Order Tray | Contact Us | Home | SIG Lists

[wxsig] Wind Speed Spike

Will Beals will at beals5.com
Sun Jul 9 13:55:33 UTC 2006


At the bottom of the "project status" page on the weather station website is
the explanation of the diagnostic data which with a little sleuthing points
out not only what I suspect your problem is, but a boundary case I thought I
had taken care of but not completely.

The "01" (regular diagnostic messages) show that there were no communication
errors, the fourth pair of digits after the 01 is the rain gauge error
counter. It was 0x35 both before and after the event.

The "03" message is inserted whenver there is a perceived wind gust of
greater than 25MPH.  I had been tracking some errors like yours for quite a
while that turned out to be failing reed switches (I have one now too, so
occasionally get wind gusts of 99MPH).  But I digress.  

The wind speed is measured by looking at a counter that counts anemometer
ticks.  It is an incrementing counter.  To get wind speed I look at the
counter value every 5 seconds, get the difference and multiply by 0.227806
miles per hour per tick (determined emperically).

The "03" message has four fields.  The second field, 0xffff is the current
count, the third field, 0x53E2 is the previous count and the first field
0xA1CD is the the difference.  That's 41,421 ticks or 9,336MPH!  What
happened (as is probably obvious by now) is that the value of 0xffff is bad.
What happened is somehow, your 1-wire bus lost power for long enough for the
wind speed counter to go through an internal reset and the value after that
reset is 0xffff.  The 1-wire bus is designed to pulse low in order to
communicate, and during those low pulses, the self-powered devices on the
bus are designed to have enough power stored in internal capacitance to say
alive for those pulses.  Something happened on your bus to have the bus go
low for more than it should have.

I knew this was possible and have a check in the T238 to see if the current
count is less than the previous count.  What I failed to realize was that
the reset value of that counter is 0xffff, not 0x0000.  A value of 0xffff is
greater than 0x5e32, so my checker didn't fire and you got 99MPH instead of
an incrementing error counter.  I will add this to my bug list.  I never saw
it before as during normal testing at least one tick had come in to roll the
counter over to 0x0000 by the time I checked.

As to root cause, there could be several things from a simple intermittent
short somewhere to maybe a sensor that isn't happy and dragging down the
bus.  Your bus voltage looks pretty healthy from the other diagnostic data,
so I do not expect the latter.  For the folk with older anemometers like me,
the reed switch is an issue, but it manifests itself as a large (hundreds)
count increase, not a counter reset.

I'll answer your second question in a separate email just to keep things


-----Original Message-----
From: wxsig-bounces at lists.tapr.org [mailto:wxsig-bounces at lists.tapr.org] On
Behalf Of Ken Brown
Sent: Saturday, July 08, 2006 5:51 PM
To: wxsig at lists.tapr.org
Subject: [wxsig] Wind Speed Spike


Several times in the last couple of weeks I have had a wind speed spike of
199g099.  This is since the nearby lightning event that caused several
other, seemingly resettable problems in the weather station and other
equipment.  That was also when the MAX232 chip failed and the barometer
became erratic.  The barometer is now plugged into a test T238 that has no
outside sensors and seems to be fine.  The status messages for a couple of
hours before, the affected line and several after are included below.  The
wind sensor is an original Dallas Semi although purchased from AAG.

normal data at 1404z 01080B350D55FED1
normal data at 1504z 01080B350D0CFEBF
(line wrap) P000r000h61e1w 03AC1DFFFF53E23A normal data at 1604z
01080C350D16FCC0 normal data at 1704z 01080D350D19FDC2

I had a similar problem a year or more ago and I changed the T238, wind
sensor, and wiring.  I don't know what may have cleared it at that time.

On another subject, I notice on the schematic that you use three pins on the
CPU to generate the one wire bus.  I assume this is to improve fan-out.  Are
all three pins the same or do you transmit on two and receive on one?  I ask
because I thought I might try to separate the one wire lines going to my
wind, temp/rh/rain, and barometer, by separating the transmit and receive
lines into some external devices to create three one wire bus lines.

Ken Brown

wxsig mailing list
wxsig at lists.tapr.org

More information about the wxsig mailing list