Pioneer F-90

The Pioneer F-90 is a high-performance AM/FM tuner that features a balanced mixer, three narrow-IF filters, pulse-count detector, and a novel switched-sinewave stereo decoder. It has good sensitivity, excellent overload resistance, excellent immunity to spurious composite signals, very high AM suppression, very high ultimate S/N, and extremely low distortion.

The negatives: With its bright aluminum front panel, the F-90 does not blend well with black audio components. It has no signal-strength indicator. The IF bandwidth and mono/stereo settings are not saved along with frequency in the station memories. Two power transistors on the PCB generate enough heat to upset a careful alignment. The board must be unbolted and tilted for modification because there is no bottom access. Finally, the memory pushbuttons usually don't work because the foam between the front panel and switches has decomposed.

Circuit

The F-90 uses a tuned antenna transformer with center-tapped primary. This provides both 75Ω and 300Ω input impedances without a lossy balun. The transformer feeds a dual-gate MOSFET RF amplifier followed by two tuned circuits. This distribution of RF selectivity provides a low noise figure, as does forgoing AGC. The balanced mixer uses a pair of 2SK125 JFETs.

The mixer output is diode-switched to two IF paths. The wide path uses two ceramic filters and a tunable transformer. The narrow path uses three ceramic filters. Isolation amplifiers between all filters in both paths prevent undesirable interaction.

The IF signal is doubled to 21.4 MHz, heterodyned to 1.26 MHz, and detected by a pulse counter. This scheme greatly increases the modulation percentage to enhance S/N while providing very high linearity.

The detected pulses feed a unique stereo decoder. Described in patents here and here, the decoder phase-locks a 38-kHz sine wave to the 19-kHz pilot. The pulses gate the sine wave to the left and right channels (the multiplication circuits shown above are analog switches). Because the pulse rate is well beyond the spectrum of the composite stereo signal, no spurious mixing products result. In particular, the decoder is immune to HD Radio self-noise. The circuit includes pilot cancellation.

The FM circuit uses three proprietary Pioneer ICs. AM uses a Sanyo LA1247, which has somewhat better performance than the usual LA1245. The tuning range is 520 to 1610 kHz. AM and FM share the 75-µs deemphasis network. It provides a reasonable approximation to NRSC AM deemphasis, at least until the IF filter takes over. A 10-kHz notch filter suppresses adjacent-channel heterodynes. An RCA jack provides 450-kHz IF output for an external AM stereo decoder.

Modifications

The memory pushbuttons worked in none of the three F-90s I've restored. It's easy to replace the decomposed foam between the front panel and switches with something elastic, such as thin pieces of weatherstripping.

The power supply pass transistors mount on two heatsinks on the PCB. They generate enough heat to significantly raise the internal temperature with the enclosure in place since there are no vents. This can degrade a meticulous alignment done with the enclosure off. Experiments with a heat gun suggest that changes in ceramic filter response are the main problem.

In one F-90 I remoted the pass transistors to the rear panel. This dissipates much of the heat directly to the outside. You must electrically insulate the transistor tabs from the chassis since they connect to the collectors. With this arrangement the alignment still changed somewhat when the enclosure was replaced, but much less than before.

You can achieve a precise alignment without relocating the transistors by following this procedure: replace the enclosure, do something else while things warm up, remove the enclosure and quickly make an adjustment, and repeat. This procedure is tedious, but I found it necessary only for the wide distortion and narrow separation adjustments.

I've found ceramic filters with a frequency-code dot color other than red at F105, the second wide-IF filter, indicating that the center frequency is not exactly 10.7 MHz. R116, which determines the filter source resistance, is lower than the 330 ohms Murata recommends, as is the load resistance, the parallel combination of R137, R117, and R118. L102 and C141 resonate at 10.7 MHz, presumably to tame spurious filter responses, but L108 will alter this resonance and seems out of place. I don't know whether any of these circuit oddities influenced the center-frequency choice for F105, but replacing it with a red-dot filter improves the overall passband shape and weak-signal performance, and reduces distortion. I've always found a red-dot, low-group-delay, 250-kHz MX filter in the first wide slot.

I usually replace the narrow filters with two 110s and a 150. Three 110s provided terrific selectivity, but the audio occasionally clicked. I've also used three 150s, which yields lower harmonic distortion and 35 dB of adjacent-channel selectivity.

To improve stereo separation in narrow IF mode, the F-90 compensates the stereo decoding matrix by switching in R306, 68kΩ, with a JFET. If you replace the narrow filters, the compensation is no longer correct. You can increase narrow separation to beyond 60 dB by adding a 100kΩ trimpot across R306, outlined in red below. Adjust it for maximum separation in narrow after making the wide separation adjustments. Make sure the signal generator is exactly on frequency.

The AM audio response rolls off at high and low frequencies. You can extend the high end by removing C217, 0.047 µF, outlined in yellow above. I clip one lead and tilt the capacitor back as shown. You can improve the bass response by adding 0.47 µF across C356, 0.047 µF, as outlined in green above. Note the polarity. These changes noticeably improve AM sound quality.

Alignment

I use L instead of L+R for the stereo test signal. I tune T101 as well as T2 for minimum distortion in wide-IF mode. I turn VR101 all the way up to maximize weak-signal limiting in wide-IF mode. I turn VR501 all the way down. The two unmarked transformers shown to the left of TP17 are in the IF doubler circuit. A couple inches to their left is a 1kΩ resistor. Adjust the transformers to peak the 1.26-MHz signal at either end of the resistor. Just to the right of TP14 is an unmarked coil. Adjust it to maximize the 38-kHz sine wave at TP14. The AM alignment instructions neglect IF filter F201. Adjust it for minimum audio output for a low-deviation FM test signal exactly on frequency. This will minimize residual digital noise when receiving an AM HD Radio signal.

Measurements

For the following measurements I used IEEE 185-1975, modified as noted here. I used the test equipment listed here. The tuner had 110-110-150 narrow filters.

Mono 30-dB quieting sensitivity      W 14.8 dBf, N 11.8 dBf
Mono 50-dB quieting sensitivity      W 18.8 dBf, N 18.8 dBf
Stereo 50-dB quieting sensitivity    W 39.4 dBf, N 37.8 dBf
Adjacent-channel selectivity         N 46 dB
Stereo THD, 1 kHz, L or R            W 0.02%, N 0.9%
Stereo separation, 1 kHz             W 56 dB, N 50 dB
AM suppression ratio                 W 82 dB, N 78 dB
RF intermod                          W 88 dBf  (97.7 + 98.5 -> 96.9)
RF spur                              W 111 dBf (95.77 -> 96.9)
Image rejection                      W 84 dB   (118.3 -> 96.9)

30-dB quieting sensitivity is somewhat like IEEE usable sensitivity, except that I did not modulate the signal while measuring 3% residual noise plus distortion. The background noise does increase with modulation and this would affect a true usable sensitivity measurement.

RF intermod is the 50-dB quieting level for a third-order intermodulation product. RF spur is the 50-dB quieting level for an untuned signal. Both measurements are made in a way that sidesteps the phase-noise sidebands of the tuner and signal generators.

More is here.

Updated May 11, 2008