Temperature data spikes revisited

[Gina]

AND... just a couple of hours after plugging in the wind sensors we have a spike, or two spikes even. I checked the earths again and they seem fine.

24hrs-Temperature-only3.png

[beteljuice]

I did wonder with the earlier graph, but thought you must have plugged the wind sensors back in ....

How are you getting a windchill reading

[steve]

How are you getting a windchill reading

I wondered that too, but thought I had lost track of where the experiment was up to...

[Gina]

I think my wind chill calculation must be wrong even if it is "apparent temperature" or maybe it's just wrong to call it "wind chill". Either way it's not right. I still have plenty to do to my software!

[Gina]

The blue plot is derived from an Australian formula for "Apparent Temperature" so I guess calling it "Wind Chill" was wrong. Maybe this is better...

24hrs-Temperature-d.png

[nking]

If the spikes occur when the wind lead is connected I wonder if it's worth connecting the new anemometer without using the extension on a temp basis as it may still be something to do with the extension.

[6719jason]

If the spikes occur when the wind lead is connected I wonder if it's worth connecting the new anemometer without using the extension on a temp basis as it may still be something to do with the extension.

I think you may be onto something there. I recently moved my whole setup and relocated it, during which I disconnected it all and reconnected it and I soon noticed that the Temperature was consistently spiking and so I removed the extension piece after a couple of hours and all has returned to normal.

I must point out that before I moved the setup, it was using the extension with no spikes.

[Gina]

I might try that but I do think it's the height of the mast/sensors that causes/enables/amplifies the problem so I would not expect to see spikes. As I recall, the spikes only came when I put the wind sensors on a high mast.

Here's the latest temperature (only) plot. It covers the period since the wind cable was re-connected.

24hrs-Temperature-only4.png

[Super-T]

As a matter of interest, I have been getting occasional spikes, normally positive of .4 or .5 degrees C, nothing to worry about and easily fixed.
The thing is that I had a couple of these small spikes showing on the Temperature graph yesterday and today and I updated to the latest C1 Beta and decided to reload data from a backup on 4th October. The small spikes are no longer there.
This suggests that the small spikes are perhaps coming from within C1.
Gina, could you try reloading from a backup prior to the spikes to see if they disappear?
Next spike I get I will try reloading form a backup again.

[Ned]

If the temperature spike doesn't coincide with a data save by Cumulus, it won't get logged and won't show on a historical review or data reload. So if Cumulus is set to log data at 10 minute intervals, a spike only has a 1 in 10 chance of being recorded.
I've noted spikes on the live screen which have almost always gone next day, as I switch off the PC every night - but the odd one will still get through. For big spikes I think the solution is to apply a software filter, as Cumulus can do. I'm sure spikes originate in the WS and not in Cumulus.

[Gina]

I'm quite certain Cumulus simply cannot produce spikes itself. I run G-Weather - my own adaptation of pywws for Linux - and I get the spikes. I know precisely what the software does with the data from the WS and know it can't produce spikes. Steve has confirmed this before for Cumulus. I do sometimes run Cumulus at the same time and get similar spikes with that.

Yes, it depends on the timing of data logging and the source of the spikes as to whether spikes are actually seen. I run my station logging at 5 minute intervals and the software reads the station logged data every 10 mins (2 samples) and uploads it to the web site. There is also the 48 second transmitting interval too but I don't know if the WS actually does anything with the data samples between the 5 min logging points.

I'm pretty much convinced the spikes are caused by some sort of radio interference being picked up by the cable feeding data from the wind sensors to the transmitter unit. Shielding the cable and earthing certainly seems to reduce the frequency of the spikes. It seems to me that the amplitude of the interference varies considerably and that the measures I'm taking reduce the affect but don't stop it altogether. I may try adding LC filters in all 4 wires from the wind sensors. I have made provision for this but not added the components yet.

I would like to know where this interference is coming from but so far all attempts to detect it separately from the WS have failed.

[Super-T]

If your spikes are coming from the wind sensors then it would probably be from the direction part because the voltage on that line is a varying voltage and more susceptible to interference whereas I imagine the wind speed is just variable pulse rate? Not having looked at the circuit I'm not sure.
I can see that the spikes could be missed because of sampling rates. I will still try a reload the next time just to see what happens.

[Ned]

The weird thing I've noticed with spikes is that they seem to increase in frequency as it gets windier - I don't remember seeing one in calm conditions. It's as if the faster pulses in the cable from the anemometer reed switch somehow provide a gateway for interference. Just guessin'

[gemini06720]

Gina, I know this is going to sound as a far fetched solution, but, how about stopping the operation of both the wind direction and the wind speed sensors - I do not mean by disconnecting the wires or anything like that - but by placing some tape on each of the sensors so that they do not move at all while remaining connected.

If the spikes are generated by some external surrounding source, then the spikes will be displayed on the graphic. If the spikes are really produced by either of the sensors interfering with each others, then there should be no spike while the sensors are tape stopped.

I am just trying to remove some of the probable cause for the spikes...

[Gina]

I don't think it's anything generated by either of the wind sensors as it didn't happen when using the original cable and the wind sensors lower down. I grant you it's a very slight possibility though and worth a try but when it's less windy than today as it means lowering the mast to get at the sensors. I haven't noticed any extra spikes in heavy wind myself. In fact it's a good Force 4 to 5 now and only one small spike in 24 hours. We are forecast a lot of wind in the next few days so we'll see if I get any more spike activity.

It's possible that it might be correlated with wind direction as the resistance of the wind vane unit depends on the direction - a higher resistance would make it more prone to picking up interference. OTOH the anemometer is an open circuit most of the time, only being short circuit briefly one per revolution when the magnet closes the reed switch. I'll go through the records later and see if there's any correlation there. It might be difficult to see with the strange way FO change resistance with direction though.