Cumulus Support

I think I've finally sussed how it works. Firstly it measures a fixed resistor, 2nd it measures the thermistor and thirdly it measures the wind vane resistance. At the measurement time it clamps one end of the appropriate resistance to 3v, with the other end going to the combined junction with the positive end of the timing capacitor. The periods ends when this junction reaches a threshold. At other times the resistance end not on the junction is open-circuit - I can see a small amount of noise here if I increase the sensitivity of the scope. Note - the waveforms are my best attempt at copying what I observed on the scope screen. Using the markers on the scope I was able to measure the pulses to 10% of the scale.
Reference pulse 44ms
Thermistor pulse 43ms
Wind vane resistance and pulse width for 3 different wind directions
- East 1K 4ms
- South 3K9 16ms
- West 120K 445ms
Spaces between measurement pulses were both 8ms.

The test rig, with the long cable, shows small spikes on the temperature plot when the anemometer is turning that go when it isn't. There are no big spikes. The temperature plot is very ragged (up and down irregularly by 0.1C) when anything is plugged in to the wind RJ11 socket but quite smooth with nothing plugged in. I'm now doing some testing with various combinations.

Gina
You can see the effect of no wind on temperature if you look at the graphs and pick a time with no wind. Zoomed temp graph is quite smooth.
It's almost as if the time between sampling of wind and temp are slightly overlapping causing the small ripple.
Maybe the effect is worse, depending on wind direction?

Yes. And yes, I would expect it to depend on wind direction.

Tests today started with wind sensors connected but no wind, then just before 17:00 I disconnection both wind sensors. About 20:30 I reconnected the anemometer leaving both wires to the vane disconnected. With the vane disconnected, the direction pulse lasted exactly 1 second then switched off. There was very little charge/voltage on the capacitor at the end of the pulse indicating that the chip stopped the charge prematurely.

Here's the last 12 hours temperature plot. The reduction in "spikiness" can be seen from just before 17:00 when the vane circuit was disconnected.

Reconnected the "common" wire to the cable and vane circuit - no change. Small spikes when anemometer turning. Put 470nF ceramic capacitor across anemometer connection and the spikes became several times bigger. I think the capacitor must be loading the circuit. This loading effect might explain why the longer cable with it's greater capacitance, causes the small spikes when the anemometer is turning. I might try putting a choke in series.

Been working things out. The chokes I have are 1mH which is pretty small and I think a resistor would be as effective and may even be better - an inductance could cause ringing with the cable capacitance.

I've measured to current to 0v with O/C anemometer switch and it's 34uA (microamps) so a 10K resistor would only drop 0.34v and should be a good logic low state for a 3v system. Anyway, I've put an 8.2K resistor in series with the live wire to the cable and anemometer. I'll leave the test running overnight with the anemometer turning and the vane pointing west (the maximum 120K resistance). I'll also test it with the vane pointing east (minimum 1K resistance).

There goes that idea!! A wind speed of about 9mph and vane westerly until 08:15 then easterly.

Just an occasional spike with easterly vane position, so better, but not really the point - the wind doesn't always come from the east. I'm now trying the RF choke (1mH) just to see if that's any better than a plain wire! Wind vane back to west.

On a slightly different note, I'm planning to set up a pulse recording system using a Velleman USB interface board (K8055). This has two analogue inputs of nominally 0-5v range. They have a 100K input resistance which is a bit low to connect to the station but have a facility to increase the gain of the input op-amp. So eg. if I set the gain to 11 I can use an external 1M resistor (like a 10X probe).

Interestingly it actually does seem better with the choke. Over an hour and no spikes yet which seems promising, though time will tell.

5 hours on and it seems better though not perfect - the odd small spike plus some low level noise.

Since the 1mH choke seemed to help I've added another in series to give 2mH and I'm running an overnight test with the usual air flow rate and vane pointing west (the worst case). I'm attaching the plot for the last 6 hours with the inductance doubled from 21:30.

14:00 on 23/11/2010 connected 1mH RF choke in series with live anemometer wire.
21:30 on 23/11/2010 increased to 2mH
10:00 on 24/11/2010 fan switched off

Here's the last 24 hours. Well, the 1mH inductance seemed to help a little but 2mH doesn't seem any better, maybe marginally worse.
With no wind there are no spikes.
With wind vane part of cable disconnected there are no spikes.

I'm now having a good think about the implications of the above and considering if more tests would add any more info and what to test.

A bit later...
Disconnected wind vane at far end of cable - both wires. Vane connections from transmitter connected to long cable.
Anemometer connected via 1mH choke in live wire, return connected directly - through long cable.
Fan on to give 8 or 9 mph of breeze from 13:15 today.

Last 36 hours :-

Testing suspended for the time being (needed the table). Some more results though, having disconnected the supplied cable from the long extension and set the fan going again.

where did you put the choke,directly between the anemometer and the cable or between the cable and the unit.
that can make a difference.
usually it's better to put it the beginning(anemometer/cable)

hans wrote:where did you put the choke,directly between the anemometer and the cable or between the cable and the unit.
that can make a difference.
usually it's better to put it the beginning(anemometer/cable)

It's between the tx unit and the cable. But thinking about it, yes, you're right - it should be between the anemometer and the cable. I'll try that tomorrow.