Today we are going to analyze my favorite from this collection, the Bugle. The Bugle is a stereo 45 amplifier yielding a generous 2.5W of power. I don't even try to sell this amp anymore because of time and sketchy availability of the 45 tube, but sound-wise

One nice thing is that a 45 is always a sin-gle plate tube and a lot easier and cheaper to find than the single plate 2A3. Also, there are rumblings that the Chinese are remaking the 45 tube, but I haven't seen any yet myself!

When designing an amplifier, I always start by characterizing the speakers and work my way back. I would estimate a good speaker of 95dB sensitivity or better at 8 ohms or higher to be right for this project.

So, let's pick the super-sounding 45 tube, which has a dissipation of 10W on the plate. In my experience the 45 is best with no more than 35mA plate current. We can take the plate voltage up to 275V at that point. I believe that when using the 45 (as well as other directly heated triodes) that self bias is best. The required bias at 275V for 35 mA plate current is -55V (actually, based on the dozens of 45s I have tested, you can expect between 52-54V in real life), which when added to the 275V plate voltage makes for a

The bias voltage is between 52-54V volts so, by Ohm's Law, we divide the mean voltage by 35mA and get a bias resistor of 1514 ohms or better yet use a readily available 1.5K unit.

We can determine the wattage dissipated of this resistor by I squared x R = 1.8375W I try to de-rate all my wattages by a factor of 5 for long life and stability. In this application, I would use at least a 10W rated part. In my version, I used the 12W Mills MRB-12 resistor because they are really nice units at that wattage.

In the Bugle, I originally set everything up for the 6072A, one of my favorite tubes. Then I also found that with the same bias-ing resistors, the 5751 also worked and provided a higher overall gain. I biased the 6072A at 2V @ 2mA at about 140V per plate in SRPP mode. The bias resistors for each section work out to be 1 K.

I find the SRPP drive stage to be good sounding, and easy to implement, offering a lot better output capacity than plate load-ing the same tubes. Because there is enough local feedback in an SRPP so that it is not beating the tube for maximum gain, I find it vastly superior in sound to all the types of the mu follower designs. Following this logic, a simple plate loaded tube should sound best, and maybe it does, but then you have to use two stages and that introduces another set of problems. In my experience, the SRPP is the best sounding simple way to go.

At 2mA, the 6072A has plenty of slew rate current. This design has an output impedance of about 15K. The 45 has a

I found that the best plan for bypassing is to use 5Hz as the limiting low end for cal-culations. Most designers go way over-board with the bypass capacitance, slowing things down. So we have 1/[2*PI*F*R) = 1/(2*PP5M70K) = 0.0677uF. I use a 0.047uF for a low end response of 7.2Hz, plenty for this type of amplifier.

Also, the smaller the capacitor the better the sound. Lately the largest coupling capacitor that I can stand is a 0.25uF in an amplifier. I would try the Audio Note cop-per or silver foil or the Hovland MusiCaps. Since there is a potential of 140V across the capacitor, a 400V or higher voltage rat-ing is required.

Okay, so we have the gain, output stage, now we need the power supply. The 45 requires a 2.5V filament at 1.5A, 6072A needs 6.3V/0.3A each, and we need a sup-ply B+voltage of 330V.

In the Bugle, I originally set everything up for the 6072A, one of my favorite tubes. Then I also found that with the same bias-ing resistors, the 5751 also worked and provided a higher overall gain. I biased the 6072A at 2V @ 2mA at about 140V per plate in SRPP mode. The bias resistors for each section work out to be 1 K.

I find the SRPP drive stage to be good sounding, and easy to implement, offering a lot better output capacity than plate load-ing the same tubes. Because there is enough local feedback in an SRPP so that it is not beating the tube for maximum gain, I find it vastly superior in sound to all the types of the mu follower designs. Following this logic, a simple plate loaded tube should sound best, and maybe it does, but then you have to use two stages and that introduces another set of problems. In my experience,

At 2mA, the 6072A has plenty of slew rate current. This design has an output impedance of about 15K. The 45 has a maximum grid resistance of 1 M in self bias mode. If we use a resistance of 470K [I found some of the globe 45s draw grid current when a higher value is used) we can determine the necessary coupling capacitor.

I found that the best plan for bypassing is to use 5Hz as the limiting low end for cal-culations. Most designers go way over-board with the bypass capacitance, slowing things down. So we have 1/[2*PI*F*R) = 1/(2*PP5M70K) = 0.0677uF. I use a 0.047uF for a low end response of 7.2Hz, plenty for this type of amplifier.

Also, the smaller the capacitor the better the sound. Lately the largest coupling capacitor that I can stand is a 0.25uF in an amplifier. I would try the Audio Note cop-per or silver foil or the Hovland MusiCaps. Since there is a potential of 140V across the capacitor, a 400V or higher voltage rat-ing is required

I prefer a choke loaded power supply over a Pi filter because the bass is so much faster and punchier with choke input. For this design I would use the 5AR4/GZ34 rectifier because of the low B+. If you are building an amplifier with a higher B+ voltage, then you may have to use its big-ger brother the GZ37. To determine the transformer necessary for the desired B+ output using the curves supplied by Amperex we see that we need a plate to plate voltage of 700-720 for the full dissi-pation of some 80mA [stereo circuit).

What I do to trim the voltage is to add the input capacitor Cx to get the B+ up to where I want it. The reason the circuit does not come in at the exact voltage shown in the curves is because of the resistance in the choke and losses in the power transformer. A small input cap can be used with a choke input to raise the out-put voltage to make up for these losses.

Ihave found large chokes and smaller capacitors to be the best case for choke loaded supplies, but high capacitance or multiple stage supplies are necessary for low ripple on the supply. Here we will use 2-l00uF in parallel with a 20uF SCR film for a high-C, single-section supply. That will drop the ripple down to less than 30 mV p-p with a 30H Hammond 157G choke for each channel.

Since we have a B+ of 330V we need to drop this voltage for the 6072A by around 55V @ 2ma, requiring a resistor equal to 27.5K ohms. A 27.4K with a wattage of 0.5W or better should do the trick.

That's it for design, now some parts notes: I use Shinko tantalums for all my 0.5W and 1W needs these are available from Angela Instruments [301)725-0451. The Mills and most of the capacitors are avail-able from Michael Percy [415)669-7181. The chokes are available from Handmade [610) 432-5732. The chassis are available from Experiences Sonores [418)652-8788 FAX-see ad in SP #72 page 50, right pic-ture is the Bugle chassis. The output and power transformers are from my faves MagneQuest [what's that number...) [215) 288-4816. I present this work in the spirit of DIY for the personal, non-commercial, hobby use of SP readers only. I must restrict copying of this schematic, without written permis-sion from Sound Practices and Wavelength Audio. Also, please do not call

Thanks and have a good time. Gordon.

   circuit diagram of the 45 Stereo Amplifier