Last month, I walked through the design of a simple 1.5 watt SET poweramp using the 12B4A triode. This month I will explain how I put this amplifier together into a "finished product" of sorts.
The first step to actually building the amplifier is to work out the proper layout. Layout is very important. It is what keeps nasty things like hum and buzz from invading your listening space. A good layout is also crucial to being able to easily put the amplifier together. Its important to be able to be able to get your fingers, soldering iron, needlenose pliers, etc into the chassis to work.
Before deciding on a layout, you need to decide how you want your chassis to look. Do you want all your input, output jacks, volume controls and switches all on one top plate? Or do you want a more traditional looking amplifier, with power switch, and volume controls on the front, output jacks on the back, and tubes on top? I mentioned last month that this was intended as an office system. As such, a "traditional" easy-to-use approach is warranted. Since I now know that there will be input jacks, binding posts and a fuse holder on the back, a volume pot, power switch and selector switch on the front, and the tubes on the top, I can now draw a layout.
Looking at the layout, we see that the power supply area is in the back, and the amplifier area is in the front. Shielded wires will be run between the input jacks in the back, to the source selector switch up front. Along the other side of the chassis, the power cord comes in, and goes to the power switch, and the power indicator lamp. The output transformers are mounted lengthwise in the chassis. This is to save some space for other things inside. The power transformer's windings, and the filter choke's windings must be mounted at a 90 degree angle to the output transformers. This is done in hopes that any stray electromagnetic "spray" is cancelled out by the 90 degree difference in windings.
To be able to pull this off, I need some way of mounting components on the top, front and back. This means ill need some metal. There are two ways to do this effectively. Use a piece of metal that is bent over, or build a wooden box, and cut a hole in the wood on the front and back and bolt on pieces of metal. You could always buy a Hammond or Bud project box, but I prefer to do the metal work myself. A trip to Sears landed me a nice piece of steel; 8" wide, 18" long and 1/16" thick. I normally use aluminum, but since I need to bend the metal, steel works better. Using a brake, I bend two 1/2" flanges on the ends of the steel to act as lips. Then I bent another 2.5" to make the front and back. This leaves about 11.75" across the top, and 2.5" front and back - with a small lip that curls in. Now that I have something that looks like an amplifier box, I can punch the holes for the tube sockets, transformer mounts, etc. Using a set of sheet metal punches, I punched holes for the 5 tubes, the big JJ Can style caps, the power switch on the front and the fuse holder on the back. The punches work by drilling a pilot hole in the metal. A bolt is then fitted to the hole, containing the punch. The punch is essentially a cup and a pair of teeth that "bite" a round hole into the metal. The punch is then screwed down with a socket wrench, and a hole is born. The rest of the holes are small enough to be made with a drill. These holes consist of holes to mount the transformers, retaining rings for the tube sockets and caps, etc. Once all the holes are punched, the chassis is primed and painted with a dark grey hammertone.
At this point we are ready to mount all the parts to the chassis. Once that's done it starts to look like an amplifier. In the photo below I have added the tubes for aesthetic purposes. Nothing is connected as of yet.
Now to start connecting. I started with wiring the input jacks and the power switch. Those seemed like they would cause the most grief since they required running wires from the back all the way to the front. For the input jacks, I used a shielded wire because the low level input signal will be subject to nasty AC interference as it runs near the power supply. A shielded wire is simply a wire that has a copper braid surrounding it. It is important to remember to ground the shield of a wire at only one end. Grounding at both ends can cause nasty ground loops and oscillation. The input wires are lead to the selector switch, which is then wired to the volume pot. Next came the power switch wiring. The power cord is run up to the front of the chassis, and to the switch, going through little green Radio Shack neon lamp along the way. The lamp isn't necessary since there are already 5 power-on indicators on the top of the chassis. But I like the way small neon lamps look :)
Next to be wired is the heater circuit. Doing this carefully is CRITICAL to keeping hum out of the amp. I chose to go with AC filaments for two reasons. One, its simpler (and I like simple). Two is that there isn't too much space available under the chassis for a rectifier and cap bank needed to make a clean DC supply. Going with AC filaments means some precautions must be taken. The wiring must be tightly wound on itself. This prevents the high level AC signals from interfering with sensitive signal carrying wires. The twisting helps cancel out some of the junk that leaks out, much like why we oriented the transformers 90 degrees from each other. The twisted heater wires are also run close to the metal chassis in hopes any more of the AC interference is absorbed by the chassis.
Now that we have the input, power and heaters wired, the only thing left to do is wire the power supplies and the amplifier circuit. The power supplies were wired directly to the solder lugs of the capacitors. Dual Mono supplies are used to help prevent any cross channel noise, or oscillation between channels. Technically this is a pseudo-dual mono design since I use one power transformer and each channel has its own filter circuit. The amplifier circuit is built between the 6AU6 terminals, 12B4A terminals, and a terminal strip mounted to the same screw the output transformer is mounted to. A pair of oak rails are added to the sides to finish off the look as seen above. A picture of the completed underside is seen bwlow.
From my brief test in mono, with one channel finished last month, I knew this amplifier sounded nice. I wasn't prepared for just how nice. This thing sings with a nice clean and clear sound I have only heard from single ended triodes. There is a slight hum, sounds like 60hz... presumably due to the AC heaters. This hum is only noticeable on the sensitive horn speakers, and not on the 'average' sensitivity monitors in the other room. In any case, its something for me to tweak out later; the hum is entirely unnoticeable when sitting at the listening position. I had just inserted a solid state integrated amplifier into my system the day before, just to make sure it still works. So I had the sound of a solid state amp in my mind when I hooked this little guy up. It was free of any solid state hashy garbage. This amplifier seems to be on the warm side of neutral, it has good extension top to bottom, even with only 1.5 watts. It is definitely missing the soft warmness that most people associate with tube amps, although there is no mistaking it with the SS integrated. In fact, this amp is quite fast and peppy. It throws a nice stereo image with no smearing. I will attribute that to the pseudo dual mono power supplies more than anything else.
It's missing something however, so I need to tweak the circuit a tad. When I breadboarded the amp last month, I had the driver tube's cathode bypassed with an electrolytic cap. Figuring the less capacitors I put in, the better, I decided to leave out the cathode bypass cap on the driver tube. I still got the required gain to drive the power tube so I figured it couldn't hurt. I noticed the completed amp sounded a little more "closed in" than the breadboarded circuit. I'll hazard a guess and say this was caused by the local feedback around the driver's cathode due to the lack of a bypass capacitor. So I went ahead and stuck the cap back in. Well, the change was subtle, but definitely necessary and audible. The amp was much more open sounding now. Capacitors must not be as evil as everyone has made them out to be.
Capacitors may not be evil, but their quality is definitely an issue. Especially any capacitor that is directly in the path of your precious signal. The DC blocking cap that sits between the driver stage and the power stage must be of excellent quality. I used an inexpensive Sprague Orange Drop in this position. The Orange Drops are not bad capacitors to begin with - but one can do better for only a few more dollars. I put in a pair of Kimber Caps in place of the Orange Drops, and after a few days of letting them run in, the sound opened up even more. A slightly wider soundstage with a clearer midrange is what I noticed. Nice change for only $10 total. The Kimber caps were $5 each, compared to the Orange Drops, which were $1 each. I may try an Auricap in place of the Kimbers at some point. The Auricap is my current favorite coupling capacitor, but I had never tried the Kimbers, so I decided this would be a good place to try them.
Since I took the time to improve the quality of the capacitor that sits right in the signal path, why not look at other parts? A very simple and effective tweak is to get the volume pot out of the signal. This pot is your standard $4 Alpha 100k log taper pot. It would be nice to get it out of the signal path as much as possible. The way to do this is to run the volume control in shunt mode. This is done by sticking a resistor that is about half the total value of the pot in series with the input tube's grid. Then connect the grid to the wiper of the pot, and ground the bottom of the pot. Now instead of varying the resistance between the grid and input jack using the pot, the input signal is fixed (using a decent quality resistor) and the difference between grid and ground is what is changed. A stepped attenuator would be even better, but as an inexpensive solution a shunted pot is hard to beat. However, I honestly cant tell if this changed the sound at all, but it seems to me as if the volume control itself works more smoothly.
Now to take care of the slight hum. Since my oscilloscope is telling me the hum is all 60hz, i can safely assume is being caused by the AC heaters. I would rather avoid going to a DC setup, so I am going to try a few tricks up my sleeve. The heaters at the moment are at ground potential. This means they could pretend to be a virtual cathode with respect to the real cathode (like how a directly heated tube works). This could be the source of the 60hz hum. The solution to this is to bias the heaters positive with respect to the cathode. This way any leakage noise leakage will be in the direction of the heaters instead of the signal carrying parts. I added a voltage divider to the power supply, and attached the heaters to the +50V DC point. This lower resistor is bypassed with a .1uF capacitor to further help the hum rejection. Now with the heaters at least 25 volts positive with respect to the cathode, the hum has died down to barely audible levels. Now its time to put this little guy back in the system.
OK, I am going to call this amplifier done. I'm sure I could go nuts and replace the transformers with Magnequest, change out the inexpensive metal film resistors with Holco or Rikens, etc. But in all honesty, I like the sound of this amplifier as it is. I can get it to play loud on my horns, but when mated with some relatively inefficient monitors (Sound Dynamics RTS-3s at 91dB/W/m) and played at low levels, the sound is very satisfying. There is a nice tube warmth, but is rather lithe at the same time. This amplifier makes my toes tap uncontrollably (which makes it hard to type this since my laptop is, well, in my lap). I'm just going to listen to my sweet 1.5 watt amp just as it is.
Well, this little design adventure has been fun. I have learned that unpopular tubes make for a fun project, and can yield great sound. I hope I haven't bored anybody with my technobabble, but if you have read this far then I guess I haven't. Should anyone wish to clone this amplifier, feel free to email any questions you may have to me.