A guy asked on the Swedish forum if it was possible to come up with a forum designed tube amplifier that is easy to build and with reasonable power output. I decided to join in and contribute with a design. My first idea was to use PL519 in screen driven mode but after a while I changed to a more easily built design. One of my goals was to reduce the number of low frequency poles to as few as possible. Tube amplifiers with negative feedback are not known for their bass reproduction and the reason is usually the excess amount of poles within the loop.


A way to get rid of poles is to use differential amplifier stages and DC connection where possible. Thus, this amplifier is fully differential from input to output. This first design is intended for unbalanced signal input and has a single feedback network but it is easy to convert to differential input and feedback (if the output transformer has a 16 ohm winding with 4 ohm tap). There are only 2 low frequency poles in the loop, the transformer and the coupling capacitors to the output tubes. While it is possible to DC connect the output tubes it would complicate the power supply too much to be an easily built design.


The design is a classical in disguise, a single input stage DC-connected to a cathodyne phase splitter much like my old EL34 design. I replaced the single triode input with a differential stage with a constant current sink in the tail. This stage takes over the role of phase splitting from the cathodyne so it could be converted to a fully differential stage with a resistor in the tail. These two stages are DC-coupled.


Here is the current schematic: (Note: Commercial use of this design is not allowed without my permission)


Figure 1 Schematic of the HFTA2.


The schematic is still preliminary and subject to modifications but a prototype is running and works just fine. A lot of measurements are left to do to verify the design. This version has been built and was demonstrated at a DIY meeting in Gothenburg Apr 2005. The tubes in the prototype are RCA 5691 for V1, Dumont 6SN7 for V2 and Radiotron EL34 (Philips dual D-getter X4f), reasonably matched.


The input stage uses 6SL7 and a constant current sink with two standard NPN transistors in a cascode configuration and a red LED as reference. The sink is set to 2.5mA with R102 which should be adjustable. Input signal goes to the upper triode and feedback to the lower. P1 is used to balance the stage. The choice to use semiconductors in the current sink is based on the usage of a standard power transformer, else a tube sink can be used here.


The driver stage is lifted up above ground to allow DC-connection to the input. The diodes are for protection during warm-up, these can be omitted if the power supply has slow start. While it is possible to use a constant current sink here I decided that it was a bit over the top for a simple design. I have found it to be the best to adjust P1 for a good balance between the anodes of V2. The diodes on the driver stage protects V2 during warm-up if the power supply has no delayed HT.


R21 and C5 is the phase compensation for high frequency. It is set for output transformers having its first HF pole around 53 kHz. C4 rounds off the HF response to avoid ringing. The first pole is set to around 19.5kHz. Phase margin is 70 degrees and the phase shift at 20kHz is 22 degrees.


The output stage is an ordinary fixed bias push-pull. With 7 kohm output transformer and with bias at 37mA per tube, the output power is 30W, all the way down to 20Hz. Itís not optimal yet, some adjustments will be done.


As said above the design is in a preliminary stage. The bias network is a quick hack using what I had in the junk box and will be redesigned. Nevertheless, the amplifier sounds good and has a deep, quick and dry bass, very much unlike classical designs that often have a fat and soft bass. Jazz music with double bass is a joy to listen to. The highs are detailed and easy to listen to.