This driver circuit was designed to work with a 300B tube. As a 2A3 is essentially half of a 300B, this circuit should – with a few component tweaks – work with a 2A3 as well.
Driving a 300B well, is not easy. As the grid-to-cathode voltage, Vgk, approaches 0 V, the tube starts drawing grid current. If a coupling capacitor is used on the grid of the 300B – as is the case for most 300B driver circuits – the coupling capacitor will start to discharge when the 300B draws grid current. This causes the voltage across the capacitor to change, hence, the bias point for the 300B to change momentarily. In severe cases where the amplifier is driven hard by a sharp transient – say the beat on a snare drum – it can cause the output stage to ‘lock up’. To guitar players, this is known as ‘farting out’. The result is that the output of the amplifier remains at a constant voltage until the output stage recovers. This can take 100’s of ms and causes some rather rude noises to be produced by the speakers. The solution is to use a driver that has a low output impedance, and ample current sourcing capability. A source follower is an example of such a driver.
The circuit below deals with high voltages — LETHAL voltages. If you are not comfortable or qualified to deal with these potentially deadly voltages, please do not attempt to build this circuit. The design is provided as-is with no warranties or service agreements whatsoever. It is provided in the spirit of DIY and may only be reproduced for non-profit purposes. Proceed at your own risk, expense, and responsibility.
Q3 is the source follower mentioned above. The source follower can be biased by either a constant current source (CCS) consisting of Q4, R14, R15 or a resistor, R16. The bias point for the 300B output tube connected at J4 is set by the resistive divider of R17~R19. The input stage is a fairly standard LED biased grounded cathode stage with CCS load. This stage has a gain of about 20 V/V. The second stage is an inverting amplifier – by some called an anode follower – with a gain of roughly -2.5 V/V. The total gain of the 300B driver circuit is, hence, approx. -50 V/V. This allows the 300B to be driven to the point of clipping with an input voltage just above 1 Vrms. I designed this circuit for use with a Jensen JT-11P-1 line transformer. If this transformer is used, skip R12 and replace C4 with a wire jumper. I used 6J5 tubes. If desired, one 6SN7 (= 2 x 6J5) may be substituted.
Traditionally, directly heated triodes (DHT) have been heated by AC voltage. However, this results in the induction of hum. Various methods have been devised for reducing the hum, but none of them cancel it completely. In this modern day and age with high-current IC voltage regulators available, DC heating of DHT’s is much preferred. Linear regulators such as the LM350 are quite suitable for this application. There’s just one caveat with the linear regulator; significant power is dissipated in the regulator and will need to be gotten rid of in the form of heat. I did the math on a linear regulator design for a stereo 300B amp based on the schematic above. I concluded that under worst-case conditions, the regulators would dissipate more power as heat than the 300B’s could deliver to the speakers. I found this rather ridiculous. Hence, I decided to implement a switch-mode supply for the heaters.
The 300B filament draws approx. 1.5 A at 5 V. The LMZ12002 is just about perfect for this application. A couple of caps and some resistors is all it takes to get going. Unfortunately, its maximum output voltage is 5.5 V, so for the 6.3 V heaters, a different regulator IC is needed. The LM2734 was chosen as it’s available in an easy to solder SOT-23 package, handles 1 A, and requires only a few external components to work. Both regulators operate at about 85~90 % efficiency, so even though their only heat sink is a small-ish area of copper on the circuit board, they only reach 45~50 degrees C under normal operating conditions.
The circuit boards are designed to have the tube sockets mounted on one side of the board so the tubes can poke through the top plate of the chassis. The remaining components are then to be mounted ‘below deck’. The bias adjustment pot can be mounted on either side of the board.
The boards are professionally made. Double-sided, FR-4, with plated-through holes and white top and bottom silk screen. They measure 4.0 x 4.5 inches.
A Bill-of-Materials is available here: BOM_6J5_300B_DRV.pdf.
For further developments, see this DIY Audio thread: “RFC: 300B driven by anode follower“.