Matjaz Vidmar, S53MV

NOAA HRPT Receiver

VHF Communications 3/1997

1. Reception of NOAA HRPT Images

The reception of weather satellite images always attracts lots of interest among radio amateurs. Of course, after the initial experiments one always wants to improve the results, trying to obtain even better pictures with less noise, better contrast and better geometrical resolution. The current receiver technology allows us to quickly reach the limits imposed by the satellite sensors and picture transmission standards.

Almost everyone starts by receiving the simple APT/WEFAX analogue picture transmissions. After perfecting the APT/WEFAX receiver, the next logical step is to switch to digital transmissions. Most weather satellites offer an improved signal-to-noise ratio and an improved geometrical resolution on their digital picture transmissions.

The main drawback of the digital picture transmissions is that they are much less standardised than the simple APT/WEFAX transmissions. Almost every satellite is using a different image data format, that requires different hardware for reception. In addition, the digital transmissions usually require a larger antenna operating on higher frequencies, making the ground station much more expensive.

The most popular digital image format is certainly the NOAA HRPT format, which has been used since the launch of the TIROS-N satellite in 1978. The NOAA HRPT format offers an improved geometrical resolution of 1km and an excellent signal-to-noise ratio (10-bit quantization) when compared to the analogue APT format with 4km resolution originating from the same satellites of the NOAA series (1), (2), (3). The same data transmission standard is still used today by many weather satellites (5), (6).

At the beginning, NOAA HRPT reception was a rather difficult technology for radio amateurs. The first amateur experiments (4) required hand-steering of the antenna and immediate conversion of the digital data into an easier-to-handle analogue APT-like format. There were no suitable computers available and microwave semiconductors were very expensive (the first GaAsFETs like the famous CFY11 were priced over £80 each).

Today the situation is completely different. Inexpensive computers can be used both for antenna tracking and for image storage and display. Parabolic dishes of various sizes and AZ/EL antenna positioners are easily available too. Microwave components became inexpensive as well and top performance low-noise preamplifiers can be readily built (7).

The only missing part is a suitable receiver to process the NOAA HRPT RF signal and output the data in a suitable format to a computer. Such a receiver will be described in this article, based on a design that has been operating for more than five years in the author's receiving station and has been successfully duplicated by many other radio amateurs as well.

2. NOAA HRPT Receiver Block Diagram

The general block diagram of the NOAA HRPT receiver is shown on figure 1. Using a state-of-the-art LNA (7), a 1m diameter dish antenna with a RHCP feed will provide an error-free reception at elevations above 10 degrees. Of course, the antenna needs to be installed on a computer-controlled AZ/EL rotator to track the quickly-moving polar-orbiting NOAA satellites.

A similar antenna/feed/LNA combination for the reception of amateur satellites in the 2.4GHz band has been described in (8). The latter can be easily scaled to the lower frequency band of 1.7GHz. Since the helical feed described and the LNA are both wideband, even the original 2.4GHz version actually operates in an excellent way at 1.7GHz as well, making it possible to use one antenna and one LNA for both weather satellites at 1.7GHz and amateur radio satellites at 2.4GHz.

To receive NOAA HRPT transmissions, most amateurs usually use different downconverters and IF strips. Such designs tend to be unnecessarily complex and the overall receiver performance is poor due to the unsuitable hardware. Therefore it was decided to develop a completely new design to avoid such problems.

The circuit shown In Fig. 1 is a single conversion receiver with an IF of 36 MHz. The latter is a convenient choice since widely-available television receiver SAW filters have just the appropriate bandwidth for the NOAA HRPT signals as well. Since 36 MHz is a relatively low value when compared to the input frequency in the 1.7 GHz range, the downconverter needs to be designed carefully. On the other hand such a design is still simpler than a double or even multiple conversion receiver.

Besides the choice of the IF frequency one also has to consider the required demodulator(s). The demodulation of a NOAA HRPT signal can be split into three different steps. In the first step, a digital signal is obtained from the modulated RF carrier in a PSK demodulator. The second step involves bit-rate synchronisation and bit conditioning. Finally, the third step is the frame synchronisation.

All three demodulation steps are included in the described receiver that provides a serial data stream, a bit clock and frame pulses as its output. These signals are usually required by the computer interfaces used. However, if a computer interface already includes the frame synchronisation or even the bit-rate synchronisation, the corresponding circuits can simply be omitted. >/P> A matching HRPT interface to the "DSP computer" was presented in (11). Of course, there are many different interfaces available for the "IBM PC compatible" computer family. Finally, one can also feed the digital data to a D/A converter and convert the signal to a 2400 Hz amplitude-modulated subcarrier, but in this way both the signal-to-noise ratio and the geometrical resolution of the images are degraded.

3. NOAA HRPT Downconverter

The circuit diagram of the NOAA HRPT downconverter is shown in figure 2. Since the downconverter is included in the indoor receiver, it is supposed that an external LNA with a gain of 25-30dB is used all of the times. Therefore the downconverter is not optimised for the best noise figure.

At 1.7 GHz one has to consider the cable loss between the LNA and the indoor receiver. At a typical distance of about 25m it is easy to keep the cable loss below 10dB, for instance by using the low-loss coax cables developed for satellite TV IF. Of course the circuit of the downconverter includes the supply network for the LNA that provides +12Vdc on the RF input connector. >/P> The downconverter includes two RF amplifier stages (two MRF571s) and a harmonic mixer with two BA481 Schottky diodes. Due to the low IF of only 36 MHz, filtering the image frequency, just 72 MHz above the desired frequency, is not easy. Since high-Q resonators can not be built in microstrip technology, the downconverter is using band-reject filters to attenuate the unwanted image response. >/P> The downconverter includes three almost identical filters, each consisting of four microstrip resonators. The two inner microstrip resonators (for example L4 and L5 in the first filter) operate as a rather wide band-pass filter with the bandwidth in the range of 200 MHz. Of course L4 and L5 alone are not able to provide any significant attenuation of the image frequency.

The two outer resonators (for example L3 and L6 in the first filter) operate as absorption traps for the image frequency, 72 MHz above the desired reception frequency. The overall combination is a band-pass/band-reject filter. Three such filters provide more than 40dB attenuation of the unwanted image frequency. The main purpose of the two RF amplifier stages is to compensate for the loss in these filters.

The mixer uses two antiparallel Schottky diodes (BA481) and requires a local oscillator signal at half the conversion frequency (around 870 MHz). The main drawback of this simple circuit is a higher noise figure, usually around 15dB.

The downconverter is built as a microstrip circuit on a double-sided FR4 glassfibre-epoxy printed-circuit board with the dimensions of 80mm x 125mm. The upper side is shown in figure 3, the lower side is not etched. The component location is shown in figure 4. The construction of similar microstrip circuits has been widely discussed in (9) and (10).

An adjustable-frequency signal source is required for the correct alignment of the image frequency traps. In practice, L3, L6, L8, L11, L13 and L16 usually need to be shortened by about 1mm during the alignment procedure. Other microstrip resonators hardly need any adjustments. L17 may be made slightly longer for the best mixer noise figure.

Since the mixer includes two antiparallel diodes, no DC voltage is generated during operation. The local oscillator chain is therefore adjusted by connecting an ohmmeter to the IF output and tuning the multiplier stages for the minimum resistance. The maximum LO drive may not correspond to the best noise figure, but in this case it is better to have some safety margin on the LO signal level.

4. NOAA HRPT receiver Local Oscillator Multiplier

NOAA HRPT transmissions are usually encountered on two different frequencies: 1698.000 MHz and 1707.000 MHz. The 1698.000 MHz channel is usually assigned to the morning/evening satellites while the afternoon/midnight satellites transmit on 1707.000 MHz. All NOAA satellites carry onboard three 1.7 GHz transmitters and in addition may transmit HRPT signals on 1702.500 MHz in the case of failure of both primary transmitters.

However, the transmitter on 1702.500 MHz is connected to a LHCP antenna and requires a polarisation switching capability at the ground station. Maybe this is the main reason why the 1702.500 MHz transmitter has never been used for HRPT transmissions although it has been used for other data transmissions. A NOAA HRPT receiving station therefore only requires two channels, 1698.000 MHz and 1707.000 MHz, both with RHCP antenna polarisation.

Other satellites may transmit on different frequencies. For example, the Chinese FENG-YUN satellites were transmitting fully NOAA HRPT compatible images on 1695.500 MHz and 1704.500 MHz. Although this receiver is not suitable for the digital transmissions originating from geostationary weather satellites, due to the different data rates and modulation techniques, it is nevertheless useful to have the reception capability of at least the main WEFAX channel at 1691.000 MHz. The reception of the latter is convenient for antenna rotator calibration and LNA/RF front-end checkout.

Since a NOAA HRPT receiver only needs to be tuned to three or four different frequencies, the local oscillator can be crystal-controlled, followed by a multiplier chain. The crystal-controlled local oscillator is shown in figure 5. Each crystal has its own oscillator and channel selection is performed by turning on the desired oscillator.

The following multiplier chain is shown in figure 6. The overall multiplication factor is 64, but the last frequency doubling is performed by the harmonic mixer itself. The oscillator transistors are already able to provide the second harmonic at around 54 MHz. This frequency is then multiplied by four to about 217 MHz, then doubled to 435 MHz and finally doubled to 870 MHz.

The NOAA HRPT receiver local oscillator multiplier is built on a single-sided printed-circuit board with the dimensions of 80mm x 100mm, as shown in figure 7. The corresponding component location is shown in figure 8.

L1, L2, L3 and L4 have an inductance of about 1.2uH and their main purpose is to force the crystals to oscillate on the desired third overtone at 27 MHz. In addition these adjustable coils allow some fine frequency tuning to compensate for the crystal tolerances. L5 and L6 have about 0.3uH each. In practice L1, L2, L3 and L4 have 10 turns each on a miniature TV IF transformer core while L5 and L6 have 5 turns each on the same type of core.

L7, L8 and L9 are self-supporting coils with three turns of 1mm copper enamelled wire each, closely wound on a 5mm internal diameter. All three coils should have the same orientation to obtain the proper magnetic coupling. L12 is a self-supporting coil with three turns of 0.5mm copper enamelled wire, closely wound on a 3mm internal diameter. Finally, L10, L11, L13 and L14 are etched on the printed circuit board.

The oscillators should operate immediately without any alignment. The multiplier chain however requires alignment. The latter should be performed stage by stage by checking the corresponding signal level on the base of the following multiplier transistor. In fact it is enough to check the DC voltages, since the BE junction operates as a rectifier.

If the operation of the last multiplier stage is found unstable, the wire leads of the BFY90 transistor should be shortened. If this does not suppress parasitic oscillations, the spacing of the turns of L12 should be modified.

5. NOAA HRPT Receiver IF Amplifier

The NOAA HRPT receiver IF amplifier is shown in figure 9. The design of the IF amplifier is based on the available 36 MHz SAW filters. The overall bandwidth of NOAA HRPT transmissions is around 3 MHz, so a PAL television receiver SAW filter with a bandwidth of about 4-5 MHz is a very convenient choice. SAW filters have a flat passband and very steep edges, but most important of all they have a flat group delay so that they do not distort fast digital signals.

The remaining circuit of the IF amplifier is also based on available components for television receivers. The first low-noise amplifier stage with the BFR90 transistor is followed by yet another amplifier stage (BFY90) to compensate for the high insertion loss (15-20dB) of the SAW filter. The SAW filter is followed by an integrated IF amplifier TDA440. The latter also includes an AGC circuit.

In addition to the SAW filter, some selectivity is also provided by the tuned circuit with L1. On the other hand, there is a wideband transformer L2 on the output of the TDA440. The video demodulator inside the TDA440 is not used except for steering the AGC. Although the NOAA HRPT PSK signals can be limited, operating the IF amplifier in the linear region makes the bandwidth of the filters much less critical.

The NOAA HRPT receiver IF amplifier is built on a single-sided printed-circuit board with the dimensions of 40mm x 100mm, as shown in figure 10. The corresponding component location is shown in figure 11.

L1 has about 0.3uH or 5 turns on a miniature TV IF transformer core. L2 is a wideband transformer with the primary inductance of about 2uH and the turns ratio of 5:1, in practice 10 turns and 2 turns on a miniature 10.7 MHz IF transformer core. Of course, L1 needs adjustment while the core of L2 can simply be adjusted for the maximum inductance.

Any 36 MHz SAW filter with a single input and output should be suitable for the IF amplifier. These filters are available in two different packages: TO-8 metal can and plastic single-in-line. The printed-circuit board has holes for both types of packages. A slight offset of the SAW filter centre frequency or local oscillator crystals can always be corrected with L1.

The 10kohm trimmer sets the AGC threshold or the output signal level. Warning! The circuit does not work with some integrated circuits, in particular with the TDA440S. Therefore be careful to obtain a TDA440 without any suffix letters!

6. NOAA HRPT Receiver PSK Demodulator

The 665.4kbit/s NOAA HRPT serial data stream is first Manchester (split-phase) encoded. The Manchester-encoded signal then modulates the phase of the RF carrier. The phase modulation amounts to +/-67.5 degrees nominally.

The RF spectrum of such a signal includes an unmodulated carrier at the centre frequency and two symmetrical sidebands with the maximum offset +/-655.4 kHz from the centre frequency. The unmodulated carrier level is about 8dB weaker than the overall signal strength. In other words, about 0.7dB of the transmitter power is lost in the residual carrier and the remaining power goes to the information-carrying sidebands.

In spite of the rather wide signal spectrum and power loss in the residual carrier, such a transmission standard was probably selected to keep both the satellite transmitters and the ground station demodulators as simple as possible. A coherent demodulator only requires filtering out the residual carrier, phase-shifting it by 90 degrees and multiplying the regenerated carrier by the raw RF signal. All of these functions can be performed by a single, simple phase-locked loop.

The phase-locked loop NOAA HRPT receiver PSK demodulator is shown in figure 12. The VCO and multiplier are included in the integrated circuit S042P. The same multiplier is used both as the PLL phase detector and as the PSK data demodulator. The PLL loop filter includes a 358 dual Op-amp, while the output data is converted to TTL level by a 311 voltage comparator.

The locking range of the PLL should allow for VCO drift and for the Doppler shift, which amounts to about 100kHz for a polar-orbiting NOAA satellite at 1.7 GHz. From this point of view, the choice of 36 MHz as an intermediate frequency is a fortunate coincidence. Finally, the described PLL PSK demodulator is also an efficient FM demodulator that can be used to receive the analogue WEFAX transmissions from geostationary weather satellites.

The audio-frequency monitor is however mainly intended to check the proper operation of the receiver and detect any interference. The PLL demodulator also includes a fine tuning control. When the latter is set correctly and the demodulator locks on a valid NOAA HRPT signal, the audio frequency noise simply disappears.

The NOAA HRPT receiver PSK demodulator is built on a single-sided printed-circuit board with the dimensions of 60mmX60mm, as shown in figure 13. The corresponding component location is shown on figure 14. L1 has an inductance of about 0.3uH or 5 turns on a miniature TV IF transformer core.

The alignment of the PSK demodulator consists in tuning the VCO to the desired frequency. In practice this means setting the inductance of L1 so that locking occurs in the middle of the fine-tune control. The fine tune potentiometer should supply a voltage up to about 9V and this power supply should be well filtered and stabilised. On the other hand, a single-turn potentiometer (100kohm lin) is sufficient for the fine-tune function.

7. NOAA HRPT Receiver Monitor AF Amplifier

A simple audio-frequency amplifier is required as an AF monitor in a NOAA HRPT receiver. The circuit diagram is shown on figure 15. The TAA611 may be an obsolete type, but it has a low power drain and does not generate much interference to the other circuits.

The single-sided printed-circuit board measures 40mm x 60mm and is shown in figure 16. The corresponding component location is shown in figure 17. A practical value for the volume potentiometer is 100kohm log.

8. NOAA HRPT Receiver Bit-Rate Synchroniser

The PSK demodulator supplies a Manchester-encoded serial data stream at 665.4kbit/s. The main function of the bit-rate synchroniser is bit-clock recovery. In addition, the bit-rate synchroniser also removes the Manchester encoding and supplies the original NRZ data stream.

A Manchester-encoded signal always includes a signal-level transition in the middle of a data bit. Additional signal-level transitions may be present at the beginning and/or end of one bit period. All of these transitions may be used for bit-clock recovery.

The circuit diagram of the NOAA HRPT receiver bit-rate synchroniser is shown in figure 18. The bit clock recovery includes a transition detector with a delay line (74LS164 shift register) and an EXOR gate. The output of the transition detector is fed to a PLL with the VCXO operating at 16 times the bit rate (10.6464MHz).

The PLL includes two phase detectors, implemented with EXOR gates. The in-phase detector is used to indicate the lock condition while the quadrature detector feeds the feedback amplifier. The polarity of the obtained 665.4 kHz clock is still ambiguous, since the PLL may lock on the beginning/end transitions or on the middle-bit transitions. An additional circuit is therefore required to resolve the correct clock phase.

The Manchester decoding is performed by a simple EXOR operation between the encoded data and the synchronised, square-wave clock. Since there are two possible clock phases, two identical bit-conditioning circuits are required. The latter use one 74LS161 synchronous counter each as an integrator. Both integrators count for 15 CLK*16 periods and dump the result to a buffer in the 16th CLK*16 period. In addition, the result of the first integrator is intentionally delayed by one-half bit period with an additional buffer to be available at the same time as the result from the second integrator.

The final decision is simple: a count between 0 an 7 indicates a logical zero, while a count between 8 and 15 indicates a logical one. In addition, one can assume that a count between 0 and 3 and between 12 and 15 indicates a "good" bit, while a count between 4 and 11 indicates a corrupted bit. A corrupted bit may be generated by noise, but it may also indicate a wrong clock phase.

Corrupted bits from both bit conditioners are integrated the corresponding RC networks and then fed to the decision circuit with the 339 comparators. The final decision is stored in the RS flip-flop with the two 74LS02 gates. The output of this flip-flop drives the 74LS157 to select the the bit conditioner driven by the correct clock phase.

The 74LS157 also drives three LED’s. The green LED indicates the bit-rate lock. The yellow LED indicates the missing transitions at the beginning/end of single bits and therefore signals the Manchester mid-bit transitions. Finally, the red LED indicates the missing mid-bit transitions and is an early indication of poor signal quality or increasing bit-error rate.

Without a valid input signal or when just noise is present at the receiver input, the green LED is off and both red and yellow LED’s are on. When a signal is applied, the green LED will go on. As the signal strength increases and the signal-to-noise ratio improves, the red LED will slowly go off. The yellow LED should stay on all of the times. If the yellow LED goes off or starts flashing, there is something wrong with the satellite transmission like long sequences of logical ones or zeroes. During normal HRPT reception the yellow LED makes just perceptible intensity variations, synchronised to the 6Hz HRPT frame period.

Since the NOAA HRPT modulation format includes a residual carrier, the polarity of the demodulated data stream is not ambiguous. However, if the RF signal is converted to a different frequency, the polarity of the phase modulation will be reversed if the signal spectrum is reversed in the frequency-conversion process. The bit-rate synchroniser therefore includes a polarity switch right at the input of the circuit. This polarity switch is only needed if the downconversion scheme is changed. If the bit-rate synchroniser is only used in the described receiver, the above mentioned polarity switch should be left open all of the time!

The NOAA HRPT receiver bit-rate synchroniser is built on a double-sided printed-circuit board with the dimensions of 80mm x 100mm. The upper side is shown in figure 19 while the lower side is shown in figure 20. The corresponding component location is shown in figure 21.

The alignment of the bit-rate synchroniser should start by bringing the VCXO to the desired frequency range with a frequency counter. The VCXO is using a crystal with a higher nominal frequency of about 10.68 MHz, so that the pulling range with the BB109 varicap diode may be made wider with the aid of L1. The exact value of L1 depends on the crystal used and may be as large as 30uH (40 turns on a 10.7 MHz IF transformer core) to pull the crystal down to 10.4646 MHz. The function of L2 (1.2uH or 10 turns on a TV IF transformer core) is to prevent the oscillator to jump on the overtone crystal resonances.

Without any input signal one should first obtain 2.5V on the varicap diode by the corresponding 10kohm trimmer and then tune L1 for 10.4646 MHz. The bit-rate lock threshold trimmer (10kohm) should be set so that the green LED just goes off. When a valid NOAA HRPT signal is applied and the green LED goes on, one should finally check that the voltage on the varicap diode is close to 2.5V and eventually correct the setting of L1.

9. NOAA HRPT Receiver Aux Frame Synchroniser

The HRPT transmission includes the data originating from all of the sensors onboard a NOAA satellite (2), (3). The serial data is organised into words and frames. Words are 10 bits long and the MSB is transmitted first. One frame includes 11090 words or 110900 bits so that exactly 6 frames are transmitted in one second at a speed of 665.4kbit/s.

The frame rate is also synchronised to the main radiometer (AVHRR) mirror rotation, so that one frame contains the data of one AVHRR scan line. The AVHRR image data takes most of the frame and consists of 10240 words, starting with word 751 and ending with word 10990. The first five words of the AVHRR data correspond to the first pixel data in five spectral channels. The following five words correspond to the second pixel. Finally, the last five words of AVHRR data correspond to the last, 2048th pixel.

Frame synchronisation is achieved by detecting known patterns in the frame structure. In the NOAA HRPT frames there are two such patterns: a 6 words (60 bits) long frame sync at the beginning of a frame and a 100 words (1000 bits) long auxiliary frame sync the end of a frame. Both patterns are not arbitrary, but are generated by mathematical algorithms, known as binary polynomial division. Besides interesting mathematical properties, these patterns are also easy to generate with shift registers and EXOR gates.

However, in real receiving conditions, some bits may get corrupted due to a poor signal-to-noise ratio. This is especially true in an amateur receiving station with a small antenna. Of course, bit sync and frame sync should not be lost with moderate bit-error rates. Therefore the frame sync circuit should not just be able to detect the sync pattern but should also tolerate a certain amount of errors in the sync pattern. Simply speaking, the frame synchroniser should lock on a signal even if just an arbitrary part of the sync pattern is received correctly.

The circuit presented in figure 22 is designed to look for the auxiliary sync pattern in a NOAA HRPT transmission. The auxiliary sync pattern is generated by a 10th degree polynomial (10 stage shift register) X**10+X**5+X**2+X+1. The frame synchroniser includes an identical shift register with an identical EXOR network. The incoming data is fed to the shift register and compared with the locally computed result at the same time. If the two results match, the incoming sequence corresponds to the given polynomial.

If a match is found for 64 consecutive bits, the 4520 counter will signal that frame sync has been achieved. However, the exact position of the detected sync sequence inside the 1000 sync bits is not known yet. Therefore the 4011 gates switch the EXOR network back to the shift register input and the circuit is clocked on until the shift register reaches the all-ones state.

The all-ones event is detected by AND gates and is finally signalled out a valid frame-sync pulse. The frame pulse also triggers a timer (4017 and 4020) that inhibits the aux frame-sync circuits until the next expected aux frame-sync pattern. The timer output also drives the frame-sync LED.

The described auxiliary sync pattern detector is also sensitive to a long sequence of zeroes. Since the latter do not represent a valid sync signal, the circuit is inhibited by the 4029 counter as soon as 12 consecutive zeroes are detected. In a valid aux sync pattern there are at most 9 consecutive zeroes, since the 10-stage shift register pattern generator never goes through the all-zeroes state.

The described frame synchroniser checks 64 consecutive bits. Since the shift register needs to be filled with 10 valid sync bits first, a correct reception of 74 sync bits is required to trigger the described frame synchroniser. These 74 bits may occur anywhere in the available 1000 aux sync bits. Therefore the described frame synchroniser will lock reliably even at bit-error rates as poor as 10**-2.

The NOAA HRPT receiver aux frame synchroniser is built on a double-sided printed-circuit board with the dimensions of 60mm x 120mm. The upper side is shown in figure 23 while the lower side is shown in figure 24. The corresponding component location is shown in figure 25.

Being a completely digital circuit, the described aux frame synchroniser should not require any alignment. It is however necessary to understand its principle of operation in the case when troubleshooting is required.

Warning! The circuit includes a "difficult" component: the 4068 AND/NAND gates. These devices should be supplied by the companies RCA or SGS. The 4068 devices from other manufacturers are only NAND gates with the output on pin-13, while pin 1 is not connected. If the latter are used, two additional CMOS inverters are required to perform the AND function on pin-1. 74HC4068 devices were not tested yet, so these may be AND/NAND devices.

10. NOAA HRPT Receiver Construction Notes

In the above description, the operation and construction of the various parts of the NOAA HRPT receiver was discussed. When assembling the complete receiver, one should however notice that the overall gain is rather large and the signal levels range from the noise floor up to TTL levels.

To avoid harmful interaction it is recommended that the downconverter, local oscillator multiplier and IF amplifier are located in one shielded container. The PSK demodulator, bit synchroniser and frame synchroniser may be located in a separated shielded container, maybe together with the computer interface. More information about the installation and shielding of microstrip and other RF circuits can be found in (9), (10).

Before building the described circuits one should first check what kind of signals are required by the computer interface. If the latter requires the frame-sync pulse, some variable delay may be required using shift registers and/or counters. The polarity of the bit clock may also be reversed. Even if the receiver is not used with the "DSP computer", it is recommended to check (11) to understand the operation of NOAA HRPT computer interfaces in general.

Of course it is hoped that everyone is able to reproduce high-quality pictures like the one shown in figure 26.