Temperature data spikes revisited

[Gina]

Now both the vane connections show 3v. The anemometer shows 0v and the rectangular wave as the reed switch opens and closes as the cups rotate. That short was the cause of the mess on the temperature plot!

Which confirms that interference on the wind vane wires can affect the temperature plot.

Yes, indeed it does.

I'm wondering if the anemometer square wave could be causing interference in the wind vane circuit. If none of the reed switches is closed this would be a high impedance circuit that might pick up interference quite easily. I think you've mentioned that some of your reed switches don't close Gina? It might also explain why some (or most?) of us don't have a spike problem.

Switching the scope to AC and increasing the sensitivity I can see small pulses of about 0.1v quite often but occasionally there is a spike which shoots right off the scale (0.1v/div) so that is several volts. These aren't pulses of RF but steep fronted single pulses on the display.

The small pulses may result from the anemometer square wave - but now and again, several tens of seconds apart there's a much bigger pulse. I've now changed v/div and the pulses are a couple of volts high.

I'll check the wind vane resistance but I think there is always at least one vane reed switch closed. The newer unit shows preference for the 8 primary points like other people's.

[Gina]

Just swapped to the board connecting anemometer only so I could measure the vane circuit resistance with digital multimeter. Over a period of a couple of minutes it showed resistance varying from about 15K to 120K - it never went O/C in that period.

Having brought the test rig and transmitter indoors, I must say, it makes testing much simpler Should have done it before but I didn't want to take the station off-line for too long. Now my patience with trying to keep it on-line has run out I've changed my mind!

[Gina]

I've connected a 470nF capacitor across the wind vane connections and now have both wind sensors connected. Over an hour so far without a spike.

[Gina]

It didn't last!

6hrs-Temp-only.png

[Gina]

More testing with the scope this morning... Anemometer looks normal but wind vane shows occasional pulses of a volt or two negative going from the 3v normal level on BOTH lines. I can't tell if they are both at the same time as it's only a "single beam" scope. Measuring with a digital voltmeter shows only a few millivolts drop across the vane resistance. With such a small current flowing through the resistors it's no wonder it's sensitive to interference. I'm wondering though if I'm missing something here. Are the measurements taken only occasionally by applying a pulse to the vane unit? I checked if the pulse timing matched the transmission, but it doesn't.

I think it needs more sophisticated test equipment. I do have a USB interface unit which has 2 analogue inputs and 5 digital inputs. I should be able to run this with suitable software to produce a low frequency scope with two inputs. It has 8bit resolution.

[nking]

Hi Gina,

I don't know if you've answered this before (sorry if you have) but have you tried using your new anemometer/wind vane and does this produce the same results?

[mcrossley]

I'm wondering though if I'm missing something here. Are the measurements taken only occasionally by applying a pulse to the vane unit? I checked if the pulse timing matched the transmission, but it doesn't.

That would seem like a good battery saving scheme. I have a spare transmitter here, but it is faulty (doesn't transmit after it locks on to the time signal), I could try and see if that exhibits the same behaviour (it would be open cct though as I don't have any spare sensors).

[Charlie]

Just a thought, if you put a series resistor in a leg, you could scope across it to see if the current is continuous or not. You might also swap the reference to the positive rail from the negative one to confirm it gets switched on by pulling low instead of high (using a transistor switch to connect to gnd instead of +3V). Maybe those decoupling caps to ground should have been decoupling caps to +3...

[Gina]

I don't know if you've answered this before (sorry if you have) but have you tried using your new anemometer/wind vane and does this produce the same results?

Yes, this is the new kit. The wind vane produces results much like other people's whereas my first one favoured a couple of the in-between positions rather than the 8 primary points.

I'm wondering though if I'm missing something here. Are the measurements taken only occasionally by applying a pulse to the vane unit? I checked if the pulse timing matched the transmission, but it doesn't.

That would seem like a good battery saving scheme. I have a spare transmitter here, but it is faulty (doesn't transmit after it locks on to the time signal), I could try and see if that exhibits the same behaviour (it would be open cct though as I don't have any spare sensors).

Thanks You could just use a resistor in place or the wind vane unit, something in the 10K-100K range I should think.

Just a thought, if you put a series resistor in a leg, you could scope across it to see if the current is continuous or not. You might also swap the reference to the positive rail from the negative one to confirm it gets switched on by pulling low instead of high (using a transistor switch to connect to gnd instead of +3V). Maybe those decoupling caps to ground should have been decoupling caps to +3...

A series resistor is a good idea. So is using the +3v side as reference. I've just done that and, yes, there are negative going pulses of about 1.5v. The shape looks like they charge up a capacitor, with a sharp leading edge and an exponential tail chopped off (if you see what I mean). I saw groups of two short and a long pulse. I'll see if I can adjust the scope triggering to capture a pulse train and then take a photo.

At the same time as checking for pulses with the scope I used the scope earth croc clip to connect the cable screen to a battery terminal rather than real earth. I seem to have pretty much stopped the spikes. (Now just watch them come back again!!).

6hrs-Temp-only-f.png

[Gina]

Isn't it amazing - no spikes for several hours but as soon as I publish this fact another of the little blighters comes along!

6hrs-Temp-only.png

[Gina]

More scope results... I have the scope referenced to the +ve rail and 1v/div. There are negative going pulses on both vane connections with one showing overshoot with a longer time constant than the decay of the pulse. The pulses are 3v in amplitude and decay to half that before turning off. About 4 or 5 seconds after these pulses, the unit shows transmission (red LED).

I currently have a 470nF capacitor across the vane connections which I shall disconnect and look at the pulses again. The pulse widths are about 40 and 60ms but vary (depending on the vane position I guess).

Removing the 470nF capacitor had no effect on the pulses as seen on the scope. Maybe not on the temperature plot spikes either, though the latest spike is a bit bigger.

4hrs-Temp-only1.png

[Gina]

Conclusions so far :-

1. I have seen no trace of RF interference using an oscilloscope. But the 3v pulses arising from the microcircuit make detection very difficult so can't actually rule it out. With the sensors cable unplugged from the transmitter, noise IS visible but to be expected feeding a very high impedance.
2. The anemometer switches 3v on and off.
3. The wind vane resistors are in a timer circuit (similar to the ubiquitous 555) and both connections are "live".
4. The wind direction is measured a few seconds before transmission.

[Gina]

It looks like the 470nF across the vane circuit cut down the spikes a little bit - they're worse now. May try capacitors to +rail instead of across the sensor.

4hrs-Temp-only.png

[Gina]

470nF from each vane sensor connection to battery +ve. I can't say there's much improvement though.

4hrs-Temp-only1.png

[Gina]

Disconnected one of the capacitors about 22:40 yesterday and the base line "rubbish" has gone but not the large spikes. An improvement though. Seems the capacitor on that line to battery made things worse.

12hrs-Temp-only.png

4hrs-Temp-only2.png