Overview My original plan for this project was to build a hybrid amplifier with a solid state preamp section and a tube power amp section. The reason for this was that I wanted the reliability and size advantage of a solid state preamp section coupled with the “fat” tube tone of a power amp section. I also felt that with a solid state preamp mounted on a PCB, I could create a preamp stage with more gain while taking up less space. For that design I would need a separate power section for the solid state preamp since it does not run on the high voltage that the tubes do. After talking with Harry Powell he suggested that I might as well go with an all tube design. That was fine with me because I wanted to know more about designing a tube preamp anyways. My design specifications were pretty simple; I wanted to design a single channel 50watt amp using 12ax7 preamp tubes, EL34 power tubes, and a baxandall tone control. The baxandall circuit attenuates or amplifies two ranges of frequencies, bass and treble.

Preamp The preamp stage has two 12ax7 preamp tubes and each 12ax7 holds a pair of ECC83 triodes. 12ax7 models and information on how to create them on Pspice were obtained from duncanamps.com. The preamp basically consists of common cathode gain stages place in series, a baxandall tone circuit, and a phase splitter. The amount of gain coming from the common cathode stages was determined by the relationship between the resistor on the anode and resistor on the cathode. The larger the resistor on the anode in relation to the cathode resistor, the larger the gain is. Below is a basic schematic of the common cathode layout.
Common Cathode Gain Stage
[pic]
Instead of the baxandall tone control being a separate entity, it was integrated into the beginning stages of the preamp to have maximum effect over the incoming guitar signal. Typically, for guitar players bass frequencies range from 100-300Hz, middle frequencies range from 300Hz-1kHz, and treble frequencies are anything above 1kHz. The traditional tone control used in guitar amps is the tone stack which has three controls; bass, mid, and treble. I chose to use the baxandall tone circuit over the tone stack because it is more versatile. For example, in the figure below taken from the duncanamps.com website is a simulation of the traditional tone stack with the bass and treble potentiometers turned down and the middle potentiometer turned all the way up.
Simulation of traditional tone stack with bass and treble turned down and mid turned up
(taken from duncanamps.com)
[pic]

As can be seen, it is incapable of producing a middle frequency boost even with the mid control maxed out. The baxandall on the other hand can produce quite a large middle frequency boost as seen in the figure below.

Baxandall simulation with bass and treble turned down to create a middle frequency boost (taken from duncanamps.com)
[pic]

The last stage of the preamp is the phase splitter which should produce equal outputs that are 180° out of phase with each other. These two signals feed into the power amp stage and should have a maximum clean voltage of 25Vpk. The preamp circuit along with its simulations is included in Appendix 1.

Power Amp & Output Transformer The class AB push-pull output stage consists of two EL34s feeding a Hammond 1650K transformer with the secondary impedance set to 8 ohms. The secondary is configurable between 4, 8, and 16 ohms. I chose 8 ohms because all of my speaker cabinets are rated at 8 ohms. Each EL34 receives the signal from the output of the phase splitters. The purpose of the output transformer is to step down the high impedance from the output stage to a low impedance suitable for the speakers. To achieve a 50watt output, the voltage seen on the 8ohm load should be approximately 28.3Vpk or 20Vrms (using Vrms2/RL = wattage). The primary impedance on the Hammond transformer is 3400 ohms. Using the same equation as before, Vrms2/RL = wattage, the voltage seen on the primary side should be 583Vpk. Pspice models for the EL34 were available on duncanamps.com. The power amp schematic and simulations are in Appendix 2. The Hammond transformer has ultra-linear screen taps that can connect directly to the screen grids on the output tubes. The other option was to pull the voltage directly from the power supply. I ended up using the ultra-linear screen taps for simplicity.

Power Supply The power transformer is a replacement transformer for the Fender Bassman amp, but its specifications work out quite well for what I need. The primary is 117V at 60Hz, and the secondary is 330-0-330V with a 53Vrms bias tap. Filament windings run at 6.5V. To rectify the AC voltage out of the transformer I used a full wave rectifier. The Duncan Amps website has a convenient power supply simulation program where one can choose from multiple topologies and filter types. I ended up using a basic full wave rectifier going into a Capacitor filter with a load resistor. The capacitor value was 220uF. Bypass capacitors were used for the voltages going into the plate of the 12ax7s and those were 56uF. As seen in Appendix 3, the voltage out of the power supply was 485VDC.

Bias Circuit The bias circuit controls how much voltage is going into the grid of the EL34s. The less negative (towards 0) the bias voltage, the more the tubes conduct. The more negative (towards -∞) the bias voltage, the less current the tubes conduct. This is demonstrated in the graph below:
[pic]
To create a negative voltage going into the grid, the typical design is to use a half wave rectifier with the diode’s cathode connected to the input side. As mentioned before, the power transformer used has a convenient 53Vrms bias tap designed for this purpose. Below is the schematic used for the bias circuit. There is a resistive voltage divider before the diode to step down the voltage. Output voltage is taken from RL.

[pic]
The simulation with a 60Hz, 53Vrms input is as follows:
[pic]

The actual result:
[pic]
Conclusions Like the 407 project last semester I still get a thrill out of designing and building a working amplifier. I do plan on making more amps and trying out that hybrid configuration I mentioned in the introduction. Tube technology seems to be pretty outdated so I am not sure how valuable this kind of experience will be in the real world, but it is something that I am really interested in so perhaps more will come out of it. As far as I know, the only tube amps being made these days are guitar amps, and there is still quite a market for a good tube amp among guitar players. A lot of digital amps on the market try to replicate the sound of well known tube amps like Marshall and Fender. Getting into whether they do a good job is another discussion, but most agree nothing sounds like a real tube amp turned up. As for this project, I’m pretty satisfied with what I did. The design was fairly simple so there was not much to go wrong on, although I did make a couple of wiring errors. The amp sounds pretty good for being the first tube amp I have ever designed and built. Future improvements/modifications might include putting in an extra gain stage, multiple channels, a gain boost function, a mosfet for the phase splitter, different output tubes, a different placement of the tone control, and a solid state preamp.

Output waveform on 8ohm load, top is output, bottom is input
[pic]
As seen from the picture above, the output delivers about 39Vpk-pk, or 19.5Vpk assuming a perfect sinusoidal signal. This translates to about 24watts of power. The source of distortion is something I plan to look into and improve in the future.

Appendix 1 – Preamp

Preamp Circuit
[pic]
Demonstrating the baxandall’s mid boost with peak at around 400Hz
[pic]

Output of the phase splitter, input = 200mVpk, output ≈ 25Vpk, A = out/in ≈ 125
[pic]

Appendix 2 – Power amp

Power amp schematic, 01ohm resistors present for simulation purposes
[pic]

Bottom is input to power amp, top is at the 8ohm load, gain ≈ 1
[pic]

Appendix 3 – Power Supply

High Voltage
[pic]

Power Supply Schematic
[pic]

Power Supply Simulation from 0-300ms
[pic]

Ripple voltage zoomed in
[pic]