GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO PHOTO 1: 6336-A all-triode push-pull amp, front view. 6336-A-BASED ALL TRIODE P.P. AMP This amp design features the 6336-A power triode tube, along with push-pull class A operation, no overall feedback, low distortion, approximately 20W of output power, reliability, and high-performance sound. BY KEES HEUVELMAN is an easy-to-use medium-gain tube ductance, which is the degree to which & WIM DE HAAN with a very low Rp. Moreover, it’s low- the flow of current through the device priced and easy obtainable. changes as a response to a change in ne of my favorite small signal grid voltage relative to the cathode. Otubes is the 5687, which I have INPUT STAGE By dividing the idle current by the used in several designs with very good In this design (Photos 1 and 2) I wished transconductance, an input overload results. Its specs are outstanding and to use a different circuit as input stage. voltage results (Vth). If this Vth volt- the sound is therefore really musical. Some time ago I ordered the Tube age is exceeded in a negatively charged The result can be a very dynamic and Cad Circuit simulation program and way relative to the cathode voltage, evolving/bold sound. played with it on many occasions. This the tube stops conducting any current. I generally use the 5687 made by tool provides many interesting details, This voltage serves as a rough figure of General Electric, but I have also used including circuits, one of which is the merit for a tube circuit, as the higher the Philips/ECG with success. In pre- so-called current sourced grounded the value, usually the better the sound. vious applications I used a very basic cathode amplifier. In this circuit the Because this Vth is much higher than grounded cathode circuit with standard plate resistor is replaced with a triode- in my standard grounded cathode cir- values—Ra=15K and Rk=680E—and based current source, which results in cuit, I chose this circuit. The basic idea with a decoupled cathode resistor. I more gain and better PSRR. for the circuit is shown in Fig. 1. achieved the best results with the two According to Tube Cad, each triode In other applications I noticed that systems in parallel. The 5687 tube exhibits a certain amount of transcon- a mu-follower (looks quite similar to May 2005 25 GA505-whole.indd 25 3/23/2005 1:33:41 PM GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO 26 May 2005 GA505-whole.indd 26 3/23/2005 1:33:52 PM GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO May 2005 27 GA505-whole.indd 27 3/23/2005 1:34:01 PM GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO the used circuit) with the signal taken LEDs as voltage reference and added a The 300B tube is out-of-budget, and, from bottom triode anode sounded potentiometer P2 to adjust the circuit for example, a 50C-A10 is expensive surprisingly better than if taken from for current and for minimal distortion. and rare. The 6AS7/6080 is an option. the cathode of the upper triode—this I used an additional potentiometer P1 However, after listening to a 6336-A- was confirmed by several other hobby- to adjust the AC symmetry/balance of based amplifier and knowing that I ists. The relatively high output imped- the long tail. had several of them somewhere in a ance of the circuit is less important, In the original concept the current shoebox, I made my choice just like because the following driver stage will source was intended to be connected that. have a high input-impedance. to a negative voltage. However, to keep The 6336-A tube is intended for Throughout this amplifier, I used things simple, and because of good re- power supply use and contains two WIMA FKP caps as coupling capaci- sults with the prototype, I built the separate triode systems (Fig. 2). Mu is tors. These low-priced capacitors give final amplifier without using this op- as low as 2 and power consumption is very good sonic results. Feel free to tion. Reference voltage is taken from as high as 5A at 6.3V; maximum plate use more “dedicated” audio capaci- the main high-voltage supply using a dissipation is 30W. Because plate re- tors such as Hovland, MIT, or Jensen, simple resistor. sistance is very low, the 6336-A tube which I’ve used before, but this time I The 5687 input circuit and the can be used in output transformer- stayed with the FKPs. 6SN7-based phase splitter yield enough less (OTL) circuits. Another choice is gain to make the overall circuit work the 6528, which is very similar to the PHASE SPLITTER without overall feedback. Because of 6336-A, but mu is 9. This would make I chose a long-tail circuit as phase the relatively low gain, feedback is it easier to drive, though this tube is splitter/driver stage. This widely used not possible using this configuration, very rare. Strangely enough, not too circuit yields fair gain and fair bal- though one of the design goals was no many circuits can be found using the ance—it is perfect for the job. In this use of overall feedback. Overall gain/ 6336-A—not on the net, in maga- application I chose the 6SN7 in a con- sensitivity of the design is about 20dB, zines, or other publications. However, figuration with a current source, which which is a good figure. I know that Luxman designed a push- I built around a small transistor with pull amplifier (the KMQ80) using the OUTPUT STAGE 6336-A; unfortunately I was unable to In searching for obtain a copy of the circuit diagram. the perfect out- I used the 6336-A in full class-A put tube, I dis- using cathode bias. It would work covered that ini- perfectly well with low primary im- tially the choice pedance to yield high output power; was more difficult however, my goal was for a high 8k than expected. load for low-distortion figures with The transmitter less output power. Output power is tubes are not my less important. choice; I consider Because output transformers are the very high volt- sensitive to DC unbalance, I added a FIGURE 1: Basic circuit of the 6336-A based amplifier. ages a problem. small meter and a variable resistor to PHOTO 2: 6336-A all-triode push-pull amp, top view. FIGURE 2: Power supply 6336 amplifier for two channels. 28 May 2005 GA505-whole.indd 28 3/23/2005 1:34:09 PM GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO • GLASS AUDIO the cathode circuit to solve this prob- Using the tube’s Chatham datasheet, to “implant” the 25W resistors. The lem. Selecting the 6336-A for balance I selected a 1450Ω resistor for biasing 2mm aluminum main chassis is used would be impossible to do. This circuit the 6336-A, thus using an effective as a heatsink; the eight 25W resistors is optimized for toroidal output trans- anode voltage of 335V (power supply, are mounted on an aluminum u-bar formers, such as those manufactured cathode bias). In this case, bias voltage that is affixed to the chassis. Airflow by Amplimo/Plitron. would be around 114V; bias current should be more than adequate; but Due to the high transconductance, would be approximately 70–80mA. you might consider installing a small cathode bias is the only option. Nega- The prototype used two 680Ω resis- fan just to keep the air flowing. I chose tive bias will not give the 6336-A the tors in series as cathode resistors; bal- stability it needs. In my attempts to ancing was done with a tandem 100Ω use negative bias, I had no good re- potentiometer (Fig. 3). Because these sults. With cathode bias, the 6336-A are hard to find, I chose the configura- worked without problems and stability tion as drawn in the final circuit (Fig. was excellent. 4). You must take special care with the In the final design, the mains trans- cathode resistors of the 6336-A. The former has an option for four second- high bias voltage and current will ary windings: 320V, 330V, 340V, and produce enormous heat; each resistor 350V. I set the high voltage using the will need to dissipate more than 7W. 320V winding, yielding a +410V DC Because of the class-A operation, the voltage with all loads attached. The 6336-A becomes really hot, so pre- amplifier worked very well and beyond heating the tube for some minutes be- all expectations. Bias voltage is around fore switching the main high-voltage 100V, giving a 70mA bias current per is necessary. triode. Because of the high bias voltage and FIGURE 3: Alternative output circuit BIASING THE 6336-A current, the bias resistors dissipate a 6336-A using a tandem 100Ω In the original concept, the 6336-A serious amount of heat.