One for the technically minded - WMR928NX adds an extra 1mm rain intermittently.

Hardware Weather Station Equipment
1

I’ve been troubleshooting a rain gauge related problem with my relatively new WMR928NX (about 2 months old now) and I’m hoping somebody here might be able to add some ideas to the potential cause. The problem, simply put, is that every now and then the weather station console (and therefore WD) records 2mm of rain when the bucket has only tipped once in my rain collector, but only intermittently.

The problem first became apparent when looking at my rainfall graphs in WD, I noticed on occasions the graph had jumped by 2mm in a single minute. I thought this a little strange, as anybody with a 1mm rain gauge would know, you need quite heavy rain in order to collect 2mm within a single 1 minute period and I knew the rain at the times of these recordings was light to very light. I started to investigate by un-installing my rain gauge (to eliminate environmental factors, wind, etc), bringing it inside and doing some manual testing.

To test, what I did was this:

  1. Reset the daily total on the console to 0mm of rain.
  2. Let the rain gauge send 3-4 data packets (about 3-4 minutes) to confirm there was no buffered data waiting to be reported to the console from the rain gauge transmitter.
  3. Manually tip the bucket once.
  4. Wait for the total rainfall on the console to increment by 1mm
  5. Back to step 3 and repeat.

This would work flawlessly up until around the 7mm mark and then the problem becomes apparent. At or around 7mm, a further single tip of the bucket causes the rainfall to increment by 2mm instead of 1mm, even though the bucket has been physically tipped only once. This meant that the console was now displaying 1mm of rain more than the total number of times I had actually tipped the bucket in the collector.

I say around 7mm as sometimes the jump occurs from 7 to 9mm and other times the jump occurs from 8 to 10mm and sometimes 6-8mm, but in each case the console is advancing 2mm when the bucket has only tipped once, and therefore over-reads the rain by 1mm at that time. This of course becomes a cumulative error over time and by the end of the year my totals would be significantly out.

I also tried doing multiple tips within one update cycle of the transmitter. This had no bearing on the symptoms whatsoever. In otherwords, I tried tipping the bucket 7 times within one minute (OK, that’s really heavy rain) and to no surprise, the console read 8mm when the transmitter sent the next packet. I also tried combinations, like doing 5 tips, wait for update, 1 tip, wait for update, etc, etc. It didn’t matter when the tips occurred in relation to the update cycle of the transmitter, the net results were still the same. At some point between 6 to 8mm, the console increments by 2mm when the bucket has tipped only once.

The problem is very easily reproducible in that I can rarely (one in ten times maybe) get the console to get to 10mm without jumping 2mm on a single tip at some point.

To try and resolve this problem I have tried:

  1. Resetting the console
  2. Resetting the solar transmitter
  3. Replacing the rain gauge
  4. Replacing the solar transmitter (by swapping it with my temp gauge transmitter)

All of these things have had zero impact on the problem, which remains consistently reproducible after around 6-8mm of rain. The only thing I can conclude is that perhaps I have another faulty rain collector unit? Either that, or my console can’t count to 10! :slight_smile: I’ve googled and searched high and low but can’t find anybody else experiencing this problem.

Does anybody know of a way to test what the collector is sending to the transmitter (unfortunately I don’t have a CRO, just a multimeter which doesn’t help), or any way to verify that the rain collector is sending the correct tip count? It would be so much easier if I could narrow down where in the chain the double-up is occurring. It also strikes me as odd that it always occurs around the 6 to 8mm mark!

Any thoughts or ideas would be more than welcome as I’m running out of ideas! Thanks everybody!

Cheers,
Graham

2

Perhaps the tip is in fractions of an inch and when it’s converted back to mm the console had to do a bit of adjustment to correct the rounding errors?

3

I have seen several instances of this on WD graphs from various people using WMR928s.

It happens to me too - not very often and I used to adjust in WD when it did - but now I just leave it and treat it as a quirk of that particular model rain guage! It last happened for me on March 31st.

I noticed though that for me it can happen any time - even after only 1 or 2mm recorded and not 7mm.

4

Sharp as a tack and quick as lightning, as usual Chris! :smiley: I did consider this possibility and did some investigation and unfortunately it looks like a dead end. The unit appears to operate internally in units of 1mm, which is then converted to inches if you choose to view in inches. This causes some minor adjustments if viewing in inches as 1mm isn’t quite 0.04 inches, so once you get to about 0.32 inches of rain, it jumps to 0.35 and then 0.39… This supports the data that the unit is internally metric (despite being American). I have also tested using the imperial display, and the anomaly occurs there as well (in other words it jumped 0.8 inches, the equivalent of 2mm) on a single tip.

The only explanation I can come up with along these lines was if the bucket itself was slightly over a 1mm/0.04 inch tip, in which case the unit could potentially add 1 mm every now and then to compensate. However the specs data does not support this (everything claims the bucket tip is 1mm). I would also have expected it would be more consistent if it was an internal adjustment and not vary as much as it is.

Excellent thought though!!

sunrainhotcoldwindsnow thanks for those observations. I had been comparing my data with a relatively nearby site which also runs a 928 and he didn’t seem to be having the issue, but granted I haven’t studied that many other WMR928 user’s sites. This is disappointing if it is a “known feature”. It’s funny I couldn’t find anyone else reporting it on the net (at least that I could find via google).

I’ve just done some testing up to 56mm. By the time I got to 56mm of rain on the console, I had only tipped the bucket 49 times!! So the cumulative inaccuracy is 7mm by 56mm. Given that this doesn’t account for any inaccuracies in the gauge itself (in terms of bucket tipping action, etc), it’s quite some way off the target. I sure do hope I can track down a solution to this. It just doesn’t seem right to me. :roll:

5

I would guess Brian may be able to add in a check to WD for this station type to only add 1mm rather than 2mm if WD receives that amount for one bucket tip??

6

Good thought! Certainly might be an option if this is an issue for a good quantity of the WMR928 user base.

Of course I would much prefer to address the root cause if possible, but if it turns out we’re stuck with this “bug”, it might be worth my while dropping Brian a line and see if he would be able to accommodate such a function.

7

Do you know how the tip is sensed? Is there a possibility of contact bounce (if magnet + reed switch) or perhaps bucket bounce if it’s optical with the optical path being closed/opened/closed by the bounce in rapid sucession?

8

On the days that has happened does it happen to be windy? I have that happen once in awhile but it seems to only be on windy days. I figure that the wind shakes the rain gage and causes it to trip once in awhile.

9

I would say it’s a magnet+reed switch which is very unreliable or a bad optic circuit, had lots of problems with equipment using reed switches as they double register when the magnet goes by. It can be fine tuned but it takes a lot of time and effort.
Peter

10

Hi Chris,

It’s a magnet/reed switch combo. I’ve tried putting a multimeter across the reed switch for continuity, but it always appears open circuit despite the position of the magnet? Perhaps other things are at play in the circuit though? I suspected switch bounce in the original unit I had, but I would have thought the chances of getting two gauges with an identical problem would be fairly remote. It was also the consistency of when the 2mm increment would occur that threw me (always around the 6-8mm mark after resetting the totals). I would have expected to see much more randomness if it was switch bounce, but I’m certainly not ruling it out! It could very well be the cause!

Any ideas how I can test for this? Unfortunately I only have a relatively slow-response digi-multimeter and I’m still puzzled as to why the switch always appears open circuit to my DMM? :?

11

I rulled this one out fairly early on in the piece. I’m testing this rain gauge indoors and am tipping the bucket manually so I can be sure of the number of tips the bucket has physically had versus what the display was reading. Thanks for the thought though, and it was definitely one of my first suspects!

12

i do know this happened with the WM918 stations
which were wired
and the solution was to have a opto isolater couple switch put in the wiring to the base
but the wmr968 is wireless…but there is still a short wiring to the rain gauge…maybe thats getting interference?

13

:smiley: sorry, just cant but think of you sitting there for hours trying to pour in 56 tips worth of water. I dont know if I should admire you or think you’re sad… :wink:

being serious for a mo, glad you have identified this, as I’m seriously considering getting the wm968 in about a months time, at least I’ll be informed!

Woody

14

It’s possible that the reed switch is normally closed apart from the very brief period when the magnet goes past and opens it. If it’s contact bounce there are circuits that can de-bounce, but you’d probably need to know a fair bit about how the circuit worked to plumb them in properly without affecting the overall operation of the unit.

15

Hi Brian, yes I remember coming across this while I was googling for possible solutions. It looks like a bit of a design flaw with the 918 cabled setup. I’m not sure if the same is applicable to the 928 cabling between the solar TX unit and the sensor. One would hope O.S. would have learnt their lesson but that might be a dangerous assumption! :smiley: I’ve tried testing with the cable wound up, fully unwound and the transmitter in various different positions but it didn’t affect the symptoms at all. The other thing that put me off this theory a little was the fact that on the 918, it caused the rain gauge to read when there was no movement of the bucket at all, whereas this only occurs if the bucket does actually swing. Still a possibility though!

Woody your post made me laugh!! :lol: :lol: I think SAD is definitely the right answer to the above question! :smiley: I must confess though, I’ve got the funnel off the rain gauge so I can tip the bucket by hand… much easier than pouring water through it and makes much less mess on the floor! :slight_smile: :slight_smile:

Yes I have a feeling that is all going to be well beyond me I’m afraid! I have e-mailed the UK Weather Shop anyway, to see if they have any further ideas on it. In my mind, it kind of blows away the accuracy stats that O.S. publish about their rain gauge. I’d love to know how common this problem is. If it is a common issue I would have expected to see more posts about it but there just seems to be no info about this problem out there, making it all the more frustrating to try and pin down. Of course, maybe I’m just unlucky and have managed to score two bad rain collectors in a row. I sure do hope it’s something simple like that!

Thanks everybody for your input so far… I’m always staggered by the amount of knowledge and helpfulness that abounds these forums!!!

16

Q:
does this rain also show on the console?

17

Hi Brian, yes the “extra” rain shows on the console and is also passed to WD.

18

I am having same problem with my WMR928NX and i am wondering if anyone found solution how to get correct mm’s

Have a nice day

Iztok

19

Hi Iztok,

After trying a replacement gauge from the UK Weathershop, and finding the behavior identical, I have come to the conclusion that this is not a fault condition, but rather a calibration to ensure “accurate” rainfall data. The reason I have concluded this is because:

  1. Swapping the transmitters did not affect the symptoms
  2. Swapping the rain gauge did not affect the symptoms
  3. The UK Weather Shop confirmed the same behavior could be replicated on a different console (even a different model of console that uses the same rain gauge)

and the most important one

  1. My rainfall totals are very close (within 1 mm typically) of the monthly totals of a very nearby station (about 2 miles away line-of-site), despite the addition of the mystery 1mm every every so-often.

I have concluded that the rain gauge itself is not quite a 1mm tip, but rather, tips when slightly-over 1mm falls. I believe the mystery addition of 1mm every now and then, is a calculated adjustment made in the console, to compensate for the fact the rain gauge is not quite a 1mm bucket. Why this is, I do not know, and the UK Weathershop never got and answer out of Oregon Scientific. One thing I am sure of is that the behavior is “as designed” and I wouldn’t worry about it. I think you will find you will get accurate rainfall totals, despite the extra 1mm being added on every now and then.

I hope that helps!

Cheers,
Graham

20

I noticed an identical result over a 12 month period Graham.
After much changing about, I let it get on with it and it seemed to ‘level out’ each month.
Making a modified funnel with a more sensitive output speeded things up, but there was always the underlying increase which I put down to not having an exact value (single digit) per tip. The offset was 0.3798