Clifton Laboratories 7236 Clifton Road  Clifton VA 20124 tel: (703) 830 0368 fax: (703) 830 0711

E-mail: Jack.Smith@cliftonlaboratories.com

I've had  the Weather Envoy remote wireless transceiver running for more than a day receiving signals from the Integrated Sensor Suite via the console repeater. (See 28 June 2008 for details on what this means.) The plot below shows the percentage of transmitted packets received. I made the change to repeater mode around 7 AM yesterday, which is roughly the middle of the data set in the three day plot below. The result is highly consistent reception, with 80% of packets received and relatively few and shallow dropouts.
 

This afternoon, while I have the use of a WJ 8617B receiver, I dialed up 916.5 MHz, 100 KHz bandwidth and sure enough, there's a data burst every couple of seconds. With the 8617B connected to my discone antenna, there's plenty of received signal strength, perhaps 30 to 40 db over the noise level. But, that's (a) with an antenna that is relatively close to the weather station  transmitter and (b) with a reasonably sensitive receiver that cost the US government as much as a new Cadillac when purchased 20 years ago.

I then tuned the Advantest R3463 spectrum analyzer to 916.5 MHz, added a broadband preamplifier and connected it to the discone antenna. By operating the spectrum analyzer in zero span mode, we get a plot of signal amplitude versus time.  Unfortunately, the choice of sweep speeds is limited, and the fastest speed is 5 ms/division or 50 ms for the entire 10 division sweep.

A typical data burst is shown below. I didn't want to drag my TDS430 digital oscilloscope across the basement, so we'll use just the R3463 images for the analysis.

First, the visible databurst is about 11.4 ms long. (Marker 1 is at the trailing edge of  the burst.) I don't know how much of the burst's leading edge is lost, but probably not too much. (This can be looked at with the  TDS430 digital scope if I have some idle time.) The modulation mode appears to be OOK or ASK or "on/off keying" or "amplitude shift keying" which are fancy  terms for keying the 916.5 MHz carrier on/off to transmit data.

The shortest data elements look to be on the order of a few hundred microseconds (really need the TDS430 connected to see this type of detail). Let's say 250 microseconds as a rough estimate. This means 4 bits per millisecond or 46 bits in 11.4 ms. If I had to guess, my money would be on 48 total data bits and a burst length of 12 ms if the 250 us number is correct. There could well be a synchronization preamble lead-in as well that is lost to the left of  the trace.

First, assuming the second burst is of equal length to the first burst, it seems that the 12 ms figure is reasonable. Both bursts also start with a similar length high followed by a short low. This might well be a synchronization lead-in. (Something similar can be found in IR remote controller transmissions.) Second, the data transmitted in the two bursts is not identical. There's also a "channel ID" somewhere in the data burst as well as a checksum to help weed out corrupted data.

The second burst could be the console's repeater signal. There's a slight amplitude difference between the two bursts, with the second burst being a couple dB weaker than the first burst.  This supports the two signals being from different sources.

If one wanted a project to keep their idle hands busy for a while, the data format looks to be slow enough and simple enough to be decoded with an 18F PIC with time to spare. However, the current production Vantage Pro II uses frequency hopping transmission, which means that you will also have to build a frequency agile receiver and program it to track the hopping transmission. Frequency hopping is mostly an anti-interference measure as allows several similar weather stations to operate without mutually interfering, or at least statistically spreading the interference. It has a perhaps not so accidental side benefit of making it more difficult to build a compatible receiver as one must know the hopping sequence. In this case—where it is unlikely that the hopping sequence is intended to provide military grade security—one could likely deduce the hopping sequence with a series of measurements.

Since the older Vantage Pro equipment I have uses a single RF channel, sharing amongst multiple weather stations uses a different approach. (The ISS transmitter has a "channel set" DIP switch, whilst the console and Weather Envoy set channels via software. These are logical channels, as all the gear operates on a single frequency, 916.5 MHz.) With a short burst transmisison, 12 ms or so every 2.3 seconds, there's a good chance that transmissions from two ISS units will not overlap. However, Davis has added an additional anti-interference measure. Each logical channel has a slightly different frame rate, i.e., the data bursts are transmitted at slightly different intervals. Logical Channel 1 bursts are, say, 2.3 seconds apart, whilst logical channel 2's bursts are 2.4 seconds apart. (not the actual values). This avoids the statistical chance that if two Vantage Pros are nearby, and both have the same burst rate, they could be in perfect synchronization so that every data burst from one transmitter interfered with the second. Given component tolerance and temperature drift, that's not likely, but it does mean that there could be longish periods where the two were close enough in synchronization to have the bursts overlap, at least in part. If the two transmitters, in contrast, start in synchronization but with intentionally slightly different burst periods, they will drift in and out of mutual interference. If the burst periods are selected with sufficient care, this may provide quite useful periods of interference free operation compared with the "all transmitters use the same burst rate" approach.
 

Rick, VE7TK, has written a tutorial on using a Z10000-U buffer amplifier to extract a wideband 9 MHz IF sample from a Ten-Tec Orion (model 565), available from Rick's web site at http://www3.telus.net/ve7tk/9MHz_IF.pdf (the document contents are still subject to change, Rick says). He notes that the suggestions likely apply to the Orion II as well, but can't confirm that.

I've also added a section to the Z10000 page linking to the four radio-specific installation notes. These are, in addition to Rick's Orion page:

28 June 2008

The last few days have seen me working on linking data from my Davis Instrument's Vantage Pro weather station to the Internet. When visiting Mike, W4XN, earlier this month, he demonstrated  Davis's WeatherLink software he uses with his Vantage Pro II. Both my Vantage Pro and his Vantage Pro II are wireless stations, with an outdoor integrated sensor suite (temperature, humidity, rain, wind speed and wind direction) linked back to a display console via a low power 900 MHz packet radio link. (Davis employs an interesting power mechanism with the ISS—a solar cell array charges a "super capacitor" with a lithium  battery for backup during extended cloudy periods. During periods of reasonably daytime sunshine, the lithium battery isn't used even at night.) The photo is of a Vantage Pro II, but there's little difference in the console appearance between the VP and the VP II, save for the legend.

In order to extract data from the ISS, Davis sells a "data logger" that plugs into a port on the rear of the display console. The data logger saves all data readings for several days and connects to (different models) either a computer's serial port or a USB port. In both Mike's case and mine, our display consoles are not near our computer rooms. Fortunately, Davis has a solution—the "Weather Envoy," a stand-alone transceiver that communicates with the ISS over the radio interface. The Weather Envoy has no display, containing only the radio transceiver, local temperature and humidity sensors and a connector for the data logger.

Mike gave me the Weather Envoy used with his Vantage Pro before he upgraded to the incompatible Vantage Pro II weather station.  I ordered the separate data logger module and PC control software, WeatherLink 5.8.0, from Davis and installed it Thursday.

Although other software compatible with Davis Vantage Pro equipment is available, it all requires the data logger module, which Davis sells only as a package with the Weather Link software.

Installing WeatherLInk starts with a multi-screen "Setup Wizard" that is generally well thought through with useful help screens. When completed and connected to the ISS, the main screen presents the indoor and outdoor data, as shown below.
 

What I found was that the signal, as measured by the ISS Reception parameter, which measures percent of received packets, 0...100%, showed long periods of good reception but with extended dropouts. Moving the Weather Envoy to different locations made changes, but I could find no position that would consistently ensure reliable communications. The data plot below shows 24 hours of signal level with four periods of signal failure.
 

The figure below shows the effect of switching to repeater reception. I made the change at 7 AM just after the deep but narrow dropout. The reduction at 9 AM is a result of experimenting with a different antenna position.

Although the direct reception mode shows some periods of 100% packet reception, it also shows periods of zero reception, with major swings. In contrast, repeater reception fluctuates a limited amount, with 80% reception the norm. Missing 20% of packets is not a serious matter, as the data is repeated to account for this eventuality.
 

In addition, I've used WeatherLink's web server features to add a weather page to this site. The graphics are a bit crude, but serviceable. I've also noted slow updating of  the scrolling display at the top of the page, which I believe is the browser not refreshing its display and instead showing a cached out-of-date version of the scrolling display data. You can see the page by clicking here, or by using the navigation bar at the top left of this page.

In experimenting with the software and making changes, some of the weather data became corrupted. The peak wind speed was not 175 MPH in June, regardless of what the weather station data says!

I've made a couple of small changes in the Watkins Johnson 8617B page WJ-8617B Receiver Impressions to correct the source for a couple of stories and to add comments based upon using the NRT squelch system more.
 

Mike, W4XN, loaned me his Watkins-Johnson WJ-8617B HF/VHF/UHF receiver last week and asked me to use it and give him my opinion of it. I've decided to memorialize my impressions of  the WJ-8617B  receiver at this site as well. You may read it by clicking here or via the navigation bar at the top left margin of this page.

If you're not familiar with WJ products, they have been widely used by the US military and by three letter agencies. This particular 8617B tunes from 2 MHz to 1100 MHz and receives AM, FM, CW, SSB and pulse modulation, with a wide selection of IF bandwidths. It includes a panadapter module.

Oh, yes, I should mention that when new, this receiver cost about as much a new Cadillac automobile. WJ's 1992 list price (without options) shows the receiver starting at $24K, and Mike's receiver has at least $12K in "options."

As anyone who has read even small pieces of this site understands, I like numbers, measurements and real data. Lord Kelvin said, as restated by Prof. Michael Sherba, my transmission line professor in electrical engineering at Wayne State University many years ago, "if you can't measure it, it's not worth a damn." What Lord Kelvin actually said was:

When you can measure what you are speaking about, and express it in numbers, you know something about it; but when you cannot measure it, when you cannot express it in numbers, your knowledge is of a meager and unsatisfactory kind: it may be the beginning of knowledge, but you have scarcely, in your thoughts, advanced to the state of science.

Along these lines, I've read today an on-line book Sustainable Energy – without the hot air, written by David J.C. MacKay of the Department of Physics University of Cambridge. The book may be downloaded or read on line without charge at http://www.withouthotair.com/

What makes this book worth reading for me is the author's attempt to quantify current energy consumption and possible long-term sustainable replacements with hard numbers, or at least as hard as the numbers can be given our current information. I don't agree with the author's viewpoint when he veers away from the physics of his topic and enters the political realm, but he has well separated opinion from fact. The book is still in progress so you'll find a few gaps and notes to "fill this section in" here and there.

My wife has spent a a couple weeks in the Greek islands studying painting and whilst she was away, I visited my friend Mike, from Charlottesville VA, W4XN Monday and Tuesday this week. My wife returned last night after what turned into a long flight from JFK to Dulles so things will return to normal around Clifton Laboratories World Headquarters.

One thing I did while visiting with W4XN was help him install a Z10000-U buffer amplifier into his FT-1000MP transceiver, to obtain a sample of the 70.455 MHz first IF ahead of the crystal filter. The photo on the right shows Mike working on the FT-1000MP.

It took both of us close to a full day to install the Z10000-U buffer amplifier. Electrically, the installation was simple enough. We found a point to connect the amplifier in the schematic. Mechanically, however, it was a different story. First, in order to reach the high IF PCB, the FT-1000MP must be disassembled, including removing the power amplifier module and the cooling fan module. Then, you must devise a mounting arrangement for the Z10000-U board. We spent a couple of hours debating various mounting approaches, as all the normal methods, such as using a threaded standoff, would have required much more disassembly than either of us cared to do. Ultimately, we attached it to the power supply enclosure with tape. Not the most elegant mount, but it avoided further disassembly.

The Z10000-U buffer amplifier's purpose is to provide an IF sample to be used with a Z90 digital spectrum analyzer. 70.455 MHz is considerably out of the Z90's normal operating range and we found  that although the Z90 has adequate sensitivity, it has some undesirable spurious responses. I'm working on a solution for that.

Because there may be wider interest in obtaining an IF sample from the FT-1000MP, I've added a new page discussing the installation. Read the page by clicking here, or via the navigation bar at the top right.

A long time friend, K8AQC, asked about the AC current and power characteristics of the compact fluorescent lamps he uses.

I've made measurements of the AC current and power and also looked at the broadband and line spectrum noise coupled back into the AC line by a Sylvania 100 watt (equivalent) 23 watt actual screw-in CCFL. The data is presented at a new Compact Fluorescent Lamp page.
 

I've tried several ways to store multi-meter cables, ranging from plastic Zip-lock bags to a small tool chest. A couple days ago, I stopped at the local Office Depot to purchase supplies and ran across an inexpensive ($2.99) nylon pencil holder that makes a great test lead holder.

The pencil holders are available in several styles and colors, and the one pictured below best fits my needs. The front is clear plastic so you can see which leads are in the bag. Additional protection is provided by the black nylon mesh. The pouch has two compartments, each with a separate zipper. I've placed the test leads in the main compartment.

The three grommet protected holes are spaced for a 3-ring binder, but could also be used to hang the bag from a hook, or to attach the pouch to the multi-meter.

Below is 24 hours of line frequency data, taken with a Racal 1992 frequency counter. I've also plotted a 15 minute moving average to better illustrate changes and suppress the sample-to-sample noise. The plot has about 35,000 data points, taken 2.33 seconds apart.

The moving average data seems to show peaks and valleys with about a 1 hour period. It also shows a major drop in line frequency at 18 hours into the collection, corresponding to 7 AM EDT.

After a bit of  tinkering with the hardware and software, I have an 18F4620 PIC reading the AC line frequency with reasonable accuracy.

The image below is the line frequency taken over a period of about 1000 seconds with a Racal 1992 frequency counter. As you can see, it isn't exactly 60 Hz at any given instant, although a long term average shows it quite close to 60.000 Hz, and if it wasn't electric clocks would not keep accurate time.

During a 100 minute sample period, I found the Racal 1992 data showed:

Projected over 24 hours, this slight deviation of 0.0138 Hz from 60.0000 Hz would cause an electric clock to run 19.7 seconds slow. That's much worse than experience shows, so we can expect the local power company to offset this slow period with a slightly higher frequency later.
 

All in all, the agreement is acceptable.

I've been working on my power line frequency and voltage project intensively over the last several days.

The photo shows the PIC controller board (small board at right center), a board with the master power supply and squarer circuit (the board made from PCB material) and a A/D converter with optical isolation from the PIC (the large plug-in board). 

This represents about 75% of the total required circuitry. Missing is the anti-alias filter, which is reasonably straightforward but needs to be breadboarded.

At this stage of development, I'm concentrating on proper functioning, not size, as should be obvious. It's much easier to work on a large layout than with parts jammed together. It's also easier to work with single op-amps than quad packages during the breadboard stage. When transferred to a PCB layout, multiple section op-amps may make space and cost sense.

I'm now reading 60 Hz powerline frequency within 0.005 Hz based on a single cycle of data.
 

I've added a new page, (click here to read it,) still under development, discussing a few issues related to measuring 50/60 Hz power line frequencies accurately and quickly. Or, click on the Measuring 60 Hz Frequency link in the navigation bar.

This is a project under development, so it is necessarily incomplete. I plan to update it as my project progresses.

As usual at the first of the month, I've moved May's updates to an archive page, viewable by clicking here or via the navigation table at the top of this page.

May has been a wet month in northern Virginia - clocking in with more than twice the normal rainfall. My backyard weather station shows 8.28 inches of rain during May, with 1.25 inches of the total falling in less than one hour yesterday afternoon.

May is the wettest month in Fairfax County, with an average of 4.60 inches of rain, so we are not quite twice average. It's going to take several months of well over average perception to offset the the drought we've been in for the last year.

Today's Washington Post has  the official figures, taken at National Airport, about 15 miles east of Clifton:

Yesterday afternoon, wind-driven rain hung a gray curtain across the horizon, as a downpour left 1.45 inches of rain at National.

That brought the month's total in Washington to within .03 inches of the record for May of 10.69 inches. The average l for May is 3.82 inches. Last year's figure was 1.75 inches.

My 4.60 inch May average is from http://www.erh.noaa.gov/marfc/Climatology/vappn.htm and is for Vienna VA, which is perhaps half the distance from Clifton as is National Airport. The MARFC data also represents a shorter  time span, 1971-2000.

As far as radio projects go, I've been working with the color LCD I mentioned last month, with some associated measurement hardware and software. I'll have more to say about it in a couple weeks, I hope.