The Technics ST-S7 receiver was introduced about December of 1979, and was top-line equipment back then. Due to an unfortunate power problem at my house, my ST-S7 was "ruined", in that the power transformer (primary) was burned out, and worse, the vacuum florescent display was also destroyed (burned out filament).

When I realized that the VFD was dead, I was quite dismayed. Parts have been unavailable for decades, so I can't buy a new one. I looked online for generic replacement VFDs, to no avail. Then I wondered if I could make an LED replacement for the VFD. The service manual is available, with a full schematic, even including voltage readings.

Perusing the schematic showed me that the display was "normal", in that it was multiplexed, with a select line for each digit, and 7 select lines for the 7 segments of the digits. The microprocessor is normal, too, running on 5 volts. This might be do-able.

Next step was to hack in a replacement for the toasted power transformer. The stock transformer has three windings: 30v, 9v, and 3.2v for the VFD filament. We can forget that last one. So I cobbled up two transformers from my collection, and with 30v and 9v applied, the unit "worked", in that it responded to button presses (and would have probably made audio if I had hooked it to an amplifier).

Studying the schematic, I see that the digit selects are active low, and the segment drive is active high. I dug out some common cathode 7 segment displays from my parts collection, and with some experimenting with bread boards later, I had a functioning LED display.

The digit selects run through the transistors in a ULN2081, which then pull the cathodes to ground as each digit select goes high. The segment drives connect to the gates of P-channel MOSFETS, which then supply 5V via some (initially 220 ohm) resistors to each segment.

Many hours of fiddling and experimenting later, I figured I had the circuit dialed in. I fired up KiCad, drew the circuit, and then generated PCB artwork. I've uploaded this to JLCPCB and am awaiting the boards now (August 16 2026).

While I wait for shipping from China, it's time to work on the power supply.

Power Supply Repairs and Modifications

The original transformer had three secondary windings, of which we now need two. With the tuner running, I can measure the current required from each different winding. It turns out that the 30v winding only needs to supply 15mA max, and the 9v winding needs to supply 110mA when the unit is "off" (for the microprocessor and digital clock), and 360mA when it's on.

This means that it's wrong to try to get one replacement transformer to supply both voltages, as the replacement will not have one voltage at a much different output current than another voltage (winding or tap). They'll all be the same, which means that the 30v winding will be grossly overbuilt. This makes the transformer much to big to fit in the tuner case.

So the answer is to use two transformers, one that is good for the 9V and 360mA, and another one that is good for 30v at 15mA.

This worked out well, and some eBay shopping later, I found my two transformers. Fortunately, there's enough empty space in the tuner case to easily hold the two transformers.

One new power supply requirement is for more 5V current, since the LEDs need much more power than the VFD. I measure about 70mA draw from the 5V supply just for the LEDs. I'm afraid to pull this from the stock power supply, because it's not designed for that additional load.

Fortunately, the transformer I found for the 9V AC supply has duplicate windings, and each one is good for more than 400mA. So I built an additional 5V source, consisting of a diode brige, filter cap, and 7805 regulator, that is dedicated for the LED display.

One note: I intentionally skimped on the filter capacitor on the new 5V supply. This "small" cap (47uF) will mean there's a lot of ripple (about 5V peak to peak) in the unregulated DC. I want this, because this lowers the heat that the 7805 will generate, because it'll be dropping a lower average input voltage to the 5V output.

The (New) Circuit Board

There isn't room in the space occupied by the original display to hold all the support circuitry along with the LED displays, so there will have to be two boards. I designed the display board to be the same size as the original display, and the other board was made big enough to hold the rest of (most of) the driver circuitry.

To make this easier to build and debug, I created one "big" pcb with both boards' circuitry on it, but segregated so that it can be cut in two when ready to be installed. After cutting the board in two, ribbon cables will be used to re-connect the circuits.

The Colon and Decimal Point

The stock display is "funny", in that the colon for time display occupies one whole digit width, and the decimal point in FM mode also occupies one whole digit width. (Note later discussion below about the two different decimal points.)

In order to replicate this odd spacing, I glued some tiny LEDs on top of the third and fourth digits, and ran 30ga wires from those LEDs to the PCB. I tried to bend the wires to conform to the gaps between digit segments, so the digits would look normal when the colon or decimal point were not illuminted.

The decimal point deserves special mention. There are actually two different decimal points used on the display. The one I use is the one that appears when the 200KHz/10KHz switch is in 200KHz mode (i.e. US FM). This decimal point occupies the entire digit 4 space.

The other decimal point appears when the switch is in 10KHz mode (European FM). This decimal is the one built into digit 3. My design initially ignored this because I didn't understand how the decimal points worked, but eventually, I figured this out and modified my design to show the two decimals correctly.

Locating and Mounting the Display

The display part of the PCB is the same size as the original VFDisplay. There is a piece of foam glued to the steel front support, and mine had a compressed region where the original display was pressed into it. This gave a vertical position for the display. The right edge of the display aligns with the edge of the folded down part of that steel support.

I cut a second piece of foam and used E6000 to glue that to the steel support on the left side, where it will be behind the left-most seven-segment display. The thickness of this foam should be about 1/8 inch (3mm). I also had to remove the foam on the right side, and slice off half of it to get it to the same thickness (and then of course glue it back on).

I also had to cut the rounded top off of the "Stereo" LED (blue in my case) because it was a bit too tall to fit behind the front panel.

Finally, glue everything up with E6000 and clamp it in place overnight.

Other Fixes: Quartz Lock Lamp

The "Quartz lock" indicator is an incandescent lamp, which of course on a nearly 40 year old piece of equipment, is burned out. I replaced it with an LED, and some experimentation suggested that a yellow LED sort of reproduced the off-white color of the original lamp.

The problem with the LED Quartz Lock indicator is that because the LED was much more sensitive than the lamp, it glowed dimly at all times. The solution was to put a 5V Zener diode in series with the LED (and of course a current limiting resistor). The resistor value that worked for me was 100Ω. The open circuit voltage of this indicator circuit is 10V.

Other Fixes: Battery Backup

The battery backup for the clock and radio station presets is three 1.5v AA batteries. Of course, if you leave batteries in there for several years, they leak and corrode the battery box terminals.

A lithium-ion battery is about 4v when charged, so it will work for the backup battery. I put an old vape battery (a 13400) and a USB Lithium battery charger in the battery box.

The battery charger is powered from the new 5V supply. I tested the charger/BMS, and find that it stops supplying power to the load when the battery voltage drops below 2.5v. I hope that is sufficient to protect the battery from damage.

I have measured the drain current from the battery when the unit is unplugged at 600µA. I measured the battery capacity at 350mAH, so this should be good for in excess of 300 hours. Self discharge might cause the voltage to drop faster than the load current does, but I didn't test that. Discharge test below.

Schematic

Gerber Files

Installation Instructions

• Populate the PCB. Do not cut it in half yet.
• Remove the VFD.
• Remove IC902, IC903, and resistor pack Z901.
• Install 8 pin socket in pins 1-8 of IC903.
• Install 7 pin socket in pins 2-8 of Z901.
• Rig up a new 5V supply capable of 100mA.
• Make up 7 pin cable to connect J1 to Z901 socket.
• Make up 8 pin cable to connect J2 to IC903 socket.
• Make up 4 pin cable to connect the following to J3:
  1. Ground from new 5V supply.
  2. Stereo indicator from Q910 collector.
  3. 5V from new 5V supply.
  4. Digit 6 drive from IC902 pin 12.
• Plug the display PCB into the three cables, and test operation.
• Once the board works and you've adjusted resistors as needed to get the LED brightness right, cut it in half. In particular, you may need to adjust R1, R2, R3, R4, R8, R9, and RN1. (This adjustment needed because your LEDs are not the same efficiency as my LEDs).
• Make up two cables to connect the halves. The cables must be able to be disconnected so you can pass the cable under the steel front panel support bracket. Make sure the connectors are thin enough to pass under the steel bracket.
• Make/modify the foam spacers as discussed in the text.
• Glue the display board to the foam and that assembly to the steel bracket.
• Insulate the back of the display driver board. I used Kapton tape, but any insulator would work. I crammed the driver board upside down behind the display.

Service Manual

FAQ

Q. Will you fix mine?

A. No, I'm retired and so not interested in a job.

Disclaimer/Warning

This is just my documentation for my repair. I don't claim that doing this is safe or recommended.

Soldering irons are dangerous, be careful. Oh, and don't eat the solder.

William Dudley
June 27, 2026
August 27, 2026