Message-Id: (m0wgO4D-000HyUC@beacons.cts.com) From: km6sn@beacons.cts.com (Rod Davis) Subject: [HomeBrew] Spectrum analyser project Date: Mon, 23 Jun 1997 22:23:01 -0700 (PDT) Hello fellow Homebrewers, In the last year I built a spectrum analyzer, and this is a description of the architecture and some things I learned in the process. The project evolved in three pieces: (1) a front end converter with sweep oscillator, swept from 50 to approx 450 MHz, IF output in the 57 Mhz range (2) a tuneable RF voltmeter, with a DC voltage output which is linear in decibels, 90 db dynamic range. When used in the spectrum analyzer application, it is set to the 57 MHz front end output (3) an oscilloscope for viewing the results, used in X-Y mode with the DC control voltage from the front end as the X input, and the RF voltmeter as the Y input THE FRONT-END CONVERTOR I used a surplus cable TV tuner which is voltage tuned from about 50 MHz to about 450 Mhz. Some units are available which tune as high as 800 MHz. An analog DC ramp generator with adjustable slope and limits controls the tuning. A sample of the DC sweep voltage is buffered and presented at an output for the 'scope X input. THE RF VOLTMETER ----------------------------------------------------- Local Oscillator/Mixer section: I made a JFET tuneable oscillator which tuned 64-144 MHz. Using a 74AC74, I did both analog-to-digital level shifting and a divide by two. (Read the Motorola lit on 74AC74 - the clock line can be directly driven by analog slow-rise signals! ) This yields a tuneable 50% duty cycle digital level oscillator which can be applied to diode ring mixers via a coupling capacitor! Further divide-by-two stages produce LO ranges of 32-72, 16-36, 8-18, and 4-9 Mhz. When used with a 4MHz IF system, this provides converage from DC to 68 MHz. No attempt was made to eliminate unwanted mixer images because (1) in my HF applications images 8 mhz away are usually not an issue, and (2) in the spectrum analyzer application, the input signal bandwidth is constrained to less than 8 Mhz. IF section: The IF is at 4 MHz. I used a single stage tuned buffer, a W7ZOI type ladder crystal filter at 4Mhz (from an earlier SSB receiver project), and a Phillips NE604 IF system chip. The NE604 is an amazing part- it does the IF amplification, and produces a DC Received Signal Strength Indicator (RSSI) output which is linear in decibels with a 90 db dynamic range. That means that full scale is 90 db, half scale is 45 db, quarter scale is 22.5 db, etc. I DC buffered the output to protect the NE604. Use care in your layout - read the Phillips lit, and use an interstage filter (a simple W7ZOI crystal filter is enough, the objective is to limit the wideband noise output from the first stage). Be aware that the crystal filter must have )90 db stop-band attenuation, and use adequate shielding. I built mine in boxes made from PCB material soldered together, using "ugly" construction for the Xtal filter and front end, and thru-hole/pad-cutter construction for the NE604 section. I use a swicth selected crystal oscillator with 9,10,and 11 MHz crystals followed by a 74AC logic series buffer. This produces marker pulses will into the UHF region, and makes frequency measuring very easy. THERE ARE NO SCHEMATICS AVAILABLE FOR THIS PROJECT. IN FACT, THERE ARE NO SCHEMATICS HERE! My advice is to build the RF voltmeter first. A fixed tuned version would be fine for the spectrum analyzer. I wanted variable tuning throughout the HF region for other applications. The performance of the spectrum analyzer is excellent. After using one, I can't imagine how I got along without it. The next thing I will add is an up-convertor to shift the HF region up to the 50 Mhz lower limit of the front end. GL, and 73, Rod Davis KM6SN ---- Submissions: homebrew@qsl.net