Sony TAN-8550 VFET Amplifier Maintenance

This Sony TAN-8550 VFET Stereo Amplifier came through the shop recently for some work. The owner reported that while it was working and sounding good, it still had the original “death diodes” installed which are known to degrade with age, and aren’t safe to leave installed even if they’re working at the moment.

The Sony TA- and TAN- line of VFET amplifiers are uncommon, very interesting pieces of equipment. They use VFET output transistors, a type of depletion-mode output FET which is unlike any other manufactured. Sony used them for a couple of years in their top of the line amplifiers, and Yamaha in a few rare and expensive pieces, but that’s about it. They’re easily damaged by loss of bias, such as provided by the VD-1221 varactor diodes known as “death diodes”. In addition, VFET equipment must never be powered up slowly using a variac or the expensive output devices will be destroyed and you’ll be left with a doorstop, as there are no replacements for burnt-out VFETs except to harvest them from another amplifier.

The TAN-8550 produces 110W into an 8 Ohm load, and features both A and B speaker connections along with a direct speaker connection which bypasses the output switch. Speaker “B” has a separate level control, as well.

It’s a nice, open layout inside for the most part although, good for cooling. The main power supply capacitors in this unit are an uncommon Nichicon dual capacitor, seen here, which will be a challenge to replace as that style is no longer manufactured. Fortunately, the owner requested only a limited service at this time and these capacitors are good, so no replacement required.

Remove a few connectors and the chimneys and the power amplifier modules slide right out. The right channel must have suffered a catastrophic failure at some point in the distant past, as several of the VFETs on the bottom and several of the transistors on the top were not original, and a large power-handling trace on the back of the circuit board looks like it burnt off and was replaced with a thick piece of copper wire jumping across the break. The shop did good work, and based on the age and availability of the components, this probably happened not long after the unit was new.

Each power amplifier module has 3 of the multi-junction varactor diodes, replaced with pairs of new diodes in series. Testing showed the old and new were matched on Vf to within about 10 mV.

In addition to the diodes on the amplifier modules, there is a diode in the power supply, and a diode on the Class A Amplifier board, the driver board for the power amp.

Next up were assorted adjustments for bias, DC offset, indicator lamp voltage, and output meter calibration. Bias on this amplifier was considerably off, 9 mV on one channel and 18 mV on the other, vs. a spec of 125 mV. This resulted in a cool-running amplifier, but would have limited the output power and low volume distortion performance.

Bias reset to about 100 mV, a little lower for a safety margin but still delivering enough power for good performance. This is later confirmed by the performance measurements.

DC offset was good, but tweaked up slightly down to 3/6 mV. Zero is the goal, but under 20 mV is generally acceptable.

Lamp voltage had drifted a bit high, up to 8.2V and was reset back to 8V.

Finally, the power meter was uneven; it was reset to match the output power. It’s not a precision indicator, but tracks more closely now. Finally, time to test the amplifier and see if it needs any further adjustments.

The amplifier meets frequency response specifications, 20 Hz – 100 kHz +0/-3 dB and channels are extremely closely matched.

The unit exceeds its distortion specification easily.

Performance is great at high output, too, although somewhat less closely matched. This could be due to the original electrolytic capacitors losing their performance, but is only noticeable with sensitive measurement equipment. As both channels exceed the published specification, this is not a problem right now, although we recommend the owner monitor for any changes to the sound quality which could signal a need for further service on the remaining components.

The amplifier played into a dummy load at 50W for an hour, and then powered off ready to go home safe with new diodes installed.