The ST-40: A Scratch-Built Tube Stereo

Update: this project was featured on Hackaday!

In the summer of 2014 I decided to build a vacuum tube stereo from scratch.

If you're unfamiliar with vacuum tubes, they are electronic devices that preceded transistors. They were once used in everything from computers to spacecraft, but today they survive almost exclusively in audio amplifiers. Transistors are better than vacuum tubes in almost every way: they are smaller, faster, cheaper, and more efficient. However, some folks believe vacuum tubes produce higher quality sound than transistors. I don't know if this is true or not, but I do love how tubes sound. I also think the way they work (using principles discovered by Thomas Edison while working on the light bulb) is fascinating—and of course they look incredibly cool.

I had tried for a long time to find a Dynaco ST-70, the most popular tube stereo ever, to restore. But these are expensive, even in disrepair. A friend's generous dad had given me non-functional Heathkit AA-100, but I wasn't in love with the look of the amp (the tubes are hidden under the chassis). It also doesn't have a reputation for great sound. So I decided to use the tubes and transformers (the expensive stuff) from the AA-100 to build a clone of an ST-70. A google search revealed a DIY amp builder named Tom McNally who had done the same thing in 2005

The first thing I did was draw up a schematic. There is a wealth of information and parts online for ST-70s. Tom McNally had used a PCB from Parks Audio called the DIYTube ST-70 Driver for the input stage. I did as well. The output stage was taken directly from the Heathkit AA-100 schematic. For the power supply, I decided to make my own PCB based on several designs for ST-70 kits I found online. For this I got some tips from Tom and the awesome folks at Audiokarma.org (a huge thank-you to dgillespie and kward). I simulated several sections of the circuit using LTSpice and PSUD2. The complete schematic is linked below.

Next thing I tackled was the chassis. I knew I wanted some combo of wood and metal. I found photos  of several amps online that served as inspiration:

Using these as a starting point, I used Google Sketchup to create a model of the chassis. I made it roughly the same width and length as the ST-70, but added to the height so I could put a meter on the front. I decided to mount the output transformers on brackets instead of inset in the baseplate (like they were in the AA-100) to simplify the metalwork and free up space inside the chassis. Here are photos of the completed model (link below):

I am not a woodworker, but I happen to have a father-in-law who is an amazing one. He graciously agreed to build the wood part of the chassis. I sent him drawings; we made a few adjustments, and just a few weeks later he emailed me this photo:

Awesome, right? The wood is cherry. You can see how the finish darkens and deepens over time by comparing to the photos at the bottom of this post. The bottom panel is removable; I decided to remove it to aid convection.

In the meantime I removed the transformers and fuse holder from my AA-100:

I discovered that the transformers were rusted and covered in wax, so I sanded and painted the cores with Rustoleum:

Next I ordered all the parts (a BOM is linked below), including three 1/8" aluminum sheets cut to size for the top, back, and front from onlinemetals.com. I then began the metalwork. Incidentally, this was the most painful part of the job, and if I had to do it again would spring for custom drilling. But I learned a lot! I was fortunate to be able to use the shop at my office. First I laid out the parts:

Then I drilled and punched:

(I am super-proud of this vintage meter I got on eBay. It was made in the USSR. I removed the rectifier so I could use it with DC voltage instead of AC.)

I wasn't happy with this look. The stock aluminum finish wasn't even and it was impossible to prevent a few dents and scratches. I found a tutorial online for brushing it by hand. It's a simple process: you brush the Al with 220 grit sandpaper using a guide block to make sure all the strokes go in the same direction. Then you brush it again with Scotchbrite to soften it. I was happy with the result:

One other note: the vent holes around the tubes didn't look great after drilling and using a deburr tool. A mechanical engineer coworker suggested I pretty them up using a counter sink bit on our mill, setting the stop so I went to the same depth on each one. I took his advice and I thought the result looked pretty great:

Next was the fun part: wiring it up. Here is a photo of the PCB I designed for the power supply. I ordered it from dirtypcbs.com, which was the cheapest I could find despite getting 10 boards when I only needed one.

Here is a photo of the diytube ST-70 driver board:

First thing I wired was the tube heaters. Afterwards I powered up to make sure the heater voltage was right and everything glowed as expected. I was pretty excited when it did.

I continued wiring:

(The big power choke in the photo above came out of an amp I pulled out of a garbage dump when I was 15. Pretty proud of that. I used an LCR meter and a hi-pot tester to characterize it.)

(Warning: the next 3 paragraphs are pretty technical—if you're unfamiliar with circuits you may want to skip.)

I wired and then powered up the amp in stages using high-watt resistors as dummy loads. I started with the power supply, added the output tubes, and then added the driver board. In the course of this process I changed the operating point of the output tubes. I had originally used the same values used in the AA-100, but I decided this was too close to the max power dissipation. With assistance from the good folks at Audiokarma I redrew the loadline to operate the plates at 430V and the screens at 325V with my 7.7K transformers. I sacrificed some output power doing this but I think it was worth it for sonic improvement and longevity. Here is a link to my loadline drawing.

The new operating point required adding a resistor in the power supply immediately after the rectifier tube in order to drop the voltage to 430 (big thanks to the good folks at dcaudiodiy.com who pointed out that I could use a lower-value resistor by placing it before—instead of after—the first capacitor bank). I originally tried to drop the plate voltage by adding zener diodes in series with the secondary ground on the power transformer. I neglected that this would change my bias voltage and I ended up smoking a capacitor. It's not a real build if you don't smoke something, right?

With everything wired it was time to start testing. Using 8 ohm dummy load resistors I measured that the maximum output per channel before clipping was 20W (thus, the name "ST-40"). I measured frequency response from 20 to 20K Hz and found it was within a dB across the band. I checked for hum with my scope but couldn't find any. I put a square wave on the input and tweaked the value of an RC filter in the driver stage to reduce ringing. I then tested a variety of different capacitor loads while varying the input frequency to check for instability. Unfortunately I found that when unloaded, the amp would go into oscillation. I asked dcaudiodiy.com for some suggestions and one of the guys there graciously invited me over to have a look at the problem! We ended up reducing the gain of the first stage and increasing the shunt capacitance of the feedback network. The amp was cured.

At this point I completely assembled the amp. I added a selector switch on the back so that the meter can measure the bias points on all 4 output tubes as well as the 300V power rail.

Finally, nearly 2 years after I started the project, I took it home and fired it up with my loudspeakers. I poured an old fashioned and put on "Attaboy" from the Goat Rodeo Sessions. It sounded just awesome.

Come by some time and we'll play some records!

Standard disclaimer: tube circuits contain lethal voltages! Don't try to build this stuff if you don't know what you're doing!
ST-40 schematic
ST-40 chassis model
ST-40 PSUD2 power supply simulation

Valve Amplifiers, Morgan Jones
Building Valve Amplifiers, Morgan Jones


Marantz 1060 Restoration

When I was a teenager my dad lent me his Marantz 1060 amplifier. The 1060 was the centerpiece of his first stereo system, purchased in 1971 when he was 20 years old. I used it to play the Beatles' "Revolver" over and over until one day the amp quit. I can't remember what the symptoms were, but I know that shortly thereafter my curiosity got the best of me and I tore the amp apart. I robbed many of its guts to make a speaker switch that I didn't really need. A short while later I received a plastic boombox for my birthday, and what was left of the 1060 was packed away.

Ten plus years and one engineering degree later, I came across the 1060 rotting in my parent's basement. Here's a photo of what it looked like (the cover and faceplate are missing, with parts piled inside):

I decided it was time to undo some of the destructiveness of my youth and restore this classic to its original condition.

When I got the amp home, I took a few more photos to show what shape it was in. The mic/aux jacks, headphone jack, speaker terminals, left quad switch circuit board, input RCA jacks, tape RCA jacks, and lamp had been removed. The fuse holder, four of the switch caps, and the blue lamp jewel were all missing.

The worst part was I had completely destroyed one of the quad switch circuit boards as a teenager:

Fortunately I found a very generous eBay seller who sold me the circuit board, plus the fuse holder and the lamp jewel, for $25. I was able to find the service manual online, so getting everything back in was straightforward. The lamp was burned out so I replaced it with an LED. The hardest part was figuring out how the aux/mic jacks connected back to the input selector switch. Here are some shots after replacing all of the removed items:

The next step was a full recap. With any vintage piece of a equipment, it's always a good idea to replace all of the electrolytic capacitors which can begin to degrade quickly after 30 years or so. It's a good idea to do this first before any further troubleshooting, in case there are leaky caps causing problems. I tried to use replacements that were as close to the original voltage / capacitance / tolerance as possible. I bought Elna caps to replace the original Elnas, and United Chemi-Con to replace the Nippo Chemis. Everything came from Digi-Key. After recapping, I cleaned the circuit boards with flux remover, the chassis with alcohol, and the switches / contacts with DeoxIT. Unfortunately I didn't take many pictures of the recap process, but here is a photo of all the caps I replaced:

Here is a photo of one of the large power supply caps and its replacement, to show how dramatically cap sizes have decreased over the last 50 years. I had to wrap some foam around the new cap to make it fit in its holder.

Once the recap was done, it was time to power up and find out where things stood. When powering up electronics that are in questionable condition, it's a good idea to slowly increase the power applied. This way if any of the components inside are drawing more current than they should be, you can catch the problem without causing even more damage. Variacs are commonly used for this purpose, but I don't have one, so I made a dim bulb tester. The 1060 passed the power-up test with flying colors, and I was able to plug it into the wall.

With the unit safely powered, I performed the "maximum power output" test listed in the service manual. This test confirms that each channel is capable of delivering 30W to its load (typically a speaker) at its maximum input level. After that, I hooked up a function generator to the input and a sacrificial speaker to the output. Everything sounded good. I took a look at the output with an oscilloscope and the waveform looked nice and smooth. I tested all the inputs, outputs, and controls and everything seemed to operate normally. Perhaps one of the caps I replaced had caused the amp to fail years ago.

I'm sorry to say I did not perform any further analysis to test noise floor, frequency response, or distortion. An honest-to-goodness restoration would require these tests. I was more interested in getting the amp in reasonable working condition than making sure it perfectly matched its original spec. Maybe one day I will go back and perform these tests for my peace of mind.

The final step was cosmetic restoration. The faceplate was very dirty. I read a lot of online forums and determined that the most popular products for cleaning aluminum faceplates were Windex, dish soap, Blue Magic metal polish, and Simple Green. I tried all of these. They did a decent job, but the best performer was actually a Magic Eraser (which I was careful not to use on the screen printing). I performed a few other minor repairs such as replacing missing hardware and fixing one of the plastic feet with some Gorilla Glue. I found OEM replacements for the missing switch caps on eBay, but they were exorbitantly expensive (~$10 per cap). I kept looking and found some almost identical caps from China that were $8 for a pack of 10. I replaced all the original caps with the Chinese ones and I think the overall look is nice and similar to the original.

Once I got the unit hooked up to an iPod and my New Large Advent loudspeakers, it sounded beautiful. Here are some final glamour shots:

The cost of the entire job, including components and cleaning products, was $106. Fortunately the generous eBay seller who sold me parts included a bunch of stuff I didn't need, which I was able to resell on eBay and bring my total cost down to $58. I still have some parts left to sell so I might be able to get that down even further.

I have no idea how many hours the project took me; I worked on it sporadically from February of 2013 to October of 2014. Since I finished the amp in the Fall, I wrapped it up and put it under the tree for my dad this Christmas. He was very happy to receive it, and we hooked it up in his system on Christmas morning. I was surprised by how much better it sounded than his 90's-era Kenwood receiver. Warm, clean, crisp, beautiful tones. I was delighted with how the project turned out.

If you're interested in restoring a 1060 yourself, here are some resources that you may find useful:
Schematic, manual, and service manual from Hifi Engine
My bill of materials for the re-cap

The good photos are courtesy of Erin Scott Photography.