DCAudioDIY.com

I’m building a phono amplifier for my stereo. It will be set up for a moving-magnet (MM) cartridge, and can be used with a moving-coil (MC) cartridge through a step-up transformer. I am using tubes in the analog path because they sound better in general than solid-state components to me, and they are fun to work with. I am also using a collection of solid-state components for current control and an output servo, so we are looking at a hybrid amplifier. The power supply is also a hybrid of tube and solid-state devices.

Design Goals:

low noise,
tight RIAA conformance,
low distortion,
freedom from overload and blocking,
capability to drive 10 kOhms of cables and load, and
all-tube amplification.

Approach:
I prefer using passive circuits, not feedback, to achieve RIAA correction. The passive inverse RIAA circuit could be all-in-one, but splitting it between the first and second stages makes it easier to implement and adjust: in fact, my first and second stages have identical topology. The second half of the inverse-RIAA filter has a second zero at 50 kHz, as shown in Figure 1:


Figure 1: Phono Amplifier Block Diagram

I want a lot of transconductance in the first stage to minimize noise, so I’ll use two Russian 6S45Ps, one for each channel, biased to 30 mA each. I’ll use two 6AM4s, again one for each channel, biased to 9 mA for more gain in the second stage. The line driver (third stage) is a 12SN7 with a pair of servo drivers in the cathode circuits. The line driver is biased to 8½ mA per triode.

Because I have three schematics to load, I'll add another post to continue my discussion.

_________________
- Guy

First Stage

The 6C45P has a maximum plate voltage of 150 Volts, a maximum plate current of 52 mA, and a maximum plate dissipation of 7.8 Watts for each plate, as shown in Figure 2. Forward-biased LEDs have low dynamic impedance, low noise, high bandwidth, and provide essentially instantaneous recovery from overloads (see Morgan Jones and Stuart Yaniger.) I will use silicon-carbide Schottky diodes in the cathode circuit to bias the tube, so I’ll need a bias in multiples of 0.8 Volts. A bias of 0.8 Volts should result in about 130 Volts across the tube and 30 mA of current through it. I’ll drive the plate circuit with a constant-current source (CCS), which should have about 20 Volts across it, so the first stage will be supplied with 150 Volts. Each of the tube’s grid pins (2 and 8), as well as the plate pin (7), has a grid-stopper resistor to limit radio-frequency (RF) oscillation. The grid-leak resistor sets the input load impedance. The circuit is shown in Figure 3.


Figure 2: 6C45P Plate Curve


Figure 3: Schematic of First Stage

_________________
- Guy

Second Stage

The 6AM4 has a maximum plate voltage of 200 Volts, a maximum plate current of 10 mA, and a maximum plate dissipation of 2 Watts, as shown in Figure 4.


Figure 4: 6AM4 Plate Curve

I will use silicon-carbide Schottky diodes in the cathode circuit to bias the tube, so I’ll need a bias in multiples of 0.8 Volts. A bias of 0.8 Volts should result in about 160 Volts across the tube and 8 mA of current through it. I’ll drive the plate circuit with a CCS, which should have about 20 Volts across it, so the second stage should be supplied with at least 180 Volts. Each of the tube’s grid pins (1, 3, 4, 6, and 9), as well as the plate pin (5), has a grid-stopper resistor to limit RF oscillation. The circuit is shown in Figure 5.


Figure 5: Schematic of Second Stage

_________________
- Guy

thanks for the info, Guy, I will be interested in hearing this. I am almost finished with Stu Yaniger's "His Master's Noise" so it will be instructive to compare the two, although HMN is a LOMC phono preamp.

Driver Stage

Note that the output impedance for a cathode follower is about 1/gm, and the 12SN7 has a transconductance of 2.5-3 mA/V, which translates to an output impedance of 330-385 Ω. Add a cathode stopper and we're up to roughly 1 kΩ. Because I already need over 180V for the 2nd stage, it’s easy to use the 12SN7 instead of a lower-plate-voltage tube.

The final stage, designed by John Broskie, is a triode-based cathode follower with a small cathode resistor driven by a complaint CCS, which is powered from a -12-Volt power supply (see http://www.tubecad.com/2007/04/blog0102.htm.) The compliant CCS looks like an AC CCS to the triode, and sets the cathode-follower DC output to zero volts. The circuit is shown in Figure 6.


Figure 6: Schematic of Driver (Third) Stage

As Broskie describes , the cathode follower is loaded at its output by a compliant current source, i.e. one that does not have a predetermined quiescent current. Instead, the current source strives to maintain a DC ground-potential input regardless of the current flowing through the cathode follower. It offers high impedance to any AC signal it sees, and in this respect it is a current source. In DC terms, it adjusts its quiescent current until its input is at ground potential over time (roughly, 3 Hz). This is the result of a DC servo loop wrapped around the input of the current source and the output of the op amp U3. If the input moves positive over a long period of time, the net DC drift is fed into the non-inverting input of U3, which causes its output to go positive. The positive-output voltage will drive the MOSFET M5 into greater conduction, which will drive the output more negative. If the output moves negative over a long period of time, the net DC drift is fed into the same non-inverting input of U3, which causes its output to go negative. The negative-output voltage will drive the MOSFET into less conduction, which will drive the output more positive.

The subtlety is that M5 is within the DC-feedback loop, but outside the AC-feedback loop of U3. As the time constant of the RC network made up of the two 1-meg resistors (R21, R22) in parallel and the 0.22-µf capacitor (C6) is so long that no music can fall into it, U3 presents a virtually constant DC voltage to the gate of M5. This steady voltage sets the amount of current that flows through M5; if the tube's idle current drifts over time, U3's output will drift with it. While responding to the music signal, the Cathode Follower's output voltage and current will vary, but too fast to register at the input of U3 and will be ignored.

The 1-meg resistor R25 that connects from a bias voltage to the output gives the compliant current source a current path in the absence of a tube in its socket or at start-up when the tube has yet to conduct any current. This prevents a pop on start-up.

Next, a power supply...

_________________
- Guy

Guy,

I'm looking forward to hearing this. My 2 cents: the 6C45Pi will oscillate if you look at it funny especially at higher currents. Carbon grid and plate stoppers with the body of the resistors right up against the pins of the tube are the name of the game. PITA that tube, but not as bad as the 6GK5.

Also I always put an automatic mute on the output for turnon and turnoff transients - easy enough to do with a 555 timer or mosfet with timer circuit driving a relay.

Charlie

Charlie,

The power-on/off mute is something I didn't document but is included on my PC board for the output servo, so I have that covered.

You're right - we'll see how bad the oscillations are - gee, I wonder if they'll just not show up?

_________________
- Guy

Guy,

You are putting a highly non-linear device in the cathode circuit of your first stage. You can use one or more diodes to bias the tube, but you should pass a large dc current through the diode from an external source to linearize the circuit. One example I did was 0.2 mA through the tube and 50 mA from a very clean dc source through a wire-wound resistor. This puts the dynamic resistance of the diode into around an ohm or two and the modulating current through the tube is insignificant for further modulating the dynamic impedance of the diode. As you have it, the dynamic impedance of the diode will be fairly high and will be highly modulated by the signal. This will create far more distortion than any transistor amplifier circuit.

David Berning

Will the 30 mA current running through the tube be enough?

Dave, how is the diode non linear? Do you mean that the diode's resistance decreases with increasing current? How big is this effect? I remember checking an LED to use in a 12AX7 cathode. The voltage across the diode didn't change more than 1/10th of a volt when current through the LED varied from 1mA to 20mA. I thought it was okay.