Giant UV211 Push Pull Amplifier

Background

The reference UV211 Push-Pull Power Amplifier circuit that was originally published in an article named as "Acoustic Sound 電聲界" in about 70s --- it was too long ago to be remembered the exact date when I was still a secondary student.

The first stage is a simple voltage amplifier which is connected as a triode using an ultra low noise pentode EF86, in the common cathode configuration. A 100K loading resistor is used to obtain a high voltage gain and linearity, i.e. lower distortion. The cathode resistor is bypassed by a 560uf capacitor to lower the output impedance of this stage.

The second stage is a long-tail differential amplifier for further amplify the signal and provide two out-of-phase driving signals for the subsequent push-pull stages. ECC813 is a high transconductance (gm), low gain (mu) and low plate resistance (rp) twin triode.

The third stage is a cathode follower which is connected by two ECC813 in parallel in order to boost the current sourcing capability for driving the UV211's grid beyond the positive region.

The final stage is configured by two pieces of UV211 and a 8K 100W P-P output transformer. The high tension voltage is about 1000V which is so dangerous for non-experienced personnel, it is an absolute MUST to take care in wiring up and testing this amplifier or it will blow you dead!

However, due to the handy tubes and components that I collected for the past 20 years, I decide to modify and design a new circuit using my existing components with reference to the basic schematic shown above. The tubes are: ECC40 --> ECC813 --> 6BL7 x2 --> UV211 x2 --> Tamura F2012 5K 100W OPT.

For optimum loading of two UV211 connecting in push pull is about 10K, how can a 5K OPT be matched with the UV211 ? The solution is because of the high quality and specifications of the famous Tamura output transformer. Refer to the specifications of the F2012, the inductance of the primary is about 150H. The required minimum inductance for the OPT to sustain a 10K P-P impedance at 20Hz is about 80H, i.e. F2012 is about double the minimum. Hence, by connecting the 8 Ohm speaker to the 4 Ohm winding of the F2012, the reflected primary impedance is 10K, which will solve the matching problem. However, this will be further confirmed in the measurement.


Circuit Redesign







I have 4 pieces of Philips Miniwatt ECC40 in original box and some Mullard CV3884 in Government Box. The first stage is redesigned using ECC40 connected in parallel. ECC40 is an indirect heater twin triode. This valve is primarily intended for use as an AF amplifier, phase inverter as well as electronic counters and calculating machine in the good old days. You can see that the Ip-Vp characteristic is quite linear and evenly spaced. By observation, two sections of ECC40 connecting in parallel is very similar to the characteristic of an ECC82. However, the rare tube base (B8A) makes it not so commonly used in DIY circle.

Since this amplifier is intended to be non feedback, linearity and distortion are my major considerations. Most of the system gain is contributed by the first stage. Hence, a 200K plate loading resistance is quite high enough to achieve my intention of getting the entire gain and linearity of this tube. By drawing a 200K load-line against a 350V plate supply voltage on the graph, getting the plate current of 1.3ma at -2.4V grid bias voltage, i.e. 1846 Ohm cathode resistor. Because we are going to connect the ECC40 in parallel, the cathode resistor and plate resistor are halved to 920 Ohm and 100K respectively in order to obtain the proper operating current of 1.3ma passing through each section of the ECC40. In steady state, the plate voltage of ECC40 is sitting about at 94V, which is quite reasonable for biasing the next stage. You can observe that the plate voltage of this stage will influence the choice of possible value of the second stage's cathode resistor or plate current. By rule of thumb, 100V for the plate voltage of the first stage is fine enough.

The second stage will follow the original design, using one ECC813 / 6463 to implement the function of an phase inverter. According to the analysis of a long-tail differential amplifier, the larger the value of the common cathode resistance can achieve higher common mode rejection ratio as well as the symmetry of both output. That's why in nowadays tube circuit design will use a constant current source to replace the common cathode resistor, in which the constant current source (CCS) can provide a very large ac impedance to give it the advantages.

In my circuit, the common cathode resistor also serves one of the major purposes of biasing both sections of the tube in optimum operating condition. Since the plate voltage of the first stage is about 94V that will maintain the common cathode voltage at about 104V. Since the plate supply voltage is being designed at about 450V, that means the plate to cathode voltage is about 346V. By drawing a load-line of 40K on the ECC813's Ia-Va graph and then locate the bias plate current of 3.5mA with grid biased at -10V. The voltage drop of the common cathode resistor is 94V + 10V = 104V. That means the common cathode resistor is 104V/(3.5ma x2) = 15K.

From the simple process of load-line drawing and calculation, one may ask me what is the primary procedures to determine the optimum operating points, the supply voltage, the selection of operating current, the loading resistor and the biasing grid voltage of an amplifier. This is quite a difficult question to tell. This is a long, repetitive, trail an error and experience. Actually, most of the fundamental design concepts are based on the electronic theories, you may get them from the text books, magazine and web. We can't make an amplifier to work violating the basic theories. For an exaggerated example, you can't make a diode to conduct in reverse biased, except it is broken down.

When you look at the Ia-Va graph of a tube characteristic, there are tremendous possibilities to bias the tube to work properly within the feasible operating region. So designing a tube amplifier is an art rather then a technical matter within the bounded electronic theories.

One may determine the supply voltage first, the plate resistor, the plate current and then from the operating point to get the required value of the cathode resistor.

One may determine the bias plate current first, the plate resistor, the bias grid voltage, the cathode resistor and then to get the required supply voltage.

So there are many strategies can be used, you may adopt your own favorable strategies and procedures, in which other important issues have to be considered and reconsidered, such as the expected voltage gain, maximum swing voltage, output impedance, driving current capability, frequency response, noise, distortion, the loading effect of the subsequent stage and stability ....... etc. Hence my own suggestion is: design, build, measure, test and redesign again and again.

I am very busy right now, writing up this post is pending...........