The 833-A Stereo Amplifier

When I first began building single ended triode amplifiers the vacuum tubes of choice were the 300-B, the 2A3, and the 45. The most prized of these tubes put out only a watt or two and sound spectacularly musical through a high efficiency speaker. However, they do not have enough power for bass frequencies. For years, I admit, I contemplated, wouldn't it be a riot if somebody built a single ended triode amplifier using a quart-sized AM radio station transmitting tube. Eventually several people did and I wanted to build one. With much trial and error, combined with a close examination that revealed how simple it might be, I decided to build one.

Please do not attempt to build one of these unless you have a reasonably good idea of what you are doing. Less flamboyant do-it-yourself amplifiers have dangerous enough voltages in them and the 833-A in my design uses 1000 Volts with capacitor banks that store enough Coulombs to be deadly and some parts of the filter network cannot be grounded. Dangerous circuits must be properly caged in plastic to prevent accidental electrocution while the amplifier is running.

The sheer size of the system forced me to build it in several chassis and well-insulated wire to carry the 1000 Volts and double safety to ground the amplifiers is a safety must. Still, I wanted to build something that would capture the steampunk style and I drew inspiration from Jules Verne, The chassis that house the 833-A final stages and output transformers look like something out of a Jules Verne novel.

This is my story of how it all began:

I first became aware of the 833-A triode in Electronics
Warfare Officer School back in the 1970s, where the instructor of a basic electronics class told us of a problem he had with some neighbors keeping him awake at night with their loud sports shows on their television. Predictably, when he asked them to turn down the volume, the neighbor replied that he was entitled to play his television as loud as he wanted whenever he wanted. So he put together a simple Colpitts oscillator powered by a battery and using a transistor. The Colpitts oscillator has the advantage that because it center taps two tunable capacitors in series one can adjust the feedback along with the resonant frequency of the tank circuit. He tuned the Colpitts oscillator to the channel of the neighbor’s television sports playing station and jammed the TV signal. He gave the neighbor intermittent television malfunctions causing the neighbor to take the television to a repair shop where the shop could never duplicate the malfunction.

One night the ultimate sports game was scheduled, a final world series baseball game. The neighbor’s television was so loud that it was easy to hear the details of the game. At the last inning, the bases were loaded and the pitch would determine the outcome of this game and the world series, upon which the neighbor had wagered one thousand dollars. As the pitch was thrown, our instructor jammed the TV signal, causing the neighbor to fulminate so violently that he smashed the picture tube in his television.

This is what inspired me to build a Colpitts oscillator to take care of neighbors who play AM radio stations at loud volume in their driveways while working on their cars. However, I wanted something more powerful than a transistor and a 9 Volt battery. I chose a step-up transformer and a vacuum tube. The tube could act as its own half-wave rectifier and amplitude modulate a loud 60 Hz buzz. This is when I first discovered the 833-A triode, which could be powered by an 1800 Volt transformer. The tuning capacitors had have widely spaced plates to prevent the radio frequency voltage from arcing. The pair of tuning capacitors had to be adjusted not only to cause a loud buzz in the neighbor’s AM radio, but the feedback had to be adjusted to minimize the anode in the 833-A glowing red-hot. The antenna connection drew enough current into a long wire antenna that a 25 Watt lightbulb would light up fully when it was placed in series between the antenna connection and the antenna. Also it drew a fat blur spark when it was connected. Obviously I had to know enough about what I was doing not to tough any of the 1800 volts in the system. (If anybody is offended by this there is nothing they can do about it because it was more than the statute of limitations ago.)

It should be understood that American culture is so fanatical about its mission to convert the world and everybody in it to its radically extreme religion of top 40 pop/rock songs it had entitled itself to force everybody to hear and to demand that everybody not only accept, but to esteem as the great art form it pretended to be, as an integral part of the audio environment even in the privacy of everybody’ home. The fanaticism of this religion is so extreme that it makes the Islamic State look like the Unitarian Church; its faithful flock, have been programmed to believe that pop music is on a mission to save the world from all its problems. Thus, when I jammed the neighbor’s AM radio that was forcing its junk culture music down my throat, he reacted by ejaculating a series of imprecations and execrations so vile that it would make pubic lice blush.

The years passed and I had owned McIntosh amplifiers. During the last decade of the 20th Century, I read in Stereophile Magazine an article about the beautiful sound of the Cary 300-B single-ended triode amplifier, and later about how the 2A3, outclassed the 300-B, and the even more beautiful sounding 245 triode is the Stradivari of vacuum tubes, surpassing in pulchritude the richness any factory-built amplifier could achieve. I began my pilgrimage into the world of single-ended triode amplifiers, especially the 245 SET.

From the outset of my pilgrimage, I learned that such amplifiers that only put out a Watt or two needed efficient speakers and, notwithstanding their beautiful midrange and high frequency performance, they had nothing near the bass output it takes to match the sound of a live orchestra in an opera house. For several more years I accepted this limitation as a price I thought I had to pay for the beautiful mid and upper frequencies the 245 offered.

At this point, I remembered the 833-A and asked myself, wouldn’t it be a riot if some knuckle head built an 833-A single ended triode amplifier, never taking seriously the possibility that anyone would do such a thing as something other than a practical joke, or that I would join those who do, quoting what Virgil would have said, “Coitus speculum si non potest jocum tolerare.” But sooner or later, somebody did, the celebrated Japanese maestro, Nobo Shishido, who is the Raphael of the art of SET amplifier design. He built a pair of 833-A monoblocks that sell for $350,000. He designed them to run at over 100 Watts and in order thereto, he put 3000 Volts on the plates with an elaborate system for controlling the current through the 833-As through a special self-adjusting grid Voltage power supply with safety features to shut off the 3000 Volts if anything went wrong. Furthermore, the 833-As were driven through transformers by KT 88 tetrodes. Stereophile reviewed this amplifier as a great and mighty masterpiece. Perhaps the idea of an 833-A SET, which seemed like an exhibitionistic no balls no glory design for an amplifier was not such a joke after all. I knew that I wanted to have one in my home even though I had no use for such extreme power in my modest little Futuro house. My pilgrimage into the 833-A SET began.

Upon studying the data sheet on the 833-A, I made several discoveries that would allow me to build a greatly simplified design if I did not need the prodigious power of the Shishido design. A vacuum tube’s distortion varies monotonically with the power it is made to put out. This is because as the driving voltage deviates further from its no-signal grid bias the second derivative of plate voltage with respect to grid voltage increases making it less linear. Running an 833-A at far less than the power it can produce when driven to noticeable distortion will yield a very linear output. The 833-A has a directly heated thoriated tungsten cathode, which is recognized as the best sounding cathode in triode, with the exception of the oxide coated cathode in the 245 triode. What is especially convenient is that at a plate Voltage of about 1000, it runs best at zero grid bias. No bias power supply or cathode biasing resistor and grounding capacitor are necessary and that is something less to get in the way of the signal path. Furthermore, at this lower voltage, the 833-A can be driven through a transformer coupling by the 245. This would be a simple amplifier stage to execute.

Before I started this project, I had a bi-amped system that consisted of an amplifier with a 6SN7 choke loaded mu-follower directly driving two 245 output tubes. One output tube was connected by step-down transformer to the tweeter and midrange of a GR Research OB-7 three-way speaker and the other was connected to the woofer. An unexpected question was, how would it compare: the 6SN7 driving three 245s vs one of the 245s driving the tweeter and the other 245 driving the midrange and the transformer through a volume potentiometer that connected to the grid of the 833-A? The latter arrangement gave me a better ratio of output impedance driving input impedance. So I began by wiring one 245 to the tweeter and the other to the midrange and this made the mid and upper frequencies much clearer than anything I had ever heard before. The 245 is ideal for tweeters and midrange speakers. The 833-A would drive the woofers. When I finally got the 833-A final stages working I verified by listening with my ear near the midrange speakers that connecting the primary windings of the 833-A’s interstage transformer had no effect on how the midrange speakers sounded. All that was left to do was to use a test disc to check the polarity of the midrange speakers to be in phase with that of the woofers near the crossover frequency, which polarity made the highest reading on a sound pressure meter. This having been done, the 100k potentiometer could be adjusted to balance the woofers with the midrange, which required that the 833-As had to be turned way down. With this balance, which I can always fine-tune if a recording is bass-heavy, the speakers give very clear mid and upper frequencies and the bass has that authority, which can be excessive if you do not show the restraint of turning it down to match the frequency response of the upper frequencies, that duplicates the low frequency energy one can hear in a live performance at the local opera house.

Along my journey to my ideal dream sound system, something I could never buy even if I had the income of a television evangelist that would make it affordable, but something I had to build my way, (which is in the steampunk style that makes it look like something out of a Jules Verne novel) I performed several experiments the results I wish to share with you so that you don’t have to waste your time on some fads that do not work. The first one I tried was the use of light emitting diodes to bias the cathodes on the tubes that drive the 245s. I strung LEDs in series till I got the biasing Voltages of 5 and 10 that the 2C22s and the 27s in the preamplifier respectively. The sound was terrible, it suffered so much grain, I conjecture from the LEDs switching current draw to maintain a steady bias voltage, that I could never let anyone hear it. It would be embarrassing if I played it before guests who were visiting my home. When I reverted to the tried and true cathode biasing resistors with Black Gate electrolytic capacitors bypasses by much smaller silver foil in oil capacitors the improvement was so extreme, so smooth, so liquid, that I knew this was better than the sound I thought I wanted. Don’t waste your time with LED cathode bias. To drive the 245s I had been using 6SN7s in choke loaded mu-follower configuration. An audiophile website reported that several people praised the 2C22, also known as the 7193, as the king of driver tubes. Its characteristics are identical to that of each triode element in the 6SN7 except that the cathode to grid capacitance, and therefore the Miller effect, was less and this might have something to do with its higher frequency response and its noticeably clearer sound than the 6SN7, the 6J5, 6P5, or the 76. Not only is this a beautiful sounding triode, but its external anode and grid caps give it the appearance of a baby 833-A.

To drive this amplifier I use a separate single stage transformer output preamplifier that has number 27 triodes. One reviewer prefers the tone of the 26 directly heated triode with its thoriated tungsten filament, but this tube suffers from microphonics and it cannot keep the sound of the heater current out of the system even if you use a center-tapped transformer to balance it (the way you can with the 245 and this without any 60 Hz hum even if you put your ear against the speakers.) Also, the 27, which is sturdier and demands less maintenance, is affordably available in blue colored glass made by Arctus for those of us who enjoy the novelty of colored vacuum tubes.

The system has two high voltage power supplies; one at 300 Volts to power the 2C22 mu-follower which directly connect to the 245s that must use the same power supply for a direct connection, and the preamplifier that has a dropping resistor and grounding capacitor to isolate it from the 2C22, 245 combination; and a 1000 Volt power supply for the 833-A. The transformers in the 300 volt power supplies put out 5 Volts for the rectifier tubes and 6.3 Volts for the 2C22s and the 245s get their own 2.5 Volt filament transformers (One overdesigned at 10 Amperes to power both 245s) and separate 2.5 Volt transformers for the 27s in the preamplifier. For the 300 Volt power supplies, the high Voltage is turned on by a switch that grounds the center-tap windings of the 600 Volt windings. For the 1000 Volt power supply I did not think it safe to have a grounding switch for the center-tap of the 1500 Volt transformers which have no low Voltage filament windings. That way, one switch puts the system in standby, turning on all the cathode heaters. Two switches are necessary for the main power: one center-tap grounding switch for the 300 Volts and another switch that connects 120 Volt house current to the 1500 Volt transformers.

There were two challenges to the 833-A power supplies, the 1000 Volts for the plates and the 10 Volt 10 Ampere cathode heaters. The 833-A cannot operate with balanced AC or anything but the most prodigiously smoothed DC on its cathode. It requires a fully rectified 10 volts with a bridge rectifier that is rated at 50 Amperes to carry the 10 Amperes without getting so hot that it gives off an unpleasant smell. Even at a rating of 50 Amps it should be bolted to an aluminum chassis and flanked by heat sinks to keep it only slightly warm to the touch. Then it needs a dropping wire-wound resistor of 0.1 Ohms – following a small first capacitor of 82,000 microfarads, 25 Volts DC – which should be rated at least 25 Watts to keep it running only warm to the touch. You have to over-design in choosing the ratings of what you will put into this project and since you are not subjected to the bean counting close cost margins of a commercial amplifier manufacturer, the cost of over-rated parts is little more than pocket change. Be conservative here. The 10 Volts have to be filtered and the only way to get all of the hum out of the 833-As is to filter the 10 Volts with an automotive digital capacitor of 4.5 Farads (not microfarads.) as the second capacitor in the pi resistor filter. Such enormous capacitance must be pre-charged with a 9 Volt AC to DC converter through a 100 Ohm resistor while the cathode heaters are switched off because the initial inrush of current to charge them will exceed the rating of the bridge rectifier. Also a relay switch is necessary to disconnect the capacitors from the 833-A cathodes when the cathode heater power is switched off so that the capacitors will not discharge. A 20 Ampere fuse is a very good idea to put in line between the 0.1 Ohm resistor the 4.5 Farad capacitor.

The other challenge is the 1000 Volt power supply. Some people like to use mercury vapor rectifiers for the same reason they like to use vacuum tube rectifiers in power supplies that are only a few hundred volts. However, mercury vapor rectifiers have to be warmed up for as long as half an hour before you can hope they do not short circuit and perhaps explode when you hit them with high voltage. That is why I use a pair of fast-reacting diodes in parallel for higher surge current rating at 5000 Volts which are make little or no switching noise. The capacitor banks have to be made of groups of three capacitors rated at 600 Volts wired in series. The available chokes in the filters are only rated to isolate 800 Volts so they cannot be grounded. They have to be mounted on plastic and they like all high-voltage circuits and components, should have safety cages to prevent anyone from touching these high Voltage sources.

For convenience of maintenance in case something breaks down in this system everything gets its own fuse with a neon pilot light connected via a 200k resistor across the fuse terminals.

The selector switch and separate plugs for sources can be connected to LED alpha-numeric displays to show what sources are being monitored by the preamplifier and for such things as a DAC, disc, or phono amplifier, switchable outlet plugs on the power supply for those who are ambitious enough. The rest, the style of the cabinets that house amplifiers and power supplies is up to you. This is a big project, but it would cost you tens of thousands or even hundreds of thousands of dollars to buy such a system, and even then, you would not have the freedom of choice you get when you design and build it yourself.

If you want the option of running only one set of speaker wires instead of bi-amping or tri-amping, replace R4 in the 2C22 and 245 schematic with and T1 with 3600 Ohms and a Hammond 266M2 on each 245 cathode. That way you connect only one pair of wires to your speakers just leave out the 245 that would normally drive the tweeters and put it in a drawer to have in case one day you change your mind and bi-amp or tri-amp your speakers.

The site vinylsavor tube of the month; the 45, refers to the UX 245 which is the first of the 245s manufactured in 1929. This site esteems this tube as having the most beautiful tone of all vacuum tubes and it reports that the earlier globe shaped 245 is superior to the later 45. The author of this site says that he considers the directly heated triode with a thoriated tungsten filament is more revealing of detail than one with an oxide coated filament, but the 45 or 245 has a more beautiful tone and is so close in the details it reveals that it is almost indistinguishable from a 26, for instance, which has a thoriated tungsten filament. But the 245, unlike any triodes with thoriated tungsten filaments, sounds better when it is heated with alternating current with the center tap of the cathode heater transformer connected to the cathode biasing resistor and grounding capacitor. Such an arrangement allows a 120 Hz hum in the speakers that can be heard with one’s ear within one foot or less of the speakers. Thus, you may be tempted to eliminate this hum by heating the cathode with filtered direct current. However, if you use alternating current you cannot hear the hum at the seats in the listening position. Therefore, you should use alternating current for the heater on a 245 because it sounds better and not add a rectifier bridge and filter network here. With thoriated tungsten filaments you have to use filtered direct current because they are much more sensitive to alternating current or any Fourier series of alternating current that is not prodigiously filtered from that direct current. Thus, for the 833-A you have no choice but to use the well filtered DC to heat the cathode. The “baby 833-A” mu-follower 2C22, also known as the 7193 drivers of the 245s are indirectly heated by a 6.3 volt transformer winding and do not benefit from direct current heater current.

If you make your own preamplifier you might consider a 26 which requires direct current heater current and is sensitive to microphonic reaction to ambient sounds in the room, but has that detailed sound esteemed by vinylsavor, but many people find the 2C22 “baby 833-A” the king of preamplifier or driver triodes. The 27 is an indirectly heated triode that is less sensitive to microphonic reaction and it can be heated with AC from a Hammond 266M2 transformer. It is highly praised by vinylsavor and it is available in blue colored glass, which adds visual beauty to the preamplifier. It works very well with a 10K output transformer such as the Hammond 126C.

Might you get ridiculed by anyone who visits your home and sees a pair of quart-sized 833-A vacuum tubes that light up when you play some music? Perhaps, but that is unlikely when they hear the breathtaking realism it produces. The worst that will happen is that you will famous for building such an outrageous system.