07/26-29/1999Neal finished loading a program and connecting a rain gauge after Fred had mounted the mast and radiometer on the 26-28th. Data runs through an optical isolator and via an RS-232 cable to COM2 of the PC called pressure in room 108 but produces errors similar and larger than those at McClure. Debugging to be done. Precise direction orientation is not yet known.
07/30/1999 External 7.1 Volts added to pin 4 at exactly 19 Z on July 30.
08/06/1999 Changed sampling to 5 seconds and baud to 300 for transmission test 990806 19:15Z.
08/10/1999 No transmission errors for more than 3 days. Changed sampling to 1 second 990810 16:16Z.
08/11/1999 No transmission errors prior to fiddling this AM. Changed program to measure old Eppley and the black tipping bucket rain gauge between 14:15 and 16 Z . Also, changed the angle of the direction indicator to be very close to proper North-South orientation approx. 23Z.
09/18/1999 UW Eppley decrease during clear skies. JET cleaned sea gull residue from both Eppleys
10/20/1999 UW Eppley decrease during clear skies. Fred cleaned sea gull residue from new Eppley
10/22/1999 UW Eppley new has condensing inside the dome, old one's wire shield totally frayed, Fred and JET examining between 21-21:30 Z
01/18/2000 Added a rain gauge to the Campbell on last pulse channel. Appears to read low compared to Internet channel. See Mail 1/21.
01/26/2000 Jumpered pressure from 108 "new" sensor up to rooftop 23X, control port 5. In process broke fountain yellow line but jury rigged it for now (needs to be stripped, de-tensioned and reinserted). Using P15 on 23X in table 2 for pressure measurement shows some success. We are sending corrected station pressure to Web plot utility, currently that is pressure plus 0.01 mb, per calibration sheet. --ncj
04/10/2000 Approx. 13:20 removed "new" sensor from rooftop 23X, replaced with jumper from old sensor to control port 5. Put new sensor on portable 23X for class demo. --ncj
04/13/2000 Approx. 11:56 reconnected "new" sensor to rooftop 23X, via 21x in 108 --ncj
06/10/2002 Verified sample rate was 5 seconds, housing fan was functional, cleaned radiometer domes --nj/fw
07/27/2002 Moved pressure sensors to new shelf along North wall which was set by Steve to same height as old support. --nj
12/10/2004 Added a new P2 instruction. Copied the one which measures the temperature of the RH sensor except that in the new one I overwrite output location 7. This is because the thermocouple temp. being written there earlier in the program appears to have broken and was giving wild data. Coincidentally the time of this change was very near the UTC date rollover (15:59 local). I also unplugged the box fan which was running noisily away. When the E-type thermocouple is replaced, the extra P2 can just be erased.
10/31/2005 Took out the exra P2 instruction and replaced the bad thermocouple. RH is now mostly bad. Tried the Rotronics to no avail but there is a new extra P2 added to measure DIFF channel 10 (I think it's the 24th instruction in the program). So, RH and dewpoint are now set to missing in the new perl script on the server. We believe a new sensor may be necessary though taking the mast down and taking a look at the sensors first is probably a worthwhile step.
08/07/2006 Sampling interval is 10 seconds. Verified also that the first precip. output channel, corresponding to uwpcpa command, is the square gray tipper. Second precip. output channel is the golden cylindrical tipper. Black tipper is not connected, though some remnants are still in the Campbell program. 5 hand tips on the uwpcpa were made on this completely cloudless day but only 2 registered!
04/05/2007 Temporarily added a P2 measurement for Li-Cor pyr. from Bertschi. It's using a 60 ohm shunt resistor and is connected to DIFF channel 12, writing to ILoc. 22. This instruction was hand entered on the 23X panel and is the first instruction in Table 1. Originally it was tacked on as the last element of the output string. Later it seemed an ingestor thought the last element was always going to be voltage so I swapped the Li-Cor output to be second to last. The cable is short and positioning is poor so our resulting "good window" of sun visibility is roughly 10AM-3PM.
04/09/2007 Swapped in different resistor, added extension and reconnected the Bertschi radiometer. Added it back on approx. 16:35
04/xx/2007 Month of April 2007 seems as if rain gauge buckets were faulty. One ridiculously high, the other ridiculously low. I'm going to stick with ridiculously low, which is believed to be the Texas Elec. and valid from 2007-01 through 2008-11-19.
05/04/2007 Checked for loose wires on Eppley radiometer. Wires still tied to the rad'n mast with elec. tape. Noticed that LiCor had been tipped off of its lead brick. It was at a 45 degree angle pointed roughly westward. I put it back atop the lead brick.
05/23/2007 Removed LiCor measurement but forgot to take out the P71. Removed the LiCor itself. Plugged the fan in.
08/20/2007 Following up user notification of missing minutes in data, I found a few odd things. One is broken outlet covers on rooftop. Secondly, the execution interval on Campbell was 25 seconds, not allowed. Output interval should be integer multiple of exec. interval. I unplugged the fan. Third item of note was that the display read "Busy with COM" when I first went up. Reset sample rate to 10-seconds and problem seems to have been fixed.
11/19/2008 Applied some sealant to outlet cover. Wired up the white rain gauge in place of the Texas Electronics gauge which appeared to be non-functional some time ago. Poured some water down the gauge about 13:32.
01/07/2009 Repaired the tipping bucket with the severed fulcrum (thanks Steve!) and wired it up (Pulse Channel 2) with a long wire, finished 13:45 or so. No tips were recorded which means a bad connection somewhere. Between 15:00 and 15:50 or so began working on stop-gap fix: jerked the wires of the old hand-made rain gauge and substituted them for the long wire and wired them into pulse channel 2 which reports to data field 6, uwpcpa. This is a larger funnel so increments are really less than 0.01-inches per tip.
01/08/2009 Increased the sample rate to 6 seconds and eliminated the single tap per interval requirement for the big funnel rain gauge. Ratio overnight between big and little was 795/116 tips. Between 14:00 and 15:20 moderate rain with incorrect conversion factor in 23X program. Fixed 15:20 local, thanks Mark!
01/13/2009 Modified ingestor to use col. 13 from Campbell which is measuring the white Weather Measure Co. rain gauge with the smaller funnel. It has verified well with hand measurement. Also, the larger funnel version (col. 6) would require other downstream ingestor changes.
02/12/2009 Took a look at last month of precip. The larger funnel is lagging the hand verified bucket on channel 13. I'm adjusting the per tip calibration from .00146 to .0017132 in accord with the approximate area ratio between the two buckets.
10/22/2009 I noticed one of the output channels was reading -99999 today which could indicate an offscale input voltage and since that can cause the Campbell problems, I investigated. The channel was one that was labelled OLD_radn in the Campbell program. This is the radiometer with the loose signal wires so whatever voltage the Campbell was sensing, it had no umph behind it. Still I disconnected those wires from the Campbell (16:00 or so). There was no effect in the raw file. I turned off the box cooling fan. There are a lot of spare/leftover output channels. Someday!!
12/16-29/2010 Plugged the new Apogee sensor (12/16 18:40'ish UTC) into the Campbell using the one that was labelled OLD_radn in the Campbell program. I turned off the box cooling fan. I didn't swap the ingestor to use the new channed right away. I wanted to run them in parallel for a bit first. David swapped new ingestor in approx. 12/30 00:08 UTC.
01/13/2010 Moved .01" tipping bucket to see if siting was the problem at all. It was often lagging the hi-res big bucket tipper by quite a bit, probably more than can be accounted for by the blockage factor but we'll see.
01/18/2010 17:00 Replaced .01" tipping bucket to see if older bucket would track the hi-res tipper any better than the spiffy white .01 tipper. I poured about 5 tips of water down the spout and only got one registered tip but the Campbell software should have a 1-tip per sample interval limiter, so, it may be the case that these tips were rapid enough to have occured within one sample interval and therefore have been converted to a single tip. Rain later this week should inform.
10/18/2014 15:00 Replaced .01" tipping bucket with white tipper. It is on Pulse Channel 3. Fixed software on Campbell as well since it may have been inadvertently altered on Friday the 17th. 2-test tips on Saturday. Sunday night into Monday tips are all the real thing.
10/27/2014 15:00 Started a 1000ml. IV drip during sunny break on this Monday before the storm to get an idea what the calibration of the white tipper really is...
12/09/2014 14:00 Moved the petit file bucket (white) north of the antenna dish after making 7 tips that were all properly recorded. Connected the previous little tipper with the gold funnel to channel 6 of the Campbell output (formerly the high-res. tipper) and made it count only tips rather than inches.
12/11/2014 13:04 Moved the precipitation channel being piped to the petit files from output location 13 to location 6 (the gold topped tipper, yes AGAIN!). I did this with a perl script change, not through hardware changes.
01/16/2015 Installed a newer gauge on channel 13 output to Campbell.
01/21/2015 16:00 After comparing channels 6 and 13 precip over the weekend, swapped the perl script ingesting data to use the new channel 13 gauge.
11/27/2017 12:38 We believe the bearing seized on the anemometer sending the readings to essentially 0 since around noon local time on Wed. Nov 22. This morning David and I went up and jumpered the signal and ground wires of Cliff's sonic anemometer from the fast sampling Campbell 21x to the 23x that measures all the other rooftop instruments (at 10 sec interval). We duplicated the P01 commands used on the 21x on the 23x except for the single-ended input locations and output storage locations. This required changing the P69 wind vector output instruction on the 23x.
After checking petit file I noticed the gust was bad and re-configured the P73 (maximum output instruction) to use the new location for wind speed, but in so doing I inadvertently overwrote the sample interval with 880 seconds... rejiggered that back to 10 seconds circa 12:38 local and things have seemed reasonable since.
03/03/2020 14:38 Since the wind speed seemed low in recent high wind events, David and I undertook the mission of seeing if we could find out why. The keyboard and communications with the rooftop 23X are very problematic. This made discovering and correcting the winds difficult but after breaking the program on Monday afternoon, the keyboard seemed to start working somewhat better on Tuesday. The jumpered winds from the sonic were being stored with the calibration on the fast-sampling 21X which it turns out was producing meters/second rather knots! Neal will backfix the petit files.

There are still some superfluous measurements in the 23x program and it should be trimmed down to match with the still working instrumentation. Good summertime task. It's not clear why the communications are undependable when connected to the logger directly and yet the continuity of data seems good. Use of the keyboard must be done very slowly and deliberately. While deleting instructions, the logger would often jump into a very strange mode. After hitting #D to delete an instruction, it would often display instructions, that weren't even in the program, in numerical order. That is, after a #D it would display P01. If you hit the back key (B), it would display P135, back again P134, and so on. If instead of B, you hit A for advance, it would advance from P1 to P2... as if it was presenting a menu of instruction options. Very weird. Each time it did that, I hit *1 again and 21A to advance to instruction 21, then began deleting more of the old instructions that were serving mainly to feed voltages that were designed to limit Drumheller fountain heights. Also deleted from this batch of commands was the ones that were limiting the tips to 1 per iteration. I do not know if those were really necessary or not. If the currently recording gauge is the mercury switch gauge, probably not.