Spikes in Humidity Sensor Reading - Not HB related

[shadowhunter]

Greetings all,
Although I currently am not running any hobby-board components as of yet,(on the horizon I think) I am hoping to gain some assitance or knowledge as to an issue I am having with my 1-wire setup.

I have been noticing sporadic spikes in my humidity readings (this ofcourse affects dew point too) There does not seem to be a rhyme or reason as to when it happens. When the humidity is low ( around 50%) I will occasionally get a spike to 100% and then it drops back to the current reading, and on occasion it will drop out to 0% and then back up to normal.

Background on the station:Last year I had an electrical spike, due to a storm which took out the serial adaptor and apparently the temperature sensor in the WS itself too. The wind direction and speed are still functional. I have replaced the DS-9097U(I think) with the Link-12 and have added a TAI-8540A Temp/Hum sensor and this is where I am seeing the spikes. I am not sure if the WS is causing an issue, or if there is an issue with the Temp/Hum sensor.

This past weekeknd I re-ran all of the wire from the roof to the PC, so now everything is in true linear form and all of the same wire. Yet the spikes continue.

I am hoping that someone may be able to shed some light on things be fore I start replacing part by part, which I may do anyway

PC Specs:
PII 450
450 MB RAM
Monitoring with Weather Display


Thanks in advace, my appolgies if this is the wrong place to post.

Jim

 

[ericvic]

Jim,

How many sensors do you have in your system? I have not seen this exact problem but it sounds like it could be a 1-Wire network overload. The 1-Wire network was never designed to provide much power so it can easily be overloaded.

Eric

 

[shadowhunter]

eric,
Thanks for replying, I relaize this forum is for the Hobby-Boards devices and not the AAG, but I have not been able to get answers anywhere else relating to the 1-wire network issues. I greatly appreciate any and all assistance. I will be expanding my network once I get this issue ironed out and will be comming back here to do it..

currently I am only running the WS and the Temp/Hum sensor. I-wire viwer is picking up 3 devices which is correct. So I do not believe it is an overload situation, however in WD under the 1-wire set up for extra sensor values ( i think that's where it is) the voltage to the Temp/Hum senor is like 4.3V. Shouldn't it be closer to 5V all the time.

I thought of another test to see if the WS is loading the network or not. tonight I am going to pull the WS out of the network and run only the Temp/Hum sensor and see if I notice the spikes still. It wouldn't bother me except for the fact that every day it shows I have a 100% Humidity reading , and for data accuracy that is not correct.

Jim

 

[ericvic]

Jim,

Go ahead and try it with just the temp/humidity on the 1-Wire network and let me know how it goes.

Eric

 

[shadowhunter]

eric,
I ran with just the temp/hum senor and am still getting the spikes :angry: this is getting frustrating...The real time graph in WD doesn't show any spikes until 0553 this morning. however the data going to my web page shows I had 100% humidity at 0031 last night :blink: not sure how that works, may need to check with Brian regarding WD. Could this be related to a power issue with the PC? The spikes from this morning coincide with the time that I got up and started turning things on in the house. I am thinking that a new sensor might be in order.

Any thoughts? Again, thanks for your assitance even though currently this does not involve your HB sensors.

Jim

 

[ericvic]

Jim,

I wouldn't think the spikes would be from turning on things in the house unless you are running the 1-Wire wire right next to some AC wiring (which I doubt).

I'm running out of ideas, sorry I couldn't be more help.

Eric

 

[shadowhunter]

eric,
Thanks for your assistance. I have about run out of ideas too. Think I am just gonna order a new sensor. I have a pagoda in place for the outside temp/hum sensor, is a case required for your sensor or is the protective coating sufficient. also is the power injector needed?


Thanks
Jim

 

[ericvic]

Jim,

If you are going to place the sensor in a "protected" environment (like a pagoda) then the moisture coating will be enough. The case doesn't really protect it from moisture (except direct rain), it is more as a way to mount it and to protect it from something banging into it.

As for the Power Injector. You can use the humidity device without external power (like the AAG sensor) but we don't recommend that. If you are going to power the device you can use the Power Injector as the easiest solution or you can provide +12v DC to the device through the screw terminals.

Let me know if you have more questions.

Eric

 

[NormC]

Jim,

I have some humdity sensors on my 1-wire network also and have seen spikes like you are seeing. What I found was it is not realted to using a humdity sensor, but to the DS2438 chip. I wrote all my own SW (VB.Net) to read the devices on the OW network and what I discovered is that occasionally when getting the Vad reading from the DS2438 chip it would return 3FFh, which is exactly 10volts, the maximum value it can return. Obviously it was a bad value so I just coded the SW to ignore this value unless I got 2 of these in a row (which has never happened). Occasionally it would 'spike' to 0 also and I treat it the same way.

I don't know why this happens or what causes it.

Not sure what SW you are using, but the only option you may have is to 'filter' your data to ignore these bad values or if you can tweak the SW.

Hope this sheds some light on this for you.

Norm

 

[shadowhunter]

Norm,
Having a light electronics background what you say makes sense. I am using Weather Display Clicky by Brian Hamilton (awesome software!!) I do not know if it is possible to tweak it to ignore certain readings. You are the first person that has responded to ever seeing the same thing as I have.

I am currently using an AAG TAI8540A Temp/Hum sensor. I am considering changing to one from Hobby-Boards, but I see that they use the same chip (i believe) so I am not sure if that would solve my problem. I do notice more spikes when I have the actual weather station connected to the network. Although my weather station did suffer some sort of electrical incident last summer and the built in temp sensor no longer works, so I am not sure if that is having any effect on the readings from the temp sensor. I have ordered a new weather station to see if this helps.

I have noticed that in weather display the voltage on the sensor is 4.3-4.5V but I am unable to see what it is when it is spiking. Do you know if 1-wire viewer can do that? Also would adding a regulated power source to the network benefit anything?

Thanks for the info, hopefully I can get this sorted out.

Jim

 

[NormC]

Jim,

I have not used the one-wire viewer for a while, but if it has logging cabability then you could see the values. As far as adding a power source, I do not ever use parasite power on the one-wire network, I always run another +5VDC line and I still get the spikes, so I don't think it is related to that.

Bottom line is the data being returned from the DS2438 is bad (rarely) and needs to be ignored. I have noticed this even on a very simple 1-wire network (one DS2438, 2 foot wire length). So it is hard to imagine that whatever is causing the DS2438 to send the wrong data is going to be easy to always avoid.

Maybe send Brian a request to ignore single occurance full scale values recieved from DS2438 chips. Hard to imagine when a single occurance (within a few seconds or so) of a full scale value is going to be a valid value. If when he got that kind of value he just tried to immediatly re-read it and got a normal value then he could ignore the spike and use the good value, or if he got 2 (or more) full scale values in a row, then he could consider them good.

I know that is not much help, but I think a SW solution is best.

Norm