In other words, these harmonics are “harmonious,” unlike the odd harmonics generated by the push-pull. Do not underestimate your ears’ judgment. This sense is one of the most advanced that Mother Nature has granted us, detecting micrometric details that you would not believe existed. Odd harmonics are disturbing, even if in almost unmeasurable quantities. As a result, you become tired of listening very soon.
I received several e-mails asking for a powerful single-ended amplifier. One of these correspondents owned some glorious speaker enclosures that are hard to drive (such as the AR-pi), with sensitivity around 80−85dB. Thus far he had been using a push-pull amplifier, but I think I convinced him that the single-ended is “something different.” I decided to use three 6C33CB in parallel, per channel, giving approximately 50W RMS output.
The Design
It took me some time to discover a suitable output transformer capable of handling the resulting quiescent current of 750mA (ten times the current required by a 300B in a single-ended configuration!) and, once this big obstacle was removed, I started designing the amp around this special OPT1.
Figure 1 represents the schematics of the amplifier sections and their power supplies. From an electrical point of view, these power supplies are “stacked,” that is, the final stage’s is “on the top” of the other2. These stacked power supplies meet at one point: the negative of the upper one (power tubes) touches the positive of the lower one (driver section).
This is a must and strictly means that you must not connect the negative of the upper one to the chassis ground; otherwise, a short circuit will result, blowing the fuse(s). Keep this in mind, and everything else is simple.
In order not to have a tremendously heavy single cabinet, I placed the driver unit with its dedicated power supply and the output stage on the main chassis, and connected the power supply for the 6C33C-B in a separate box with a twin wire to the main unit.
The first triode (half of the 6SN7GT) amplifies the signal at the input and feeds it to the second half without any blocking capacitor. This system is widely used, most notably in the famous Williamson amplifier.
I call your attention to the heart of the system, the anode load of the driver section of the 6SN7GT, namely, the resistor R6 (Fig. 1). This resistor feeds the 6C33 triplet in two ways: it supplies them the bias required (about 85V negative to grid, positive to cathode) as well as an AC signal of adequate swing (80 to 85V peak). To get this amplitude, you need about 1V RMS at the input, the amplification being 60−65 times.
Power Stage
While all anodes are directly connected together and to the primary of the OPT, the grids get the signal through small resistors (anything between 220Ω to 1k ¼W would do), in order to be somewhat separated (Fig. 1). As far as the cathodes of the 6C33s are concerned, I needed to fit a variable resistance network to “align” them, due to unavoidable differences in the valves’ characteristics.
The six 50Ω potentiometers, shunting the 100Ω fixed resistors, adjust the resulting resistance from 33Ω to zero. At its maximum level (33Ω) there is a drop of about 8V that represents an individual bias for each 6C33, in addition to the common negative voltage supplied by R6. It may be necessary to have, for instance, one pot set halfway, the second fully turned clockwise, and the third in between, in order to adjust the quiescent currents of the three 6C33s to almost the same level. (I would like to add, however, that, in practice, I did not notice a big change in the sound with as much as 20−30mA difference between the valves)3.
The 6C33C-B is a wonderful valve: Sturdy (designed for the Soviet Migs and tanks), it has such an unbelievably beautiful low internal resistance (80−100Ω); which makes a difference during bass amplification. As you know, the reactance of the primary coil tends to drop at low frequencies, and, if the internal resistance of the valve is high, most of the power is kept inside the valve and not transferred (as in a voltage divider attenuator)4. This is all you need to understand about the final stage layout, which is otherwise quite conventional.
There is nothing really special in the power supplies. Figure 1 shows this portion of the amp also. I used solid state diodes because they are handier, but managed to shunt them with small capacity, high-voltage capacitors to reduce the switching peaks. The transformer is a normal line insulation step-up type. Its rating should not be less than 450VA.
Figure 2 shows the frequency range (which is in excess of most available speaker systems); it is measured at 32W output. Figure 3 shows the distortion/power relationship. This is a very satisfactory result, considering that there is no overall feedback. Because my usual listening level does not exceed 10W (due to a relatively modest room), I enjoy a rate of about 1% distortion restricted to just the second harmonic (I found no trace of odd harmonics with the HP spectrum analyzer).
In Photo 1, the protruding knobs are the bias controls for the first section of the drivers, which also determine the common bias to the power tubes, and therefore set their average idle plate current. The trimmers that take care of a finer individual alignment of the 6C33s are next to these valves, in the back. The white octal socket, in the middle, contains six test points that measure the 6C33s plate currents, using an external portable multimeter; however, a removable box comprising a selector and a meter can be plugged into this socket, to ensure easy monitoring of the anode currents.
The huge boxes in the back enclose the powerful output transformers. The connections for the loudspeakers and to the power stage are on the hidden side of the boxes. I can remove or replace each output transformer in less than a minute, including the connection, by means of safety plugs and sockets.
Hints
1. Use generous-size chassis to accommodate this design.
2. The 6C33s generate a lot of heat, so do not place the capacitors close to them, but possibly underneath.
3. The two OPTs weigh 30 lb each! My solution was to make them easily removable with safety connectors (the same used with some portable multimeters) of different colors (Photo 3), so that taking them away or putting them back and connecting the wires takes less than one minute. Do the same and you won’t regret it.
4. The power supply for the output valves is split into two parts:
a)The first comprises the mains (115 or 230V) to 210−215V transformer, plus the rectifier bridge, the first filtering capacitor, and the 1200mA/5H choke all enclosed in a separate, well-ventilated box that I put on the floor. The connection to the main chassis is ensured by a twin wire cable with safety connectors (one white positive and the other black negative). Approximate weight is 40 lb (18kgs).
b)The second part of the power supply is settled in the back of the main chassis and includes four big (1000μF/385V) capacitors, each pair being fed from the B+ of about 300V through a 10Ω/5W resistor, used as a line separator (Fig. 1). This part also includes two filament transformers (12.6V−10A each) or just one (12.6V−20A). Considering that these transformers might generate heat, the best location is on the top of the chassis. I discarded the idea of putting them with the 450VA tranny in the same separate box, because of the high current they must convey to the valve heaters. The removable OPTs solution helps keep the main cabinet’s weight at a normal level for such an amplifier (36 lb.).
5. For the wiring, always use the same colors for the same functions. I suggest:
• Red for the 300V B+ line
• Brown for the negative of the 300V line (power tubes)
• Brown, also, for the B+ of the 400−425V positive line of the driver’s power supply
• Blue for the connections to theplates
• Green for the connections to thegrids
• Combined yellow/green colored wire for the connections to the main ground (metal chassis)
• Twisted bifilar yellow wire for the heaters
• White for any other unmentioned use.
Believe me, this color code will save you a lot of headaches and expensive errors. Once you memorize it, always use the same colors (no exceptions, please).
6. I strongly advise you to follow the recommended layout (including the values of the components, their rating, and their suggested location, particularly with regard to those with high heat dissipation) if you want to build this unit successfully. I have built three of these amplifiers so far (photos show my latest unit), and this is what my experience suggests.
Operating Tips
Most of the heating elements will find their better place outside the cabinet, as shown. As recommended, do not attempt to change the values of the components. Particularly critical are P1 and R3. Depending on the 6SN7GT characteristics, you may need to use a slightly higher or lower value for R3 in order to build across R6 the correct amount of bias for the 6C33C-B.
Once you have adjusted the plate current levels of the 6C33s (up to and including the warm-up time of about 15 minutes) by operating first on P1 and then on the individual 6C33C-Bs pots, your amp will be quite steady. However, if you choose to constantly witness its behavior, you can connect a 0.5 or a 1V fs meter to a six-position selector to measure the voltage across the 1Ω resistor test points. The reading in volts across these resistors indicates the milliamps (example: 200mV = 200mA) of plate current5.
Regarding the sound, every person who listened to the amp found it astonishing, but I would like to quote especially a sound engineer who is also a teacher of sound recording techniques in Turin, Italy: “Allow me to express my congratulations for the quality of the sound that your amplifier can reproduce; I have been able to enjoy the works that I know by heart [we had been listening to some of his recordings on CDs] and discovered such a living dimension very close to the one I received when I was listening to the musical instruments before placing the microphones.”
Bass is strong, damped, and dry. Not surprisingly, the three 6C33s in parallel mean a resulting plate resistance of about 90/3 = 30Ω, which is not too far from a solid-state device but with the advantage of the valve sound.
In sum, please note:
a) This amplifier is not for beginners, unless you take time to understand everything, not to rush, check and recheck what you have assembled, and so forth. Needless to say, you must have some electrical background know-how.
b) As an experienced builder, you know that some of the voltages are lethal. Never forget that. If you are a beginner, this warning is of primary importance. aX
References
1. This OPT is available at the following address: Bob@klingbergconsulting.com under the reference OPTOPUS-70-200 or King III.
2. The system is fully described in Plitron’s site quoting an article published by Glass Audio (17 novel audio amplifier circuits), available from Antique Electronics.
3. The circuit is based on the 6C33C-Bs working with a plate current between 180 and 240mA. Immediately after switching the high voltage on, the current is 100−120mA, but within 10 minutes it settles to 200−220mA, which is an excellent working condition, and it does not move from this level.
4. As an example, consider a 3H impedance primary under the frequency of 20Hz. According to the formula
2 × 3.14 × f × L = Z (Ω), its reactance would be 377Ω. With the plate resistance for one 6C33C-B at approximately 100Ω, the primary would get a share of 377/(100 + 377), or 79% of the signal amplitude. This is better than 70%, which corresponds to the loss of 3dB. With the three paralleled valves, the result has definitely improved!
5. The sequence to operate the amp is:
a. Make sure the standby switch is off (open).
b.Turn the main chassis switch on: this will enable the driver to operate and supply the bias required by the 6C33C-B in advance of the plate voltage. Also, the heaters of the 6C33s will start warming.
c.Turn the separate power supply box switch on.
d.About one minute later, turn the standby switch on (closed).
To shut down the amplifier, reverse these steps.
Author website: www.polisois-audio.com
This article was published in audioXpress, July 2004.
