I know what you are thinking. What the hell am I going to use a 1.5 watt amplifier for? Well, as it turns out 1.5 watts is probably more than enough for an office system. As there were a couple of 12B4As in my tube drawer, so it made sense to put them to use. Since this project is an experiment, inexpensive parts will be used.
A trip to Duncan's Amp Pages told me the 12B4A is a small indirectly heated power triode in a 9-pin envelope. Its intended use is in the vertical deflection amplifiers in old televisions. Maximum plate dissipation is about 5.5 watts. Since a single-ended amplifier usually can get around 20% of the max plate dissipation to go to the loudspeakers, I figured one could achieve approximately one watt from these little guys.
The easiest way to design is from output to input. With that in mind, the first thing to do when designing an amp like this is to choose an operating point for the output tube (technically you start with the speaker, but I am assuming a standard 8-ohm speaker). The tube data for the 12B4A from Duncan Amps suggests an operating point of 150V on the plate and 35mA with bias set at -17.5V. The plate impedance at this point is about 1.3k ohms. There is a rule of thumb that says you should pick a load impedance that is approximately 2.5-3.5 times the plate impedance at idle. Since the plate impedance varies wildly with the AC swing of music (this is just a suggestion).
After looking at the tube curves, I decided a slightly better operating point would be at 180V on the plate, 27mA and a bias of -23V. This would produce about 1.56 watts when using a 5K ohm load, with just about the same distortion level that the suggested operating point gives. The easiest way to set this up is with cathode (auto) bias. Ohm's Law says that to get -23V drop with 27mA, therefore I need an 850 ohm cathode resistor. This resistor will dissipate about 0.6 watts. It is always best to underrate your components, so I decided on a 2 watt part. I had some 820 ohm, 3 watt resistors on hand - which is close enough to 850 ohms. In fact, that only moved the bias to about -22.5V, so that is not a problem. This resistor needs to be bypassed with a capacitor to maintain the correct gain and output power. This cap works out to around 50uF, and should be rated at least 50V. Any calculations that I do not mention and you are curious about, please e-mail me about them and ask.
Now, on the other side of the tube, we need a plate load. Since I decided on a 5K load, we needed an output transformer for that. The Hammond 125BSE seemed like a good choice. It can handle up to 45mA of bias current - and I am using approximately 27mA. So this output transformer will work. I bought mine from Angela Instruments. The last thing we need for the output stage now is a grid leak resistor and DC blocking cap. This resistor is generally best made as large as possible. Max rating for the 12B4A is 1Mohm. I will use a 470k 1/2 watt metal film resistor cause I have quite a few of those. The cap is unnecessary is you are going with a direct coupled design. However that would require some changes to the voltage points. The cap and the grid leak resistor form an RC high pass filter, so we need to be careful what size cap we use. It is generally a good idea to pick a nice low frequency cutoff for the high pass filter. I used 5Hz - and the cap works out to around .047uF. For this I will use an inexpensive Sprauge Orange Drop, which can usually be had for around $1. (My favorite coupling cap these days is the AuriCap, and in this value would be around $6 each - I may switch some in later on.)
Now we need a driver stage. The driver needs to do two things: First, it needs to provide enough voltage swing to drive the output stage to full output. Second, it needs to provide enough current to overcome the interelectrode capacitances of the power tube. The first requirement is easy to figure out. If we want to drive the output tube to full power, we need a AC voltage swing that is double the bias voltage. In this case, the 12B4A's grid is at -23V, thus we need the driver stage to swing 46 volts.
Now to determine how much current we need, calculate the slew rate the output tube needs to overcome its capacitances. Roughly speaking, this works out to about 2mA for the 12B4A. Any driver with at least this much bias current will do. Now to pick a tube that can do this. I have a bunch of signal tubes that could be used on hand, it is just a matter of deciding which one.
The 12AX7 is ruled out. It will swing 100 volts (that is way too much), at a max of 2 mA. We need some more leeway than that. A standard twin-triode as found in many pre-amplifiers will do, such as the 12AU7 or 6DJ8. However, since I am using such an unusual output tube, why not use an unusual driver tube? Lets go with the 6AU6. The 6AU6 is a small RF pentode in a 7-pin miniature envelope. Its intended use is in tuner sections. It is entirely unsuitable as a driver when used as a pentode - however when wired as a triode (screen and suppressor both tied to the plate) it makes a fantastic voltage amplifier.
I am running the 6AU6 at around 220V on the plate, 12mA plate current, and -4V bias. The bias is set to -4V so that the 2V output from a CD player wont clip the input. The 6AU6 is set up with a 10K load resistor and a 330 ohm cathode resistor. The load resistor will dissipate about 1.5 watts, so I have a 5 watt resistor in place. At this operating point, the 6AU6 is providing about 50 volts of swing, which is just over the 46 volts we need for the 12B4A's full output. This means the driver wont clip before the output stage clips - a good thing. Stick a 100k ohm log taper pot on the grid of the 6AU6 for a volume control and that part is done. All this can be seen in the below schematic.
The next thing that needs to be designed is the power supply. Since this amplifier is rather simple to begin with, a simple power supply is in order. The maximum voltage needed is around 350, used as the B+ for the driver stage. This works out to a transformer voltage of around 275. I selected the Hammond 270DX, which will give 275V at 90mA, center tapped, and has a 6.3V, 3A winding for the heaters. I had a 6BW4 (9-pin) full wave rectifier in my tube stash, so I will use that one. The output of the 6BW4 is sent into a CLCRCRCRC filter using a 10uF capacitor, a Hammond 156M choke, and then a JJ 20/20/20/40 can-type capacitor. The B+ for the driver is taken from the first 20uF cap. The B+ for the 12B4A is taken off the 40uF cap. The schematic can be seen in the below schematic.
It is now time to put this brainstorm to the test. Essentially, I need to make sure everything that was calculated would happen in real life. First, I needed to make sure my power supply design actually gives me the right voltages. So I wired it up using a few tack-solders and clip leads. Sure enough, there are the right voltages so that solves hurdle number one. Hurdle two involves checking to see if the driver stage indeed gives me 50V as calculated. I wired the output stage using tack-solders and clip leads as well, this time adding the DC blocking cap and 470k grid leak resistor (you need to have something for the driver to drive, even if it is just a resistor). Sure enough - I was getting about 57V from the driver tube using a 2V input from my signal generator. OK so it is a bit high, but that is of little concern. It will still work fine.
Now for the real big money - the power stage. I am starting to run out of clip leads at this point, but we will forge on ahead. With the power tube wired in my oscilloscope is telling me we are getting a whopping 1.56 watts at full output power before distortion takes over. At this point, this whole mess looked like the photo above. I think its time to listen to it. Sure only one channel is done (if you can call that mess 'done'), but that never stopped me!
The Real Test
I was fully expecting this thing to sound like absolute junk. After all, its clip leaded together, and it was a circuit I just thought up in my head. But boy was I wrong! This 1.5 watt flea powered amplifier sings! Its missing a touch of refinement that my 300B monoblock amplifiers have, but it certainly sounds like a single-ended triode amplifier. And yea, its only 1.5 watts... though my loudspeakers are efficient enough. I found myself groovin' at quite a decent volume with my 97dB horns. It had all those nice qualities like pacing and timing. Qualities I find lacking in some commercial tube amplifiers (perhaps the manufactures want to over-emphasize the 'tube sound' and end up making it sound ooey-gooey?). In fact, this amplifier sounds downright wonderful. This mess of clip leads sounded nice and fast, yet smooth. It will not play extreme loud, but within its useable volume range with is quite nice. Now I just need to find a tube that sounds like this wonderful indirectly heated triode that can give me a few more watts. This is one fun little tube!
Now all that's left is to build a chassis for it and do the final construction, and the all important critical listening test. Tune in next month for part 2 to see how this project comes out. Need some "office" speakers to go with this little "office" amplifier? Look for an upcoming article on a matching speaker...
Part 2 Coming Next Month!