture of what shields you currently have and at what location they are installed? You seem to be moving around and setting up your stations constantly, at least I'm getting that impression?

Yes, I know ... my wife tells me I've been complicated for 40 years. 
So, either working or in the process of being installed (interesting equipment):
In Portugal: 1 Pro3, 1 FARS new version, 1 MeteoHelix, 1 MeteoRain, 1 MeteoWind, 1 Comet (dormant for the moment)
In Belgium: 1 Pro3, 1 FARS reconfirmed in a new version by Barani scheduled for December (it was originally in Portugal), 1 Rad14 (which comes from Portugal), 1 MeteoRain.
A MeteoHelix is planned, where a Lambrecht rain gauge (from Portugal and tested on my Portuguese Helix) will be fitted.
I would have liked another shelter in Portugal and Belgium, but that didn't work out.
The SHT45 sensors without filters will be replaced by a version with a PTFE membrane.
As I've run out of SHT45s, I've fitted new SHT35s without filters in Portugal in the meantime.
I make no secret of the fact that I'm looking for other sensor solutions to solve this unique and undated EP measurement.
EP that I modify where I keep only the transmitter.
So on Thursday 21, 2 large parcels arrived in Belgium with my equipment and tools (Fluke, soldering station, PCB, etc.).
For family reasons I'm obliged to divide my time between Portugal (where I've been living and working since 2010) and Belgium for my 92-year-old parents-in-law.
I'm in Belgium from 23/11 to 25/01, then back to Portugal for 3 weeks and so on. 

[tobyportugal]

Yesterday 2pm -> today 2pm

 [ You are not allowed to view attachments ]
 [ You are not allowed to view attachments ]

[tobyportugal]

An uncomplicated day, while we wait for the night-time temperature jumps. Here in Portugal, this can happen even in winter.
On the other hand, for the 2nd day in a row we can see that the HR of the Pro3 is not good, even though the sensor is new!
It's permanently above 4% compared with all the others.
PS: a SHT with two separate sensors for t° and RH.

 [ You are not allowed to view attachments ]
 [ You are not allowed to view attachments ]
 [ You are not allowed to view attachments ]

[tobyportugal]

A good omen. The heating inertia after reducing/stopping the engine seems to have disappeared.
To be confirmed with time, of course.
Back to Belgium tomorrow, in December, curious to see the FARS new in the Belgian fridge. 

 [ You are not allowed to view attachments ]

[tobyportugal]

 [ You are not allowed to view attachments ]
 [ You are not allowed to view attachments ]

[Jasper3012]

Can you clarify the shields you're comparing in this latest test?

[tobyportugal]

Can you clarify the shields you're comparing in this latest test?

The Pro3 against the new version of FARS Barani.