KA7OEI Weather Satellite Page


Here are the pictures as received from weather satellites using the receiver described on this page.


Please note:

There are a number of different types of weather satellite pictures available:

A few important comments about the pictures:
What's wrong with NOAA 17?

On October 15, 2010, the scan motor on the NOAA 17 weather satellite stalled and the ability for it to produce APT images was lost.  The current status of this satellite may be seen here.

For some months prior to this the scan motor would occasionally stall, resulting in distorted pictures - or none at all - as diagnostics were being carried out.  So, for the time-being, the NOAA 17 images have been removed from this page as they didn't show anything, anyway.

Will it come back online?

It's unlikely, although testing may be continuing.  In the past other satellites have had similar scan motor problems - and some of them have "fixed" themselves - but whether or not this will be the case with NOAA 17 is anyone's guess!

(As of mid-January 2011, I've noted that, occasionally, the images  from NOAA 17 aren't totally blank, showing either calibration/test lines or evidence of the scan motor occasionally "breaking loose" .)

 
About the weather satellite receiver:
Figure 1:  The front panel of the homebrew weather satellite receiver.
Click on the image for a larger view.
Front-panel view of the homebrew weather satellite receiver


Previously, I'd been messing with the WXtoIMG program, variously using my FT-817 or a service monitor (a piece of test equipment that has, among other things, a wide-range receive capability) to demodulate the weather satellite signals - but neither of these receivers were very satisfactory.  These transmissions, from polar-orbiting satellites, include those on frequencies in the 137 MHz area.  Using FM and having fairly strong signals, exotic equipment is not required to receive these satellite, but most standard receiving gear (scanners, amateur radio receivers, etc.) isn't particularly well-suited for their reception for a number of reasons:

In November of 2008 I decided that I wanted to put together a 137 MHz APT weather satellite receiver and decided to build it from the ground up using components onhand.  A few of the critical design goals were:

Quadrafilar helix antenna and preamplifier, used for weather satellite reception
Inside the preamp and filter
Figure 2:
Top:  The "Tall and Narrow" Quadrifilar Helix Antenna with the mast-mounted preamplifier used for weather satellite reception.
Bottom:  A look inside the enclosure containing the GaAsFET preamp (left) and bandpass filter.
Click on either image for a larger version.


In reality, it wasn't the desire for a weather satellite receiver that was the main motive for its construction, but rather that I'd always wanted to build a receiver entirely from scratch - plus, I wanted to try out a few circuit ideas, too, including:


To be sure, I could have made a simpler receiver than that described that was smaller, lower current consumption and used fewer parts, but remember that the idea was to build a receiver and have fun doing it!


The antenna:

For receiving signals from polar-orbiting weather satellites one must either have an antenna that tracks the satellite or use an antenna that is sufficiently non-directional to allow it to receive signals that come from anywhere overhead where the satellite might be.  Because the 137 MHz signals from these satellite are quite strong, one need not go through the trouble of assembling an automatically-tracking antenna system to get good results.

One simple antenna that can be used for weather satellite reception is the "turnstile" antenna.  While a good, simple antenna, it was not used here as it isn't quite as "sensitive" as some of the other options  While there are several other types of antennas that may be used, I chose a rather odd-looking antenna, the "Quadrifilar Helix Antenna" (or "QHA") as seen in Figure 2, top.  This antenna, like the turnstile, is circularly-polarized to match the transmissions of the satellite's antenna:  Whereas most antennas radiate their signals in a horizontal or vertical plane, the satellite uses an antenna that imparts a "spin" on the radio waves.  This has a definite advantage in satellite work over a "linearly" polarized antenna (e.g. horizontal or vertical) in that as the satellite goes overhead, you would have to make sure that your horizontal or vertical antenna matched however that of the satellite might be oriented:  If you are trying to receive a "horizontal" signal on a "vertical" antenna, most of the receive signals will be lost - much as what happens when you try to look at a wristwatch through polarized sunglasses and the dial face darkens!  By using circular polarization, rotation of either the transmit or receive antenna becomes irrelevant.

By making the antenna "tall and narrow" it exhibits more gain near the horizon (when the satellite is farthest away) than it does at high angles - such as when the satellite is overhead (at the zenith) and is the closest with the strongest signals.  By making this trade-off one can better-receive signals when the satellite is at a low angle and not only be able to receive the signal when the satellite is farther north and south, but also have a better chance of receiving signals when the satellite is farther to the east or west.

This antenna was built on a piece of ABS plastic pipe and using thin, copper refrigeration tubing.  While it looks fairly complicated, careful attention to the drawings and dimensions make it fairly easy to duplicate with good results.

For details on construction of some weather satellite antennas check these links - and don't forget to try searching on your own, too!:

For best results, the antenna should be mounted as high as possible to avoid blockage from ground-based objects such as trees and buildings and to remove it as much as possible from nearby interference sources:  A high, clear mounting location can considerably improve signals when the satellites are near the horizon.
Technical description of the receiver's circuitry:

Note:  A "nerd alert" is appropriate here as what follows contains a rather detailed description of the circuitry contained within the receiver and antenna preamp.


The mast-mounted preamplifier and bandpass filter:

In the list of "design parameters" above, the second item - "adequate" sensitivity" - deserves a bit of explanation.  For the best receive system sensitivity, one puts the preamplifier outside and at the receive antenna to minimize coax losses.  Putting a low-noise preamplifier outside also allows one to use inexpensive, small-diameter coaxial cable - such as that used for television reception - to connect the receiver to the antenna.  It is worth remembering, then, that the more amplifier gain one puts in front of the receiver, the more susceptible it would be to overload from strong signals - such as those from passing aircraft, paging systems, or even nearby ham transmitters.  By designing the receiver to have "adequate" gain, that meant that it could be used (with "good" but not stellar) performance without the preamp, but it wouldn't be as likely to be overloaded once the preamplifier was added.  In addition to the preamplifier, the box mounted outside would also contain a bandpass filter that would reject those signals farther away from the desired 137 MHz weather satellite frequencies, further improving the receiver's overall performance.
Weather satellite receiver, on the workbench before being put into its enclosure.
With the cover removed, a top view of the homebrew weather satellite receiver and all of its pieces.
Figure 3:
Top:  On the workbench:  In pieces, but working, the Weather Satellite receiver - before it was put into it's box, of course!
Left to right:  Display, CPU board (the small one with the ribbon cable), the main RF/VCO/PLL board, and the IF/Demodulator board - before the audio amplifier was added.
Bottom:  A top view of the receiver after all of the bits were crammed into a box.  Starting from the upper-right:  IF/Audio section (on perfboard), CPU/Control board (mounted on front panel), LCD (mounted on front panel), Local oscillator compartment (just behind ribbon cable), mixer and post-mixer amplifier (left of the local oscillator, behind the display), RF amplifier and filter (the coil and the compartment with the cover) and the frequency reference/PLL board (the one with the box with the white label.)
Click on an image for a larger version.

Even though the signals from the weather satellites are quite strong when overhead, they can be quite weak when the satellites are near the horizon.  This makes sense when you think about it:

Aside from boosting faint signals, an amplifier also overcomes the losses in the cable that connects the antenna to the receiver that would further degrade an already-weak signal.

For the preamplifier (see Figure 2, bottom) I constructed a simple GaAsFET amplifier.  This amplifier has a fairly low noise figure (in the 0.5-0.9 dB area) and modest (about 18dB) gain - both being sufficient to dig a weak signal out of the noise and boost it enough to overcome moderate cable losses.

A representative schematic of this preamplifier may be found here.  Even though this preamplifier is intended for operation in the 2-meter amateur band (144-148 MHz) it is easily re-tuned to the 137 MHz region.  Note that the linked schematic does not include the bandpass filter or the power-over-coax coupling.

Following the preamplifier is a 2-stage bandpass filter:  This filter is placed after the amplifier to minimize losses that would increase the amplifier's noise figure and reduce its sensitivity.  Because GaAsFET preamplifiers have reasonably good dynamic range, only very-nearby transmitters (such as my own 2-meter transmission) are likely to overload it.  The filters themselves knock out signals on other nearby frequencies (such as passing aircraft) that could overload other stages in the receiver as well as provide a higher amount of "image rejection" to prevent signals 21.4 MHz below that of the desired frequency (which are also in the aircraft band) from getting into the receiver.  Since this filter has a bit of loss (1-3dB) it should be placed after the preamplifier where this small amount of attenuation will not visibly affect signal quality.

To conduct signals from the preamp to the receiver - as well as to provide power for the preamplifier - standard RG-6 TV-type coaxial cable is used with F-type fittings.  This cable is ubiquitous, fairly small in diameter, cheap, and exhibits fairly low-loss characteristics at this frequency so it was an obvious choice.  Again, to minimize losses the preamplifier assembly is mounted at the antenna in a weatherproof enclosure:  Doing so allows a relatively short run of coax from the antenna to the preamplifier to minimize signal degradation.

Filter details:

As it turns out, a number of people have emailed me for details on the 2-stage bandpass filter that is used "post-preamp" to restrict the bandwidth and to make the receiver less susceptible to off-frequency interference.  This filter can be reasonably expected to improve the image rejection and to somewhat attenuate nearby 2-meter amateur transmissions, but it is NOT sharp enough to reject signals from paging systems that plague some European users of these satellites:  For that, you'll need to use a 1/4 wave notch cavity.  This can be constructed of standard copper water pipe and would be approximately 20" (51cm) long:  Details can be gleaned from the internet.

As can be seen from the picture in Figure 2 this filter consists of two coils, mostly separated by a shield with the pair of coils enclosed in a larger shielded enclosure consisting of approximately 1" (2.5cm) wide strips of double-sided copper circuit board material, all built on a larger piece of circuit board material - which could be single-sided if so-desired.  Small countersunk holes were drilled in the PC board material to allow easy connection to the coils' input and output taps.  Since the preamp and filter was mounted in a rather shallow metal box, I didn't put a top shield over the coils but if they had been put in a larger box or anywhere near a source of noise - such as a computer or digital circuitry then a shield of thin sheet brass should be placed over the top of each coil, individually, and tack-soldered in a few points each with a small hole to access the tuning screws of the trimmer capacitors.

The coils themselves consist of approximately 8 turns of #14 AWG solid copper from electrical house wire, close-wound wound on a 0.25 inch form (approx. 6.2mm) and then stretched after winding to space adjacent turns about 1/2 wire diameter apart.  It is recommended that one coil be wound clockwise and the other counter-clockwise and placed so that the "1 turn from ground" tap point (see below) of each coil be located adjacent to the hole through which the connecting wire (to the tap) is passed.

The input and output "taps" are located about 1 turn from the grounded end of each coil and 2-20pF trimmer capacitors are used at the "top" end for tuning and the two coils themselves are coupled to each other using 0.5-1pF of capacitance which can either be in the form of one or two 1pF caps in series, or a small "gimmick" capacitor.  Practically speaking, the way one would wind and mount these coils they will have closer to 8-1/2 turns, but because of the available tuning range offered by the 2-20pF trimmer caps, the actual dimensions of the coil aren't terribly critical.  The gauge of wire can also vary from #12 AWG to #16 AWG, depending on what is available, but the heavier wire allows the coil to be self-supporting and rigid. 

In this filter, I used inexpensive ceramic trimmer capacitors which are adequate for the job.  This filter is narrow enough that it is best that it be adjusted on the center-most frequency of interest (about 137.50 MHz) as there will be some attenuation of the frequencies at the two extremes of the receiver's tuning range (e.g. the 137.10 and 137.9125 MHz frequencies) but since this filter is intended to follow a low-noise preamp, its slight additional loss should not measurably affect the overall signal quality:  In other words, do not put this filter between your antenna and preamp!

If one wishes to optimize performance of the filter, the more-expensive glass or porcelain "piston" type trimmers may be used - along with a 0.5-2.5 pF piston trimmer as the coupling capacitor between the two filter sections to allow its "shape" to be optimally adjusted using a sweep generator or similar piece of test equipment.  If one really is picky, the coils themselves could be silver-plated or, at the very least, enamel wire could be used (or the bare wire sprayed with clear lacquer after the filter is constructed) to avoid losses due to its oxidation.  If an even "sharper" filter is required, additional sections could be added, but the proper adjustment of such a filter would definitely require test equipment and a bit of extra homework to implement!

Receiver front end:

Refer to schematic pg. 1

To power the external preamplifier, power is fed to the input coax via J101, a standard TV-type "F" connector and L101, a 10uH choke:  This blocks the received signal while passing the DC to run the amplifier while capacitors C102 and C103 shunt away any residual RF that might make it past the choke.  DC power is supplied via Q101 and associated components which form a DC current source - the purpose of which is to limit the available current should the RF cable be shorted out to a safe value in the area of 40-60 milliamps - more than enough to run the external preamplifier.  Without it, the full current of the power supply running the receiver would be present and an accidental short (easy to do with F-type fittings!) would instantly destroy L101.  Switch S101 allows this voltage to be disabled, if desired.

The receiver front end is fairly simple - a two-stage bandpass filter with integral JFET preamplifier, Q102.  For this, a grounded-gate amplifier, based on a J309, is used with a tuned circuit on the input (L103 and C105) and output (L103 and C107.)  This amplifier offers modest performance, providing a gain of 10-12dB with a noise figure of 2-3dB - more than adequate to overcome the majority of the mixer noise.  The filtering of this amplifier - while not particularly "tight" - is adequate for the purpose:  Remember, there is better (lower-noise) amplification and filtering at the antenna!  As can be seen from Figure 4 the "input" coil has a shield placed over it to prevent noise from the digital circuitry in the receiver from being coupled into it.

The signal from the preamplifier goes to a diode-ring passive mixer U101 - but between the input and ground is a simple, series L/C circuit consisting of L104 and C109 that is tuned to the image frequency of the receiver (lower than the receive frequency by twice the IF frequency, or 21.4 MHz below) around 116 MHz.  This provides at least 20dB of additional image rejection, yielding about 55dB overall of image rejection to the receiver.  Again, this figure is "adequate" but not great, remembering that there is more filtering at the antenna that improves the image rejection to well above 75dB.

Even without the antenna-mounted preamp, the 12dB SINAD sensitivity of the receiver with the JFET preamp alone is better than a microvolt - good enough to be used with a short run of low-loss cable and a decent antenna for reception of "high" satellite passes.
Figure 4:
The "front-end" and mixer of the receiver.  On the far left can be seen the antenna input connector and the current-limit circuit.  Under the shield is a coil - much like the one that is visible to its right - along with the JFET for the front-end amplifier.  In the compartment on the right is the image-reject filter, the mixer and post-mixer amplifier.
Click on the image for a larger version.
Filter/amplifier compartment of the receiver

Mixer:

Refer to schematic pg. 1

A standard diode-ring mixer, a Mini-Circuits RMS-11X, is used at U101.  This is a passive device, having good dynamic range and requiring only a local oscillator to convert signals.  Because it has a fairly high insertion loss (around 7 dB) it is necessary to precede it with the preamplifier described above to achieve good system sensitivity.  Even though this is a surface-mount device, it may be used with "dead bug" construction if heavy ground conductors (such as copper foil) are used to interconnect its ground leads to the receiver's ground plane.

Following the mixer is a very simple "highpass/lowpass" diplexer to assure that signals being outputted by the mixer (particularly the "local oscillator plus receive signal" image that is at about 260 MHz) are properly terminated:  C102 and R108 are chosen to provide something resembling 50-ish ohms at the image frequency while the combination of C114 and L105 form a broadly-resonant circuit to allow the 10.7 MHz-area signals to reach the post-mixer amplifier, Q103.  It's worth noting that the failure to terminate the mixer at both the desired and image signal frequencies resulting from a conversion can result in undesired mixing products, potentially reducing receiver performance considerably - especially in the presence of other signals within the passband of the pre-mixer RF filtering.  Since there is always a bandpass filter preceding the mixer, there aren't really any significant signals present other than the sum and difference mixing products.  The post-mixer amplifier, Q103, is designed to terminate the mixer at around 50 ohms.

IF Amplifier:

Refer to schematic pg. 2

The amplified signal, now at the I.F., from the mixer is further amplified and broadly filtered by several IF amplifier stages, finally reaching Q203 and being passed through a narrow (40 kHz) ceramic filter, CF203:  It is this last filter that sets the ultimate bandwidth of the receiver.  While the 10.7 MHz, 40 kHz wide ceramic filter isn't particularly "sharp", it seems to be more than adequate in eliminating adjacent-channel signals - such as those from the Orbcomm satellite network.  C208 and L202 roughly match the 600-ish ohm output impedance of the ceramic filter to the 50-ish ohm input impedance of U201 and its associated circuitry.

Initially, I was worried that having fairly wide-bandwidth filtering in the early IF stages was going to make the receiver susceptible to interference from nearby Orbcomm satellite and Aeronautical traffic - either of which can be within a MHz or two and for this reason, I originally had a 40 kHz filter installed at CF202 as well.  Because of the fickle nature of ceramic filters and the fact that their bandpass response is often less-than-ideal (that is, not particularly "flat" across their passband but a more "rounded" curve) I observed some degradation, particularly on weak signals that were slightly offset in frequency due to Doppler shift so I replaced it with a 150 kHz wide filter which resolved the problems.  I was initially worried that this would make it more-susceptible to nearby-frequency interference, but "on-air" testing has not revealed any tendency for this to happen.
Figure 5:
The IF amplifier and audio board.  The 10.7 MHz IF signal enters at the right and passes through several filter/amplification stages.  U201 - the demodulator chip - is seen, along with the quadrature coil.  Toward the lower-right corner of the board can be seen U202, the LM386-4 audio amplifier.
 Click on the image for a larger version.
The IF amplifier/Demodulator/Audio amplifier board.

Demodulator:

Refer to schematic pg. 2

The demodulator uses the venerable LM3189 chip, U201.  This is a fairly easy-to-use IC and although it is obsolete, it is still readily available on the surplus market:  I used it just because I happen to have a bunch of them onhand.  This chip contains a limiter and quadrature detector and provides both an "AFC" output (Automatic Frequency Control - for determining if the received signal is on-frequency) and an "RSSI" (Received Signal Strength Indicator) output:  The former can be used to track Doppler shift, if desired, and the latter can be used to provide a signal strength reading.

When using this chip it is important that good wiring practices be employed with sufficient RF bypassing.  Failure to do this can cause the LM3189 (or its predecessor, the '3089) to do odd things, most commonly "seeing" a signal that isn't there as evidenced by elevated RSSI readings when no actual input signal (even noise) is present.  Despite all of these precautions, I originally found it necessary to use as much "pre-limiter" gain as I did in order to assure that, with the receiver running "barefoot" (that is, just the JFET preamp and not the external preamp) that there was at least some RSSI indication on the receiver's own noise floor.

In order to keep high-gain RF circuits like this happy, it's imperative that a good RF ground is provided - something that is difficult to do on perforated proto-board like this.  For this reason I have strips of self-adhesive copper foil along portions of the bottom of the board:  Without it, it is possible that some of the signal from U201 will find its way back into the input and cause instability which would degrade performance.  A bit of this copper foil is just visible along the left edge of the board as it wraps around to the top side.

Comment:
I later discovered a minor wiring error that reduced the level of the IF signal:  Now, everything works as it should - more details are at the bottom of the page.  Remember:  You learn more by making (and fixing) mistakes than you do if everything goes right the first time!
The demodulated "audio" output of this chip is buffered by Q204 and related component with R226/C226 offering a degree of de-emphasis/lowpass filtering (mostly to remove high-frequency noise components) and made available to the computer doing the processing.  Not shown in the schematic is a 1:1 audio transformer (600 ohm) that is used to prevent a ground loop between the receiver audio line and the computer - something that could introduce hum under certain conditions.

Also included is an audio amplifier, U202, based on the LM386, that is used to drive a speaker.  While an "S-Meter" squelch (one that is based on signal strength rather than signal quality) could have been implemented with the LM3189, I have not done so yet - and since the computer doesn't really care if it "sees" noise all of the time that there's not satellite being received, there doesn't seem to be any real reason to do so.  Since I only turn up the volume occasionally to see if things are working properly, the constant "hiss" doesn't bother me!

Local oscillator:

Refer to schematic pg. 1
Figure 6:
The local oscillator module.  The micro-coax resonator may be seen in the upper-right corner.
 Click on the image for a larger version.
The local oscillator/buffer section

The local oscillator is a simple Colpitts circuit using an MPF102 JFET at Q301 and it operates 10.7 MHz below the receive frequency.  The inductor L301 uses some extremely small-diameter (about the size of #18 AWG wire) PTFE "hardline" coax forming a 1/4 wave resonator.  While I could have simply used a standard inductor/capacitor arrangement, I was curious as to how well this 1/4 wave transmission line resonator would work.  One advantage of this method is that it is less-sensitive to microphonics - that is, the tendency for mechanical vibration to modulate the oscillator, the result ending up back in the receiver audio which, in the worst-case scenario, can cause feedback with the loudspeaker.

In testing, I have found that the local oscillator is very insensitive to microphonics despite the fact that the speaker is only inches away and that most of the tendency for microphonics is, in fact, from the mechanical vibration of Q301 and its associated components.  As such there is no obvious tendency of the receiver to "ring" due to microphonics and even though the loudspeaker is in the same box as the VCO, there is very little effect on the received pictures (due to microphonic action) when the audio amplifier is turned all of the way up indicating that this experiment was successful.

A varactor diode D301 is used to tune the oscillator and a range of about 126.0 to 129.5 MHz is provided, covering all possible satellite frequencies in the 137 MHz range, plus allowing the receiver to tune to 140 MHz:  This 140 MHz tuning capability is provided solely to allow the receiver to tune in a harmonic of the 20 MHz CPU clock/reference signal used in the receiver and allow a means of self-calibration in terms of the receiver tuning and the AFC.

Following the oscillator are two emitter-follower buffer amplifiers:  The first of these, Q302, isolates the oscillator from the rest of the circuitry while the second, Q303, isolates the oscillator yet again, providing a drive signal to the synthesizer section, described below.  This isolation is important as the synthesizer section contains a frequency mixer (Q401 - see below) that could inject spurious signals into the local oscillator signal and degrade receiver performance.  The output of Q302 goes to yet another stage of amplification, Q304, which boosts the local oscillator signal to about +7dBm to drive the diode-ring mixer, U101.

The power supply for the VCO and buffer comes from U301, a 78L09 regulator.  This device provides a clean, stable source of power for the VCO to prevent changes in the power supply voltage from affecting the frequency or modulating the local oscillator.

LO Converter and Master frequency reference:

Refer to schematic pg. 3

In order for a frequency synthesizer to work with good accuracy, a stable frequency reference is needed.  In this case, a 100 MHz VCXO (U401, a Voltage-Control Crystal Oscillator) is used:  This device was rescued from a piece of scrapped satellite equipment and has been found to be stable to better than +-1kHz over a fairly wide temperature range - adequate stability for our purposes.  The output of this oscillator is amplified to "TTL" level by Q404 and fed to U403, a 74F191 wired as a "divide-by-5" counter to produce a 20 MHz signal that is used to drive the CPU.

The 100 MHz output also goes to a single-transistor bipolar mixer, Q401, where it is combined with a sample of the 126-130 MHz buffered output of the local oscillator from Q303.  The 26-30 MHz output from this mixer (the "difference" between the 126-130 MHz local oscillator and the 100 MHz VCXO) is low-pass filtered by C405-C407 and L402-L402 and then amplified by Q402, broadly filtered again by C413 and L404 and amplified to "CMOS logic" level by Q403 before being inputted to U402, a 74HC4040 binary counter, where it is divided by 4096 to produce another signal that is in the 6.3-7.3 kHz region.  It is this frequency that is compared with the PLL reference signal generated by the CPU in the PLL.

Figure 7:
The LO Converter, Master frequency reference and PLL board.
The audio-frequency reference from the CPU comes in at the upper-left and is filtered and fed to the PLL chip.  On the upper-right is the LO converter that "subtracts" 100 MHz from the local oscillator signal.  The box with the white label is the 100 MHz VCXO.
 Click on the image for a larger version.
The LO converter, reference and PLL board.
Phase-locked Loop:

Refer to schematic pg. 3

The Phase-Locked Loop (PLL) is used to control the frequency of the VCO.  It does this by comparing the frequency of the down-converted LO (the 6.5 kHz-ish signal from U402, the 74HC4040) with a locally-generated audio frequency being used as a reference - also in the 6.5kHz range:  This latter frequency is chosen to be the same as that which would be produced by the VCO's down-conversion if it happened to be on the correct frequency.  The receive frequency is related to that down-converted frequency in this way:

Down-converted (audio) frequency  (in kHz) = ( (Receive Frequency in kHz - 10700 kHz) - 100000 kHz) / 4096

The PLL reference signal, generated by the CPU (and discussed below) is inputted to a bandpass filter consisting of U501C and U501D.  This filter takes the rather "rough" signal from the CPU and filters it to produce a clean sine wave, using D501 and D502 to limit it to a constant amplitude.  The output of this filter is then passed to a "slicer" consisting of U501A, R507 and C504, the output of which is a nice square wave.  This is then fed to U502 via R508, which effectively allows the 0-12 volt square wave from U501A to drive U502 which only needs a 0-5 volt signal.
Comment:  It was later determined that this bandpass/limiter filter was probably not necessary as the loop filtering proved to be adequate.  Since it was already built and working, I didn't bother removing it!
The PLL chip, U502 - a 4046 - compares the phases of the down-converted signal with the locally-generated signal and if the VCO is too-high in frequency, it steers the VCO's frequency downwards and if it is too low, it steers it upwards.  With the proper selection of time-constants, the VCO would immediately "snap" onto the desired frequency with good stability.

To do this "frequency steering" a "loop filter" is employed on the output of U502 consisting of U501B and associated components.  The heart of this circuit is U501B and its associated capacitors, forming an integrator that takes the varying pulse-width from U502 and smoothes it to a voltage that will appropriately move the VCO up and down.

On the input of this loop filter are a number of additional components:  R510, a 1 Meg resistor, together with C510 and C509/R511 have a rather long time-constant to provide considerable filtering of the tuning voltage to remove all traces of the audio-frequency reference signal.  If these were used alone, it would take a rather long time for the synthesizer to tune from one frequency to another (perhaps 5-20 seconds) - a lock time that I considered to be unacceptable - at least to someone tuning the receiver manually:  Practically speaking, in automated weather-satellite use, this time would not be a problem as the receiver is tuned to the proper frequency at the beginning of the satellite pass and the loss of a few seconds of signal would be of no real importance.

To speed up the tuning, additional components were added:  R509, a 33k resistor works with D503 and D504 and only larger-sized changes in the output from U502 will "break over" those diodes.  The output from these diodes is filtered by C508 and this voltage is then applied to Zener diodes D505 and D506.  These series-wired diodes don't conduct until the voltage exceeds 3 volts, but once they do, capacitors C509 and C510 are charged/discharged very quickly.  The effect of this is that when the frequency is "way off" U502 will slam to full V+ or ground and the  extra diode circuitry will conduct, quickly steering the local oscillator.  Once the frequency gets "close" to its intended target the output of U502 will decrease and these diodes will no longer conduct and the "slow" time constant of R510 will again take over.  It is this "dual time constant" that allows the best of both worlds - "slow" filter that permits a clean, stable output for keeping the local oscillator on-frequency and a "fast" one that allows any frequency to be tuned in under a second.

CPU:

Refer to schematic pg. 1

The CPU has several jobs to do:

Q601 buffers and amplifies the 20 MHz signal from the PLL board and feeds it to the CPU, U601.  The PWM output of the CPU is coarsely filtered by R602 and C604 to remove the highest-frequency components that might otherwise find their way into the IF or RF circuitry.  Q602 is a simple inverter that takes the RS-232 signal from the controlling computer and converts it to a voltage compatible with the CPU's logic input, using the chip's on-board pull up resistor.

Figure 8:
The CPU board.  The processor generates the precise audio frequency for the PLL reference as well as driving the LCD (display), reading the pushbutton(s) and receiving serial-port commands.
 Click on the image for a larger version.
The CPU/Controller board.
The CPU, a PIC16F88, does all of these functions using its onboard peripherals.  For the audio frequency-generation, its onboard PWM hardware is used as a simple D/A converter and DDS (Direct Digital Synthesis) techniques are used to produce a sine wave with 10 bits of resolution.  The DDS software was written to use a 32 bit accumulator which means that the audio frequency can be very-precisely generated in steps of about 0.000005 Hz - yes, that's 5 microHertz! 

Because the downconverted frequency is related to the local oscillator frequency by a division-by-4096, that means that the local oscillator frequency (at 126-130 MHz) can be controlled with far less precision than that of the reference frequency - but still, to within 0.02 Hz!  While this amount of precision is meaningless when it comes to the stability of the 100 MHz reference oscillator and tuning in an FM signal, it was easy to accomplish in software and provides more than adequate frequency resolution:  A DDS synthesis with only 16 bits of accumulator resolution (the next "logical" step downwards in terms of software) would provide steps of about 1.3 kHz - a size that I considered to be too coarse.  Because the CPU is clocked from the 20 MHz signal originally derived from the 100 MHz oscillator, it has a stable, accurate reference for the generation of the local oscillator frequency, holding frequency to 1-2 kHz over a wide temperature range.

For those readers more familiar with the design of DDS and PLLs, eyebrows might be raised as to the use of techniques with such a large divisor ratio!  One concern would be the inevitable spurious responses intrinsic to DDS synthesis causing undesired phase modulation of the 6.3-7.3kHz reference signal.  With fairly aggressive loop filtering (e.g. fairly long time constants) and the careful selection of the DDS clock frequency, one can "relocate" these spurious responses away from the loop filter's bandwidth and generate audio frequencies with sufficient cleanliness so that undesired "reference sidebands" are attenuated adequately by filtering.  Additionally, relatively simple techniques of using a "dual time constant" in the PLL's loop filter can allow the receiver's local oscillator to be on-frequency in well under a second, even with such aggressive loop filtering.

In addition to generating the reference frequency for the local oscillator, the CPU also drives U603, a 2-line by 16 character LCD.  This display is operated in the "4-bit" mode to reduce the number of connections to the CPU.  Because the CPU has a limited number of I/O pins, the two pushbuttons are also connected across the data lines that feed the LCD and when the display is not being updated, the status of these buttons are being read.  While one of these pushbuttons is used to select a pre-programmed weather satellite frequency (plus the 140 MHz "calibration" frequency) the other is presently unused - mostly because I haven't quite figured out what to do with it!  Another job of the CPU is to digitize some analog input voltages from the AFC circuit (to determine offsets from Doppler shift) as well as the RSSI output (to determine signal strength) and display them - although neither of these parameters are currently used by the receiver to affect tuning.

The LCD - a surplus 16 character-by-2-line unit - displays not only the frequency being received, but also signal strength and frequency-offset readings - mostly for my curiosity.  Finally, as mentioned above, the CPU also contains a UART that can receive serial commands from the computer running the WXtoIMG weather satellite image decoding program that not only tells the receiver what frequency to select, but also when the computer is processing a signal - a case in which the CPU puts an "RX" on the display to let one know that a satellite pass is in progress!


Measured performance:

Using my trusty service monitor, here are some measured specifications:
 Again, it should be remembered that there is also mast-mounted GaAsFET preamplifier with a bandpass filter is placed ahead of this receiver, so the actual sensitivity should easily reach that of the thermal noise limit (that is, Earth, atmospheric and electrical noise would be the limiting factor for the receiver's performance and not its intrinsic sensitivity.)

Final comments:

Having built the receiver several years ago and having it powered up continuously since then, it seems to work quite well as can generally be seen from the pictures - despite the local, low-level interference source (the "wavy" bars) that seems to be present in the vicinity of some of the weather satellite frequencies.

Were I to build another receiver from scratch, there are a few things that I might do differently:
 Is this design ideal?  No, of course not - but that's not the point!  I just wanted to try my hand at putting something like this together with stuff that I had laying around and were I truly serious about the ultimate in design/performance/elegance/simplicity I'd have done things differently!

So, would I recommend that someone wanting to build a weather satellite receiver duplicate exactly what I've done?  No, but that's because I'm sure that some things could be done better!


Want to send email?  Go here.

This page maintained by Clint, KA7OEI and is copyright 2008-2012.  Last update:  201120222
 

Since 12/2010: