Forums

a mysterious sensor

Started by Jan Panteltje April 30, 2017
Hi,
been doing some sensor stuff.
I connect those to a PIC micro and an POE ethernet module, and send UDP data that way.

Today was the turn for the relative humidity and temperature sensor DHP11.

It is supposed to be within a few percent accuracy for RH.

However with 60 % RH it displays 20 %....
I have 2 and both do the same, software is good and checksum of communication is good.

So started googling, seems MANY people have the same problem,
nobody solved it, some people just wrote some correction math...
That is what I am doing now too (and ordered an other sensor type).

But, playing with it, trying to find out what I did wrong or what was wrong with those sensors:

I noticed that if I put my finger on it the reading went way up from 20 % to over 60 %.

Heat? no, also if I put my finger 5 mm away next to it...
Tried a light bulb (Edison type), hot, no effect on RH.
Tried a hot piece of metal: no effect on RH.
Tried a dark screen, tried a high brightness CREE flashlight: no effect.
Tried radiation sources (uranium marbles, thorium rods): no effect.

Tried a wet piece of cloth next to it, way up it goes.
But but, my fingers are DRY.

So dry I need to wet those to open a sandwich bag...

So WTF is going on ?
Interesting physics, it seems to sense blood THROUGH the skin at a distance?? (theory 1).

But HOW come these sensors are wrong for so many people ?
It says in the data sheet that there is an internal lookup table in the on board micro it has for calibration (not use accessible AFAIK).

Now I remember a discussion here years ago about customs using Xrays to check things, erasing chips,
could have happened and would explain why so many people have this problem (theory 2),
especially as datasheet mentions:
 5) Light Affect
        long time exposure to strong sunlight and ultraviolet may debase DHT's performance.
 UV, Xray?

But a sensor that can detect blood through the skin?
Or has our body a moisture halo around it all the time?

Anybody here played with the DHT11?

Any application for this effect?




On Sunday, 30 April 2017 19:19:10 UTC+1, Jan Panteltje  wrote:
> Hi, > been doing some sensor stuff. > I connect those to a PIC micro and an POE ethernet module, and send UDP data that way. > > Today was the turn for the relative humidity and temperature sensor DHP11. > > It is supposed to be within a few percent accuracy for RH. > > However with 60 % RH it displays 20 %.... > I have 2 and both do the same, software is good and checksum of communication is good. > > So started googling, seems MANY people have the same problem, > nobody solved it, some people just wrote some correction math... > That is what I am doing now too (and ordered an other sensor type). > > But, playing with it, trying to find out what I did wrong or what was wrong with those sensors: > > I noticed that if I put my finger on it the reading went way up from 20 % to over 60 %. > > Heat? no, also if I put my finger 5 mm away next to it... > Tried a light bulb (Edison type), hot, no effect on RH. > Tried a hot piece of metal: no effect on RH. > Tried a dark screen, tried a high brightness CREE flashlight: no effect. > Tried radiation sources (uranium marbles, thorium rods): no effect. > > Tried a wet piece of cloth next to it, way up it goes. > But but, my fingers are DRY. > > So dry I need to wet those to open a sandwich bag... > > So WTF is going on ? > Interesting physics, it seems to sense blood THROUGH the skin at a distance?? (theory 1). > > But HOW come these sensors are wrong for so many people ? > It says in the data sheet that there is an internal lookup table in the on board micro it has for calibration (not use accessible AFAIK). > > Now I remember a discussion here years ago about customs using Xrays to check things, erasing chips, > could have happened and would explain why so many people have this problem (theory 2), > especially as datasheet mentions: > 5) Light Affect > long time exposure to strong sunlight and ultraviolet may debase DHT's performance. > UV, Xray? > > But a sensor that can detect blood through the skin? > Or has our body a moisture halo around it all the time? > > Anybody here played with the DHT11? > > Any application for this effect?
fingers are far from dry. NT
Jan   you might check the data sheet for response time. Most of the ones 
I've seen are very slow.

Hul

Jan Panteltje <pNaOnStPeAlMtje@yahoo.com> wrote:
> Hi, > been doing some sensor stuff. > I connect those to a PIC micro and an POE ethernet module, and send UDP data that way.
> Today was the turn for the relative humidity and temperature sensor DHP11.
> It is supposed to be within a few percent accuracy for RH.
> However with 60 % RH it displays 20 %.... > I have 2 and both do the same, software is good and checksum of communication is good.
> So started googling, seems MANY people have the same problem, > nobody solved it, some people just wrote some correction math... > That is what I am doing now too (and ordered an other sensor type).
> But, playing with it, trying to find out what I did wrong or what was wrong with those sensors:
> I noticed that if I put my finger on it the reading went way up from 20 % to over 60 %.
> Heat? no, also if I put my finger 5 mm away next to it... > Tried a light bulb (Edison type), hot, no effect on RH. > Tried a hot piece of metal: no effect on RH. > Tried a dark screen, tried a high brightness CREE flashlight: no effect. > Tried radiation sources (uranium marbles, thorium rods): no effect.
> Tried a wet piece of cloth next to it, way up it goes. > But but, my fingers are DRY.
> So dry I need to wet those to open a sandwich bag...
> So WTF is going on ? > Interesting physics, it seems to sense blood THROUGH the skin at a distance?? (theory 1).
> But HOW come these sensors are wrong for so many people ? > It says in the data sheet that there is an internal lookup table in the on board micro it has for calibration (not use accessible AFAIK).
> Now I remember a discussion here years ago about customs using Xrays to check things, erasing chips, > could have happened and would explain why so many people have this problem (theory 2), > especially as datasheet mentions: > 5) Light Affect > long time exposure to strong sunlight and ultraviolet may debase DHT's performance. > UV, Xray?
> But a sensor that can detect blood through the skin? > Or has our body a moisture halo around it all the time?
> Anybody here played with the DHT11?
> Any application for this effect?
Welcome back, Jan. We've missed you. 

Skin has liquid water on one side and relatively dry air on the other. The concentration gradient drives a lot of water vapour flow from the skin to the air. 

Cheers

Phil Hobbs
I seem to rember those sensor needing to be reconditioned if they have been stored wrong. Some like a 2 hour bake at 100c 
On Sun, 30 Apr 2017 15:04:11 -0700, pcdhobbs wrote:

> Welcome back, Jan. We've missed you.
+1
Jan Panteltje wrote:

> Hi, > been doing some sensor stuff. > I connect those to a PIC micro and an POE ethernet module, and send UDP > data that way. > > Today was the turn for the relative humidity and temperature sensor DHP11. >
I've been using Phys_Chem Scinetific humidity sensors in my environmantal monitoring system for 20+ years. They seem to work quite well, and although I have not rechecked the calibration, it seems to still be reasonable. These are capacitive sensors, and need a circuit built out of an Analog Devices audio chip. Jon
On Sun, 30 Apr 2017 18:19:05 +0000, Jan Panteltje wrote:

> Hi, > been doing some sensor stuff. > I connect those to a PIC micro and an POE ethernet module, and send UDP > data that way.
Welcome back! Have you tried checking the calibration of the device? One fairly straightforward/cheap way is to create a slurry of a salt and water. The particular salt (in a small enclosed space) forces a particular humidity. Do a web search and you'll find a few examples.
pcdhobbs@gmail.com wrote:
> Welcome back, Jan. We've missed you.
+1 Thank you, -- Don Kuenz KB7RPU
On 01-May-17 7:02 AM, Cursitor Doom wrote:
> On Sun, 30 Apr 2017 15:04:11 -0700, pcdhobbs wrote: > >> Welcome back, Jan. We've missed you. > > +1 >
+2