One of my first Hoppe’s Brain customers sent me this beautiful old preamp to work on. Eight years ago I had refurbished for him, a pair of Adcom GFA-565’s to drive a pair of power-hungry Eminent Technology EFT-8 planar speakers. He loved my work, and sent me this preamp to restore.

Spectral Audio is a Silicon Valley company, started in the late 1970’s. Indeed the design itself is very reminiscent of high-end test equipment from HP or Tektronix.

Where to start? Well, it’s reportedly still working, so let’s whack it up on the HP8903B Audio Analyzer and see what it does.

Left channel: 1KHz, 1V in, 1V out, 0.005% THD! Nice.

Right channel 0.86%?? Ope! That can’t be right.

Let’s pop the cover.

Dang, this thing is pretty! No expense spared on anything. Top quality parts. Teflon wire. Two-layer board with plated through-holes. (Expensive in 1981) I love the layout. Everything is on one board, and the way the board just slides out of the chassis makes it super easy to work on. The front panel is connected by gold-plated edge connectors.

First of all, I can see someone has been working on this before… The LM317 and LM337 voltage regulators have been replaced in the left channel; they don’t match the ones in the right channel.

Someone has been in here and done a competent repair.

There are some dried up caps here, so before I do much troubleshooting, I’m going to just go ahead and re-cap the whole thing. I look up the datasheet for the original Illinois Capacitor RMR series, and considering their specs, and their role in the circuit, I decide Panasonic FC is an excellent modern replacement.

But the voltage rails are still sagging on the right channel. They’re +14 and -18V when they should both be 24V. Immediately I suspect the voltage regulators, and sure enough, there is only 0.8V at the LM337 and LM317 adjust terminals, when it should be the reference voltage of 1.2V.

I replace the voltage regulators and the voltage rails come good to 24V!

So now it’s basically working, but all the adjustments are out of whack. The right channel’s bias is way too high, the left channel’s bias is too low, and the gain adjustment is set to give only 5db gain at maximum volume. The spec is 10db and so that’s where I set it. Bias current spec is 20mA, set by adjusting for 2V across the 100 ohm resistor that supplies current from V+ to the output FETs. The DC offset was already pretty well adjusted, but I gave it a slight tweak.

Here’s the schematic for reference. Pretty neat how the phono preamp and line-level stages are pretty much the same circuit.

I added some 0.047µF polypropylene bypass caps across all the electrolytic local supply bypass caps. (The white caps are a factory bodge.)

On to re-capping the protection circuit and… WHAT’S THIS hiding beneath this capacitor?

That stray solder-blob might have been causing some problems! It looks like it squirted up through the board during wave-soldering and melted the capacitor’s insulation! It might have been touching the aluminum capacitor body, or the ground plane, perhaps intermittently. This trace is connected to -24V. That could have been why the regulators fried in the first place.

Out with the electrolytics! I found some film caps that fit perfectly in place of the original axial-lead electrolytics. They’ll do the job and last pretty much forever.

My only critique of the board design: These electrolytic capacitors are located too close to the heatsinks on the transistors, which run really hot, so the capacitors are cooking for no reason. They’re not under a lot of electrical stress, so they should still last a decently long time, but moving them just a couple millimeters would run them much cooler.

That’s it! Working beautifully. More pix…