This is a summer project built between my junior year of high school and freshman year at MIT. I went through a few prototypes, figured out how to etched double-sided circuit boards, and learned a lot about amplifier construction in the process. Because I didn’t want to take too many risks with such an expensive project (actually quite cheap for a tube amplifier – the total was well under $300), I used Fred Nachbaur’s Minibloc SET design for the pre and power amplifier to ensure I built something that was guaranteed to sound good, but modified the power supply slightly. 1 watt of output power is easily enough to fill my dorm room with music with the right speakers.
Before I went out an built an entire amplifier, I decided to make one or two prototype boards that would remain chassis-less to test the circuit. I quickly Eagle’d up a single-sided board that contained Fred’s entire original circuit:
Transferring the schematic to a board layout proved not to be very difficult, but I sacrificed ground plane continuity:
The board did work once it was etched and populated, but there were issues with noise and hum. The B+ voltage supply had too much 60Hz noise (too little filtering) for my ears, and the entire amplifier would start oscillating if the input was left disconnected. Nonetheless, I was encouraged to try to fix the issues and continue with another revision. (Note that I did eventually replace that puny little tube TV set output transformer with a Hammond)
First, I decided to split the power supply and amplifier sections into two boards. The power supply ground plane is less critical, so the board was routed single-sided. I used a DC filament because I thought this would reduce the 60Hz noise being coupled into the signal pathway. I didn’t realize, however, that by grounding one side of the 12.6VAC input voltage from the step-down transformer, I would effectively force the entire DC filament supply to bounce up and down with respect to ground! Oh well, it worked just fine in the end — very little noise is introduced by an AC filament in an indirectly heated tube.
Assembled circuit board, with short filter caps specifically chosen to fit the chassis:
Next, the actual amplifier board was laid out with special care given to keep the ground plane continuous. This meant it had to be a double-sided board. Adding tiny capacitances between each trace and ground actually has positive effects; high frequency noise is bypassed into the ground plane. I also put in grid stopper resistors before each dual triode’s control grids to further eliminate noise. The input impedance of a tube grid is already extremely high, so adding some extra resistance doesn’t effect the operation of the rest of the circuit, but the grid stopper acts as a lowpass filter with the grid’s Miller capacitance, further hampering high frequency noise. Together, these two fixes solved the oscillation issue.
The board layout; C1, C2, C10, and C11 are soldered to the underside of the board not to interfere with other components on the top side, and to make it possible to mount the tube sockets flush with the top of the chassis so the tubes plug into the top:
The board assembled, from the bottom. The big red WIMA capacitors are the DC blocking capacitors in the signal path:
Here’s what I meant by the tube sockets being flush with the top of the chassis:
The wooden chassis was milled (yes, you read that correctly, no, I don’t encourage you to do it!) out of a solid piece of wood I found in a pile of scraps being thrown away by a wood shop. I didn’t have access to proper woodworking tools at MITERS, so I used the metalworking tools. I highly discourage readers from following my example (In fact, woodworking on the metalworking tools is now banned at MITERS!). Concerned machinists might feel better knowing that the mill was cleaned and re-oiled following these tragic events 😛
First, the center of the wood was drilled out:
Next, I milled the inner surface smooth and milled channels on the top of the chassis for the aluminum top plate to slide in. Note that it’s important to only mill in the direction where the sharp edges of the endmill’s flutes spin in the direction of the wood grain, otherwise, force will be transmitted down the grain of the wood, causing splitting and cracking. The PVC pipe was cut to the inner dimension of the chassis to support the force of the XY table clamp:
Here’s the completed chassis frame (next to some other fun items to scale) with a piece of sheet aluminum cut on the bandsaw to fit on top and serve as the carrier for the circuit boards, audio jacks, and transformers:
The front and back of the chassis were further milled to accommodate an IEC power jack, a power switch, a volume control pot, and speaker binding posts. Once this was done, I drilled four countersunk holes into the sheet aluminum and screwed it into the chassis so it would serve as a base while I milled, drilled, and bored mounting holes for all the components on the aluminum top cover:
Once the entire chassis was done, I unscrewed the aluminum plate so I could stain and varnish the wood. Installing the electrical components was a breeze, because for once I thought out my chassis design before everything was already being put together! I plan to do this in the future as well. The top panel was grounded, but the circuit boards were specifically installed with nylon screws so they wouldn’t be connected to this ground; the entire amplifier circuit is connected to ground through a “ground loop breaker” consisting of a 47 ohm resistor in parallel with a 1uF capacitor. This “breaks” ground loops that would otherwise occur between the source equipment (computer, DAC, preamplifier, whatever) and the power amplifier (this thing).
All of the components connected and mounted in the chassis. The output transformers are Hammond 125CSEs ordered from Digikey:
I’m totally satisfied with the sound of this amplifier. Regardless of whether the single-ended distortion actually sounds better (which I believe it does), the aesthetics of the amplifier as well as a (possible) good dose of (maybe) placebo effect add up to a worthwhile project.
I’m using Cerwin Vega VS-80 entry level speakers that I picked up in basically new condition at Swapfest for $30. I did try the amplifier out at a nearby audio store with some $500 and $1000 Paradigm speaker sets, and they sounded amazing, but I wasn’t willing to spend that amount of money just quite yet.
If you’re interested in replicating this project, I’ve uploaded all of the relevant CAD files (in an unsorted and unorganized way, sorry!) to my Github account: https://github.com/dkramnik/Eagle-Public/tree/master/audio-13EM7-monoblock-power-amplifier