The Technics ST-9030 is a classic analog FM tuner in a heavy, rack-mount enclosure. With its eight-gang tuning capacitor and balanced mixer, the front-end is nearly bulletproof. The front panel features a long linear dial with flywheel tuning, signal-strength meter, and tuning meter. Automatic circuits with manual override control IF bandwidth, stereo hi-blend, servo tuning, and interchannel muting. The tuner has variable-level audio output, fixed-level ultrasonically filtered output, and multipath oscilloscope outputs. Silkscreened on the top cover are three performance graphs and a block diagram.
The front-end has six tuned circuits in the signal path. This yields a narrow RF passband with steep skirts that greatly attenuate untuned signals. Rated noise figure for the 3SK40 at 200 MHz is 3 dB typical. The balanced mixer reduces third-order intermod (98.5 + 98.5 − 98.9 → 98.1) by substantially cancelling the necessary second-order product (98.5 + 98.5 → 197). Forgoing AGC maintains sensitivity when there are strong untuned signals within the RF passband. Unusual for a 1970s tuner, there is no 300Ω antenna input. Technics supplied an external 75:300Ω balun.
Separate IF strips for wide and narrow bandwidths include individual ratio detectors. For some reason the detectors are designed to yield inverted signals. The wide IF uses a linear-phase Sumida Electric LC filter. Narrow uses four 230-kHz Murata ceramic filters cascaded in pairs. Narrow drives the meters, servo tuning, muting, and vertical multipath output. Wide drives the automatic IF bandwidth circuit. A welcome adjustment not often found equalizes wide and narrow audio levels.
The stereo decoder uses a Panasonic AN363 with external pilot cancellation, adjustable pilot phase compensation, and individual wide and narrow stereo separation adjustments. The AN363 channel balance spec is rather loose at 1 dB maximum. The ST-9030 provides an adjustment to equalize L and R levels, although equalizing them in stereo yielded a 0.7-dB channel imbalance in mono.
The muting circuit combines signals from the ratio detector, IF chip, and a double-tuned, IF resonant circuit. According to the service manual, the latter helps with “malfunction arising from neighboring stations.”
Servo tuning is a fancy name for automatic frequency control (AFC). It adjusts the voltage to a varicap diode in the local oscillator to keep the signal centered in the IF passband. Changing frequency temporarily disables the control loop to let you manually fine-tune another station. After several seconds it automatically reengages. A pushbutton defeats both servo tuning and interchannel muting.
This compares the variable and fixed outputs at 10 dB/div for a nonpreemphasized test tone swept to 50 kHz. The notch is at 19 kHz.
The audio circuit uses emitter followers and an NJM4558 dual op-amp with a typical gain-bandwidth product of 3 MHz and slew rate of 1 V/µs. A 10kΩ dual pot drives the variable outputs, while a passive ultrasonic filter drives the fixed output.
The power supply has an AC line choke, fuses in the transformer primary and secondary, and voltage doublers driving discrete regulator circuits. Rated power consumption is 27 W.
The AN363 stereo decoder uses a squarewave demodulator that is susceptible to HD Radio self-noise. I installed the postdetection filter shown above directly on the PCB. It eliminated an annoying hiss in stereo on HD Radio signals.
In wide-IF mode I noticed hiss on several strong analog signals. Technics specifies alternate-channel selectivity in wide at just 25 dB (I measured 23 dB). With a figure that low you could reasonably expect some interference from strong alternate-channel signals. But adding a 250-kHz ceramic filter between the IF stages did not help. This turned out to be due to a design problem in the IF strip. The LC filter is located after the first µPC577 gain stage. Untuned signals can intermodulate within the µPC577 before the filter has a chance to attenuate them. I corrected the problem by placing the ceramic filter before the stage, driving it from emitter follower TR201. I changed the emitter resistor to 1.5kΩ and connected it to −12 V to prevent any clipping prior to the filter.
To provide better selectivity in the crowded FM band at my location, I replaced the narrow filters with two 110s and a 150. The first 110 precedes the first narrow-IF gain stage and the second follows it, replacing C203. Separating the filters instead of cascading them as in the original circuit lowers the IF noise figure and provides purely resistive source and load impedances. I added resistors across R206, R207, R209 and R210 to make all impedances 330Ω. The narrow passband yields a fair amount of modulation-induced noise at low signal levels, but it provides excellent adjacent-channel rejection.
With a 7-dB preamp and an antenna system with 7-dBd gain aimed toward strong local signals, I noticed a small amount of crossmodulation on one station. To reproduce the effect, I combined an unmodulated tuned signal and a signal 400 to 600 kHz away 100%-modulated with 1 kHz. The image above shows the spectrum of the ST-9030 local oscillator at 2 kHz/div with the modulated signal present and absent. The untuned signal FMed the LO when its level was between 100 and 120 dBf. Even though the LO has a tuned buffer, extremely strong RF somehow works its way in. The crossmodulation was 30 dB down worst case, though mostly much lower. Applying RF AGC to G2 of the second RF amplifier lowered the worst case to −60 dB. However, the required AGC threshold was about 70 dBf, which would have restricted ultimate S/N and promoted strong-signal desensing within the RF passband. I removed the AGC.
Unlike the first RF amp, G2 of the second is not bypassed. The PCB has pads and holes for a capacitor but no silkscreen marking. Bypassing G2 increased front-end gain 4 dB. This reduced modulation-induced noise somewhat in narrow and did not unduly aggravate the crossmodulation problem, which did not increase at all at some interfering frequencies and levels.
Because the mixer input transformer is not centertapped, gate and stray capacitances determine the two MOSFET signal amplitudes. They were quite different in my tuner. I was able to equalize them with a trimmer capacitor on one gate, but intermod performance did not improve so I removed the trimmer.
I replaced two of the automatic circuits with manual controls. The tuner provided no way to force monophonic reception so I turned the hi-blend auto/off pushbutton into a stereo/mono switch. I changed the auto/wide IF bandwidth pushbutton to a hard narrow/wide selection.
The tuner implemented hi-blend in an awkward way that required different deemphasis capacitor values for the two channels. After disabling hi-blend, I equalized them.
Even with the muting threshold set above 40 dBf, some interstation noise and grunge remained unmuted in my signal environment. Lifting one lead of R416 helped by narrowing the allowed mistuning range.
I noticed that the center-tuning meter had a small dead band. I was surprised to find that it was intentional, caused by diodes in series with the meter. I wanted to see small tuning errors so I shorted the diodes. What seems like a sliver of dead band remains. Actually it is a tiny bit of AFC, always enabled. I rather like the effect, which reminds me of a detented control, so I left it. I did add 22kΩ across R508 to increase the meter sensitivity.
I added resistors across the stereo separation trimpots to make adjustment less touchy. The postdetection filter reduces L−R and requires a correspondingly smaller L+R stereo matrix injection. A JFET engages the narrow trimpot. With the reduced circuit impedance, any variation in FET channel resistance would more markedly degrade stereo separation. To prevent this, I wired an unused section of the wide/narrow switch across the JFET.
I replaced the wide-IF and stereo indicator lamps with red LEDs with series resistors. 6.3-V, 250-mA, fuse-style lamps illuminate the tuning dial and meters. Replacement lamps are available.
Alignment instructions are here, here, and here.
Rather than using the front-end alignment procedure, I swept it. This yields precise tracking of the many tuned circuits.
The front-end has a metal cover that must be removed to align it. I noticed a slight frequency shift and a small change in distortion when I removed the cover, but not enough to bother drilling access holes. Adjusting the mixer transformer slug slightly too far clockwise compensated for the distortion change.
A block diagram is here. Schematics are here, here, and here.
For the following measurements I used IEEE 185-1975, updated as described here. I used the test equipment listed here. The tuner had all of the modifications described above, plus a few others that I subsequently removed.
50-dB quieting sensitivity, mono W 18.0 dBf, N 18.8 dBf 50-dB quieting sensitivity, stereo W 40.6 dBf, N 40.3 dBf Selectivity W 47.5 dB, N 46 dB THD, 1 kHz, stereo W 0.02%, N 1.1% Stereo separation, 1 kHz W 43 dB, N 42 dB AM suppression ratio W 73 dB, N 67 dB S/N, 65 dBf, mono W 79 dB S/N, 65 dBf, stereo W 74.5 dB S/N, 85 dBf, stereo W 81 dB RF intermod 99 dBf RF spur > 130 dBf RF image > 130 dBf RF mismatch loss 0.0-0.3 dB Noise figure, 96.9 MHz 7.3 dB Modulation acceptance W 200%, N 165% Minimum stereo pilot injection W 4.7%, N 4.7% Treble response, variable @ full W +0.0/-0.5 dB Treble response, variable @ half W +0.0/-0.9 dB Treble response, fixed W +0.0/-1.2 dB Bass response, -1 dB 28 Hz Output level, variable 1.3 V max Output level, fixed 0.6 V Output impedance, variable 2.5kΩ max Output impedance, fixed, 15 kHz 2.1kΩ
Although sensitivity is decent, the figures are several dB shy of the best I've seen. But its exceptional front-end makes the tuner a perfect candidate for an RF preamplifier. A low-noise, high-intercept, 10-dB preamp will greatly improve sensitivity while reducing intermod performance to merely excellent. Even an ordinary cable-TV preamp improved mono 50-dB quieting sensitivity 2.7 dB.
Wide selectivity is for the alternate channel and narrow is for the adjacent.