This driver board is optimized for a 300B tube. However, with a few minor tweaks, it can be used to drive many other tubes as well. 2A3 and 45 immediately spring to mind. With a separate B+ supply for the output tube, 801A and 211 in class A1 could be driven by this board as well. More about this in the Output Tube Options section below.
This 4.0×6.4 inch (10×16 cm) circuit board has been optimized for signal integrity. The purpose of this is to ensure that every amplifier built on this board will exhibit the least amount of hum and be the least susceptible to electromagnetic interference as possible.
|PCB ID||Description||Price each||QTY in stock|
|300B_Driver_R1p0||6N6P/ECC99/12BH7A 300B Driver Board, Rev. 1.0||$50||1|
Design Collateral & Quick Reference
- Schematic & Bill-of-Materials: 300B_Amp_DriverBoard_SchematicBOM.pdf
- Mechanical dimensions (click for larger view). The mounting holes are 130 mil (3.3 mm) in diameter.
The circuit schematic is below. The configuration using the 6N6P tube is shown. As with all the pictures, click for a larger view.
The driver architecture is as follows: Input transformer (TR1) → Gain stage (TU1B) → Current driver (TU1A) → Output stage (TU2).
The input transformer, TR1, provides conversion from the differential input to the single ended topology of the rest of the amplifier. For those not needing differential input, the circuit has provisions for placing an input capacitor instead of the transformer.
Common cathode stage comprised of TU1B and associated components provide the voltage gain of the driver circuit. This stage uses a Constant Current Source (CCS) anode load and LED biasing (zener biasing optional), as this provides the lowest THD and maximum gain possible in this type of gain stage. The common cathode stage, however, has relatively high output impedance, hence, isn’t suitable for driving a 300B tube directly. To provide a good, low-impedance drive for the 300B, a cathode follower (TU1A) is used. The combination of a grounded cathode stage and cathode follower stage has the added benefit of providing distortion cancellation. This is because the two stages create similar distortion components, but those of the grounded cathode stage are in opposite phase to those of the cathode follower stage. It seems counter-intuitive, but slightly increasing the THD of the cathode follower actually reduces the THD of the two stages combined. The circuit has been optimized for the lowest THD possible with the 6N6P tube. Q1 works as a source follower and provides a stable, filtered supply voltage for TU1A.
The CCS is comprised of U1 and associated components. The ratio between R1 and R2 sets the current, C2 is added to reduce noise, and C1 is required to ensure good stability and a smooth transient response of the CCS. The cascode, Q2, protects the CCS from over-voltage. This also dramatically increases the output impedance of the CCS, hence, minimizing the THD and maximizing the gain of the input stage formed by TU1B.
In addition to providing a good, low-impedance driver for the 300B output tube, cathode follower TU1A ensures that the coupling capacitor C4 remains fully charged at all times. As the cathode follower draws no grid current, the charge on C4, hence, the DC voltage across it and, thus, the bias voltage for the 300B all remain constant. This eliminates blocking distortion (what guitar players refer to as “farting out”).
The 300B is biased using fixed bias – i.e. a constant DC voltage applied between grid and cathode. This voltage is obtained by the voltage divider formed by resistors R12, R14, and trimpot R13. The grid-to-cathode voltage on the 300B is adjustable from approx. -75 V to -95 V. This should be suitable for most 300B tubes at most 300B operating points. The adjustment range can be expanded by lowering R12 and R14. The PCB layout is optimized for ease of bias adjustment. I.e. turning trimpot, R13, all the way counter-clockwise (CCW) brings the 300B bias current to the minimum value. The 300B bias current may be measured by measuring the voltage on connector J2. 10 mV on J2 corresponds to a 1 mA bias current in the 300B.
The board measures 6.4 × 4.0 inches. There are five mounting holes on the board and all components that are intended to poke through the chassis (primarily the mounting screws and tube sockets) are arranged symmetrically around the center mounting hole. In addition, the two bias adjustment potentiometers are arranged symmetrically. The intent here is to allow the user to adjust the bias through a hole drilled in the chassis without having these holes detract from the aesthetics of the amplifier. The mounting holes are 130 mil (3.3 mm) in diameter, intended for use with M3 or #4 machine screws.
The circuit is designed for use with three different tubes; Russian 6N6P, JJ ECC99, and 12BH7A. With some creative wiring, a quartet of d3A configured as triodes can be used as well. However, do note that changing from one tube to the other does require a few component changes and wire link options to be moved, as the tubes are biased slightly differently and have different pinouts for the heaters. The component changes are highlighted in the BOM. The circuit is primarily optimized for use with 6N6P and ECC99. These tubes are quite similar in electrical characteristics. Personally, I ended up preferring the 6N6P over the ECC99. The ECC99 is the more precise sounding of the two, but I found the 6N6P more pleasing to the ear. The 12BH7A is an excellent performer as well. This tube has more of a classic tube sound to it. I found it to be a bit warmer than the 6N6P, ECC99 but nowhere near as precise. The d3A is a good middle ground between the 12BH7A and the ECC99, 6N6P tubes. If you do pull off this stunt and wire in a quad of triode strapped d3As in place of the two dual triodes, use 100 Ω, 0.5 W from G2 to the anode and keep the connections from the board to the tubes as short as possible.
The pinout of the 6N6P tube is the B9AJ type where pin 9 is used for a electrostatic shield between the two triode sections. ECC99 and 12BH7A is the B9A pinout where pin 9 is the center point of the 12.6 V heater. Hence, to support both types in this board, a wire link option was introduced. With ECC99, 12BH7A (or d3A) connect options OPT1A and OPT2A. For use with 6N6P tubes, connect wire options OPT1B and OPT2B. The two configurations are illustrated below. Note that in addition to the wire options, a few component substitutions are required to switch from one tube to the next. Hence, these options should be permanently wired and not brought out to a selector switch.
There is one capacitor in the signal path, C4. The board has been designed to allow enough room to use a quality polypropylene capacitor in this location. Personally, I use a 220 nF, 630 V polypropylene cap from Solen in this location. I get them through AES. The Solens are quite large, so I’ve overlapped the footprint for the large Solen cap with one that can be used with a general purpose polypropylene cap. Some may frown at the use of a capacitor in the signal path. However, my experience is, that if a quality component is used, the capacitor does not impact the sound of the amplifier. By “quality” I mean a capacitor type that does not suffer from dielectric absorption and has very low parasitic resistance (ESR), inductance (ESL). Most polypropylene capacitors fall into this category. Some improvement in sound quality may be obtained by selecting a capacitor that has been designed to minimize the ESR, ESL components. The Solen caps fit this bill nicely.
The input transformer is a Jensen JT-11P1-HPC. If a differential input is desired, a transformer is very, very hard to beat on performance. But if you’re on a budget, or don’t need the differential input, the board offers the option of using an AC coupling capacitor, C3, on the input. To minimize the board area required by these options, the capacitor and transformer share the same space. Hence, it is not possible to use both transformer and capacitor inputs simultaneously. As with C4, the footprint for C3 allows for the use of either a bigger high-quality cap, or a general purpose kind. This to allow builders to choose what suits their personal preferences and budgets.
A note on LED biasing
With a CCS loaded grounded cathode stage (such as TU1B), the highest gain and lowest distortion is obtained when an ideal voltage source is used in the cathode bias circuit. Of course, ideal components aren’t readily available, so compromises will have to be made. Several choices for voltage sources available; batteries, bandgaps (such as LEDs), and zener/avalanche breakdown diodes. Of those, the LEDs and zeners provide the lowest dynamic resistance, i.e. small-signal voltage variation versus small-signal current variation or ∂V/∂i, thus, are as close to the ideal voltage source (∂V/∂i = 0) as possible. The dynamic resistance of an LED or zener does vary with bias current. And it does vary from manufacturer to manufacturer – especially for the LEDs. Scouring through a stack of LED datasheets led to the selection of the HLMP-series. The green HLMP-3507 used in this design have a dynamic resistance of approx. 20 Ω and provide a 2.0 V bandgap voltage.
Should you prefer to use resistor bias on the input stage, I suggest using the footprints for the zener diodes for the resistor and fitting the bypass capacitor on the footprints for the LEDs. Options, options….
The board comes together pretty easily. I suggest populating the 1/4 W resistors first. Then, while the board can still lay flat on the workbench, populate U1 and U2. From there, populate the rest of the board.
The MOS devices, Q1 through Q4 will need to be isolated from the heat sinks as their drains (tabs) are at different potentials. Failure to do this will short out the CCS and likely result in melt-down of the input tube. Use the proper shoulder washers, isolating washer, and hardware for mounting the devices to the heat sinks. Personally, I use the silicone pads for the isolating washer, but mica and thermal grease can be used as well.
The tube sockets go on the back side of the board, i.e. the side without silk screen. The bias trimpots may be mounted on the component side of the board, or the back side of the board. The latter make them possible to reach through holes in the top plate of the chassis for easy bias adjustment.
Once fully assembled, connect the driver board to the filament supplies and power supplies. Turn the bias trimpots all the way counter-clockwise (minimum bias current). I suggest powering the board up without the tubes populated at first. Verify that the correct heater/filament voltages are present at the correct tube pins. Verify the bias supply voltage (-225 V) is present on the grid pin of the 300B socket. Verify that B+ is present at the anode pin of the 300B socket. There should be roughly 85 V on pin 1 of the socket for TU1 and TU3; and roughly B+ on pin 6. If all this checks out, power off the supply and wait for the reservoir caps to fully discharge before proceeding. Now plug in tubes TU1, TU3 and power back up. Verify that the DC voltages fall roughly as indicated on the schematic. In particular, verify the grid voltage on the 300B to be -90 V or so. By turning the bias trimpot, the grid voltage on the 300B should change from approx. -70 V to -90 V. Ensure that the bias trimpot is turned all the way counter-clockwise (minimum bias). This should be the end of the trimpot that results in -90 V on the grid of the 300B. Power off the amp.
Now for the bias adjustment you will need two DC voltmeters. If you only have one, I suggest proceeding one channel at a time. Plug in the 300B tubes. Connect one DC voltmeter to J2 and one to J6. Connect a load resistor (or speaker) to the amp output and turn on the amp. The voltage measured here will be 10× the bias current. I.e. if the voltage on J2 is 850 mV, 85 mA is flowing in TU2. Adjust the trimpots for the two channels until you reach the desired bias current. This current is generally in the range of 60 mA to 100 mA. Personally, I use 85 mA with a B+ of 400 V. Check the bias periodically and re-tweak until the desired bias current remains stable (or within a few percent of the target current) for at least 15 minutes. With new tubes, you may have to re-tweak the bias after 8~10 hours of operation. After that, the bias point should remain fairly stable.
The amp is now ready for use.
If you are unable to reach the desired bias current, change R12 or R14. Lowering R12 will make it possible to reach a higher bias current; lowering R14 will extend the adjustment range towards lower bias currents.
Output Tube Options
This board has been optimized for use with a 300B tube. However, it could easily be used as a driver for other directly heated triode tubes as well. The table below shows the suggested component values for use with various DHTs. Except for the values for the 300B, of course, these values are not tested, but should be viewed as a good starting point. They were calculated from the operating points listed in the tube data sheets.
|Output Tube||B+||Vgk @ idle||D1, D6||R10, R24||R12, R26||R14, R28|
|300B||400 V||-85 V||HLMP-3507||33 kΩ||360 kΩ||680 kΩ|
|2A3||250 V *||-45 V||SHORT||39 kΩ||270 kΩ||1.2 MΩ|
|45||275 V *||-56 V||SHORT||39 kΩ||300 kΩ||1 MΩ|
|801A||600 V **||-55 V||HLMP-3507||39 kΩ||300 kΩ||1 MΩ|
|211||1.25 kV **||-80 V||HLMP-3507||33 kΩ||330 kΩ||620 kΩ|
*: This board is optimized for operation at 400 V with a 300B tube. Lowering the B+ voltage a bit – to, say 350 V – is no big deal. But operation at lower voltages is likely to push the headroom on the CCS on the driver tube, TU1B. To gain a bit more headroom, D1 is replaced by a wire jumper (i.e. shorted out). This lowers the anode voltage on the driver tube and allows for full, undistorted signal swing even at the 250~275 V B+ voltages.
**: Note that if assembled according to the BOM, this board will only support a B+ voltage up to about 400 V. Maybe 425 V if you’re using a good voltage regulator for the B+ such as my 21st Century Maida Regulator. Using B+ voltages above 400 V for the output tube will necessitate a separate 400 V supply for this driver circuit.