6N3 tube and transistor combined to carry single-ended amplifier

 
   1. Joint Loading
     The so-called "Joint Loading" structure is a new name given by Mr. Liang Jiandong, a fellow enthusiast, to circuit structures such as 6N3+2SC5200; it is different from the "front gallbladder and back stone" and "STC" structures. In the gallbladder and back stone "joint loading" structure, the electronic triode and the transistor are directly combined to share the load. Therefore, the low internal resistance characteristic of the electronic triode becomes the inherent characteristic of the composite device, directly providing high damping to the load. In the traditional "front gallbladder and back stone" circuit, the signal transmission path is first through the electronic tube and then through the transistor, and the component that actually bears the load is only the transistor. Since the internal resistance of the transistor is very high, it is impossible to directly provide sufficient damping to the load without negative feedback. "STC" should refer specifically to the direct composite circuit of the electronic triode and the electronic multipole, which can be regarded as a special case of the "joint loading" circuit.
    If the transistor and the electronic triode connected in the "joint load" mode are regarded as a whole, it is a voltage-controlled current output device with high input impedance (A1 class), high linearity, low internal resistance, and large current, and has the typical unsaturated output characteristics of the electronic triode. Therefore, the high damping of the "joint load" structure to the load is achieved by its own low internal resistance, rather than by negative feedback; it is easy to obtain a relaxed and natural sound when used in the power amplifier circuit.


    The above figure shows the output characteristic curves of 6N3, 6N3 and 2SC5200 after compounding, and the two shapes are almost the same. If you observe more carefully, you will find that the curve of the joint load circuit in the A2 part will be more linear than the original curve of 6N3. In this example of the joint load circuit of 6N3 electronic tube and transistor, the linearity in the A2 state is improved. The reason should be that the gate current and cathode current are injected into the base of the transistor at the same time, expanding the current of the collector of the transistor; the appearance of the gate current in the A2 state compensates for the problem of compression of the A2 curve area of ​​6N3 electronic tube to a certain extent, and finally improves the linearity of the joint load circuit.
    The measured gate current of 6N3 is as follows:
2. Driving circuit:
    I chose the 4556 op amp with a larger current output to drive the gate of 6N3 to deal with the gate current. Although within the designed working range, the gate current of 6N3 does not exceed 5mA. If you use the 5532 with a nominal maximum output current of 38mA, the sound is dry and rough. If you use the 4556 with a nominal maximum output current of 70mA, the sound is obviously bright and moist (although I can't see any difference on my oscilloscope). I believe there must be other op amps with better performance to replace 4556, you may as well try it yourself. The functions of the op amp are as follows:
1. Input signal buffering. The ratio of R3 and R4 determines the basic gain. I personally think that 5 to 10 times is appropriate.
2. Provide gate bias voltage for the electron tube. According to the output characteristic curve of the composite circuit, the gate voltage of the electron tube will be roughly in the range of 0 to -2V according to different main power supply voltage, load impedance, output power and other requirements. The generation of negative gate voltage is achieved by controlling the potential of the inverting input terminal of the op amp. The potential of the inverting input terminal of the op amp is obtained through the voltage divider circuit composed of R5 and R6. Adjusting R6 can allow the output of the op amp to provide the appropriate gate voltage to the electron tube.
3. Provide the gate current required for the A2 working condition of the electron tube. Since the electron tube in this design will enter the A2 state with gate current when the input signal is slightly larger, the input impedance will decrease sharply. In the designed working range, the minimum impedance of the gate ground will be about 300Ω, and the gate current at this time is close to 5mA. R7 is the load of the op amp. In order to balance the output current of the positive and negative half-cycle of the signal under the A2 working condition, the value of R7 is relatively low.
 
3. Power supply:
1. The main power supply of DC80V1A is obtained by rectifying and filtering the AC power of about AC60V. CLC filtering or voltage stabilization must be performed. The actual measurement of the CLC configuration of 4700uF+100mH+4700uF has no obvious 50/100Hz AC sound. When using 80V main power supply voltage and static current of about 0.35A, the output power reaches 10W and the distortion is less than 5%, which is no problem. You can also choose other working points according to the power supply at hand. When using ordinary switching power supply as the main power supply, my experience is that the sound quality of the above CLC filter circuit will be better.
2. After rectification and filtering, the 6.3V0.35A filament power supply can be stabilized with a LM317. Pay attention to adding appropriate heat sinks. In fact, no noise can be heard when using 6.3V industrial frequency AC power supply in this circuit.
3. The DC±15V0.1A op amp power supply has a good effect with the simplest 78/79L15 three-terminal voltage regulator. If you have extreme requirements, please choose a better voltage regulator circuit.
 
    In order to simplify the power supply, I once tried to use a DC-DC module with a wide voltage input, with a nominal output of 5V1.6A and ±12V0.3A. After the modification, the output was 6.3V and ±15V for the filament and op amp respectively. In this way, the whole machine can be powered by a single DC80V power supply. However, the use of DC-DC modules will add new problems: it is necessary to eliminate the influence of DC-DC module noise on the operation of the op amp, which is more troublesome. Therefore, the public circuit diagram still uses a traditional power supply. The circuit diagram and PCB files will also be continuously improved and updated.
    The following figure is a sample made when using a DC-DC module.
 
 
4. PCB
    In the shared project files, I provide three PCB files: a minimalist version, a DC/DC module power supply version, and a conventional power supply version. Friends can directly use them to open the board. Due to the size of the board, I did not draw the VCC-related rectification and CLC filtering circuits on the PCB. Please build a shed or draw a board by yourself. At present, I have chosen the most common plug-in component package on the PCB, just to start a discussion. You can change the component package at will, or even adjust the circuit to suit your needs. If you choose high-end components and PCB materials, the sound quality of the whole machine will definitely be better.
    The following figure is a PCB sample using a conventional power supply version:
 
5. Output transformer:    
    After multiple verifications, the single-ended output transformer with a design impedance ratio of 168:8, a quiescent current of 0.4A, and a power of about 10W is very suitable for this circuit, and is also suitable for many common "joint load" circuits of electron tubes and transistors. As an important fixed asset investment in power amplifiers, this output transformer can have more applications.
    I used to think that low-impedance output transformers were relatively easy to wind; I also customized several pairs from a famous cattle breeder a few years ago. Unexpectedly, after the master wound a wire package with the experience of winding tube amplifier output transformers, he measured that the attenuation of 15KHz had reached -1dB. Since there was no time to continue to optimize the winding parameters, he returned the order. In subsequent communications, we believe that the design experience of high-impedance transformers for tube amplifiers is not completely applicable to the production of low-impedance transformers; how to adjust them specifically requires multiple practices and tests.  
    I have several pairs of broadcast amplifier output transformers from KOHSEL of Denmark on hand. After modification, they are just suitable for the needs of this circuit, and the frequency response and sound quality are satisfactory. I have shown the measured data in a forum, you can check it out. After all, this Danish output transformer is just a coincidence. I hope that an expert can design and produce a low-resistance output transformer with better performance!
    Emphasize: In the primary circuit of the output transformer, a fuse is necessary! Once, because the contact between the transistor and the heat sink was not tight, the transistor overheated and broke down during operation, and one of my output transformers started smoking. Therefore, good heat dissipation and the addition of fuses can ensure the safety of the circuit.
   
6. Debugging and others:
1. Do not plug in the op amp and the electron tube first, and check the voltage of each group on the PCB.
2. Plug in the op amp and adjust the bias potentiometer R6 so that the gate voltage of the two electron tubes is -3V (this is conducive to the safety of the next step of debugging, and it is about 0~ -2V in normal operation).
3. Attention! There is no coupling capacitor on the audio channel of this circuit, so it is required that there should be no DC signal at the input end, otherwise it will directly affect the final stage working point. If you do not trust the output state of the audio source device, you should add a pair of coupling (DC isolation) capacitors to the input end of the op amp.
4. Plug in the electron tube, adjust the main power supply to half of the predetermined working voltage with the programmable power supply, set CC to the predetermined quiescent current, connect the transistor and output transformer of one channel, and increase the voltage while observing the current until the predetermined working voltage is reached. If the quiescent current deviates from the predetermined value, correct it by adjusting R6. Repeat this step to debug another channel. If there is no programmable power supply, the quiescent current can also be calculated and monitored by measuring the voltage drop of the primary winding of the output transformer.
5. The two transistors must be fixed on the heat sink with thermal insulation sheets.
6. The transistor pins can be extended with leads to facilitate the placement of PCB and heat sink. According to the production experience of friends, the lead is as long as 20cm and no self-excitation is found. It can be seen that the stability of this circuit is still good.
7. Since there is no temperature compensation circuit, the quiescent current of the transistor will increase as the temperature of the radiator rises; when the temperature of the radiator is constant, the quiescent current is also basically constant. Friends who like traditional cooling methods can calculate the parameters of the radiator according to the actual static power consumption. I recommend the air-cooled radiator with an intelligent temperature control circuit. The biggest advantage is that the transistor can quickly reach thermal balance and stability (this is the determining factor for the stability of the quiescent current, which is simpler, more practical and more reliable than the temperature compensation circuit). Secondly, it can greatly reduce the size and weight of the cooling system. As for the legendary wind noise, it is a legend after all! Modern high-quality low-noise fans and air duct designs are trustworthy.
    The radiator I tested is an air-cooled CPU radiator. The 12V fan is temporarily connected to the filament and powered by 6.3V DC, and the wind noise is almost inaudible. When working, the radiator temperature is basically stable at around 50℃ (air temperature 25℃). Generally, thermal balance can be achieved within a few minutes of power on.
    Here is another picture of the stall. They have been working stably for hundreds of hours, and they should find a shell for them.
 
The frequency response curve when the measured output is about 9W (the power supply voltage at that time was about 70V, and the static current was about 0.4A):
 
The output waveform and FFT parameters when the measured output is about 9W (green is the output waveform attenuated by 10 times, and yellow is the input waveform):
 
    The current audition results show that using ordinary components, the sound of this circuit is not inferior to the general EL34 and KT88 single-ended amplifiers, and it is slightly superior at both high and low frequencies; it conforms to the characteristics of low internal resistance and high linearity of the "joint load-bearing" structure circuit.
    Because the circuit is simple and there are not many components, even if all the components of fever quality are selected, the cost will not be too high; it is also easy to reflect the quality of each high-quality component. I look forward to friends working together to continuously improve this circuit, add bricks and tiles together, and work together.
I wish you all have fun!