4x 6550 200W RMS symmetric parallel amplifier.

This is an attempt of translation from Spanish to English. It corresponds to the article of October 2nd of 2010.


The original amplifier was designed in 1962 by R. A. Greiner, Department of Electrical Engineering at the University of Wisconsin.

He was commissioned an amplifier can deliver 120 W RMS continuous, with a range of frequencies from 20 Hz to 20,000 Hz, and the possibility of delivering 200 W for five minutes with a maximum of 2% harmonic distortion and better than the 0.5 % at 100 W.

The original design revolved around a power transformer UTC S-46, 2,000 V - 0 V - 2,000 V at 300 mA. This transformer is enough to build a 100 W RMS amplifier with 4 valves 6550 fed 800V on plate and screen-regulated 300 V (supplied from another transformer Stancor PM-8412, in original). Delivering 300 mA are sufficient for 100 W and can sustain 500 mA for ten minutes, so that the same circuit is capable of delivering 200 W RMS for the same time. After those ten minutes UTC S-46 is too hot and you have to wait to reach 200 W.

The amplifier circuit design is straightforward and simple. There has only four capacitors in the path of the audio signal. The main complication is the source of power, as it is necessary to heat the filaments of the 6550 for about 30 seconds before applying voltages.

In 1962 and in the United States of America was easy to acquire time-delay relays and other key pieces of the source is now impossible to find, particularly in this corner of the world (Argentina).

As a reference, only, is a list of them, for those who want to investigate the original design. There is a very comprehensive article in Spanish in the magazine "Radio Chassis TV", Buenos Aires, July 1962, page 55 onwards.

At present, it's more practical to think of doing the same advantage of solid state electronics, before you go looking for museum pieces and astronomical prices, if achieved.



Before proceeding we should clarify that this equipment isn't for beginners. The source of 800 V at 0.5 A is reason enough to clarify this. This type of voltages should be handled only by those with expertise and experience, the consequences can be fatal. At its peak of power 1,200 V of music signal are present in the primary of output transformer , while the driver gets to 140 V peak to peak. Even in the cables that feed the plates outlet valves must take precautions, should be appropriate cables, an usual wire is isolated for 750 V to 1,000 V. The equipment can provide risk disconnected if there is charge remaining in the filters.

It is also an expensive device, especially if it's tried to obtain the maximum quality of reproduction of the good circuit. In Argentina, a person who wound transformers for "Holimar" is still working on their own. Is able to provide a transformer with frequency response from 7 Hz to 400 kHz, thus allowing a smooth playback of 8 KHz square wave.

Winding a transformer that quality is an art and takes a long time, tens of hours. But no amplifier is better than your output transformer. First that you must do, it's provide the output transformer and components of the source that is able to feed the monster. It is useless to get the best 6550 if not done so first. The final sound will also depend on passive components and the assembly and adjustment.

To begin the project, it takes the following parts:



I tried to keep as much as possible the original layout of the source, changing only the delay relays to more accessible solid state electronics. Follows the scheme of the source and then an explanation of the steps that must be met on the turn-on process, so that the expensive valves 6550 not suffer, or to deteriorate prematurely.




When pressing the main switch, power of the domestic network is supplied to the filament transformer of the 6550 tubes, to the filaments of the two 5R4 rectifiers of the 800 V supply and to the 6X4 tube of the - 150 V supply. Relay 1 and a timer is fed from the 6.3 V of 6550 filaments by a diode bridge or voltage doubler, as required.

After thirty seconds, the timer actuates the relay 1 coil and closes the normally open contact, enabling the PM-8412 transformer. The valve 5U4GB will begin to heat their filament, so the tensions of 6550 screens and the plates of voltage amplifiers will grow gradually. Meanwhile, the 6X4 had his heater on, so that on receipt of voltage from the secondary of the PM-8412 transformer the - 150 V supply deliver that power in the time it takes to charge the capacitors.

This negative voltage of 150 V triggers the second relay, which closes two normally open contacts which allow entry into service the power supply 800 V and + 300 V 2, for the screens of the 6550 tubes. The stand-by switch makes it possible to disconnect the plates and screens of the 6550 without shutting the rest of the equipment when it isn't in use for a short time.

Although the power source was designed for electrolytic capacitors, it is preferable to use oil-paper capacitors, the sound is much better. But these capacitors are very expensive and not so readily available. An interesting alternative would be to use MKP capacitors for power factor compensation. A capacitor of this type is provided with spade terminals, durable plastic cylindrical box and nominal tolerance of 5%. With a specified voltage of 400 V 50 Hz, the DC insulation is about 1,200 V. Although the audio frequency tension that can receive these filters is, at the most, from 1% to 5% of the signal of audio present in the transformer, are many unexpected factors like variations of impedance in the electro-acoustic transducer or oscillations that can be presented. A single capacitor would be loaded with 800 V of DC. We have left about 400 V but, by the doubts, it agrees to leave the series as they are, to sleep calm. A less demanded capacitor will always work better. If it's necessary to widen the answer of audio in this filter, a 0.47 uF MKP audio capacitor 1,000 V DC can be placed in parallel with each one of the four of 100 uF. For the source of 800 V will be 68 uF and 100 uF capacitors; whereas, for the source of - 150 V will be able to be used capacitors of 22 uF, 47 uF and 39 uF, for which it's next to regulator OA2. It agrees not to increase there not to cause oscillations of relaxation. You can find capacitors of 20 uF, that is a standardized value for tolerance of 5%. In the commerce of the electrical branch these capacitors couldn't be obtained in all values, since they are specified in KVAr. The electronics commerce does not sell these capacitors. The best place to find all the series values is a repair parts storage for electric home appliances. The best marks are (In Latinamerica): Leyden, Elecond and Toyocon. A capacitor for 250 V 50 Hz is isolated to 750 V DC.

The voltage regulators valves deserve a few lines. Generally, the gaseous voltage regulators valves requires of some excitation to trigger the current that will produce the voltage regulation. Some valves have specified for a determined level of luminosity of the surroundings (for example: 50 - 500 lux) so that the shooting happens to a certain voltage. In the dark that voltage of turn-on usually is higher. The military valves used to have a small source of radioactivity to assure the turn-on with lower voltages, still in the complete dark. Generally, these valves came identified with the classic symbol of the radioactivity and the danger depends on the colors in which they are printed. If some of these valves arrives at your hands is risky its handling and it's required of professional aid. Although the levels are relatively low, the genetic effect is cumulative and directly proportional to the exposure time. The breakage of the glass capsule can release elements not only radioactive, but chemically toxic.

Another thing that is important to say is that with time they can lose the qualities that are described in the manuals, although is a new valve, as a NOS. Valves VR150, OD3 or VT-139 that use the two 300 V sources have a nominal voltage regulation of 150 V, shoot in 160 V and work in a rank of currents from 5 to 40 mA. The regulation is within 4 V (2.67%). With an integrated MC1466L a regulated source can be constructed to 0.01% and the maximum tension and current depends on the safe operational area of a Darlington pair. This integrating requires of an independent supplying of 25 V. More ahead we will see a circuit of application for + 300 V 1, that is almost the same that + 300 V 2, but, in this last case, with a more robust Darlington pair and a different limitor resistance.

With a transformer of 2,000 V + 2,000 v supplying 300 mA the specifications of the amplifier is:



If a transformer of 500 mA is used, the 200 W will be available indefinitely, with a maximum of THD of 2%.

We continued, now, with the power output stage:



It is necessary to respect the most perfect symmetry. The resistors of 100 ohms 5 W are used in the adjustment of the zero signal currents. If the currents that circulate around both branches of the primary of the output transformer are not identical, remains a residual current of magnetization that ruins partly all the efforts to obtain a good transformer. Equal voltage drops in each pair of opposite valves will be measured; that is, it is not necessary that the first pair has the same current at zero signal that the second pair, but that the currents of each pair is equal.

Now is the turn of the voltage amplifier stages, including driver of the power valves:



Once fit the currents of zero signal of both branches, the unique step that we have left is the balance with signal, that it is obtained driving the potentiometer in the entrance stage until obtaining even cut of a sinusoidal wave, observed in an oscilloscope connected to a resistance of 8.2 ohms 200 W in the exit of the transformer (or eight resistances of 68 ohms 25 W in parallel). An adjustable generator is used of such form that with the potentiometer of the amplifier becomes the cut even "as it's being seen" in the screen of the instrument, soon to lower the excitation and to observe that both tips of the wave become sinusoidal simultaneously.

Another way would be the one to fit with a distortion measurer, but it isn't a current instrument in the majority of the amateurs' labs.

Next follows the scheme of a egulador of 300 V 80 mA with a typical regulation of 0.01% and 0.03% + 3 mV in the worse case. All the masses of this circuit return to the starting point to the load. The scheme of the source was modified from a regulated source of variable tension between 0 and 300 V 80 mA. The pair of marked resistores as R2 were a potentiometer of 300K. This way, with R2 of low value, we say 3,000 ohms, the exit would be of 3 V. As the Darlington pair receives fixed 310 V, although the tension of rupture of the MJ413 is 400 V, its safe operational area with 310 V in collector limits the current of the source to 80 mA. For a regulated source from 0 to 250 V this current can raise 100 mA and R2 is 200K. RS, the limitor resistance of the circuit of protection, happens to be of 2.5 ohms and Co of 10 uF, whereas in the collector are presents 260 V.

This circuit is sufficient for the source + 300 V 1. In the case of the screens of the four power valves, I don't have data of the currents for 800 V in plates. The manual arrives up to 600 V and 800 V is the absolute maximum that can be reached. The engineer designer removed all the possible juice to 6550s. By this source would have to flow a maximum current equal to 5 mA - that it is the minimum current in the VR150- plus two maximum currents of screens. As the VR150 does not lead more than 40 mA and the original design uses a series, we would have to hope that it consists the maximum variation between the minimum consumption and the maximum, but, which is the total current that flows through the limitor resistance of the original circuit?

If this current were more than 80 mA, would force to change to the pair Darlington by another one with ampler safe operational area. There is something very important: the maximum current of the source divided by the product of minimum betas of each transistor of the pair must be inferior to .5 mA, so that the regulation isn't harmed. The reader will say that with 300 V fixed when coming out between collector and emitter has applied not more than 10 volts. This is certain when Co and the reducing noises capacitor of 2 uF is loaded. I cannot assure that in the small time in which these capacitors are to low voltage -when the charging is happening- it doesn't produces the secondary breaking in the Darlington, if the current is superior to 80 mA. These is something that will be to solve empirically or collecting precise data of the currents of screens in the conditions of operation.



It is a very expensive circuit and I don't believe that there are many in conditions to assemble it. Anyway, and although I am a fanatic of the sound of triodes, this 200 W RMS circuit could be between the best ones in more than a competition. In 1962 it represented the top that could be obtained without resorting to great valves of transmission of several kilowatts and unmanageable high voltages in a modest amateur's working table.

IMPORTANT NOTE: If there is some possibility that exists an oscillation by coupling feedback, it's fundamental to place spark gaps adapted in the plates and the screens of the power valves.

Cabezal Marshall 100 W RMS para bajo.

Hace por lo menos unos veinte años atrás, Hugo Racca, un músico de Heavy Metal instrumental, me llamó para copiar un cabezal Marshall de su propiedad.

Él quería tener un amplificador que sonara igual, pero que fuera menos costoso y que pudiera llevarlo a los ensayos en lugar del equipo importado. Así minimizaba el riesgo de que le robaran el amplificador inglés, muy caro para el poder adquisitivo de un argentino medio.

Como estábamos en su casa, lo único que tenía a mano era papel cuadriculado, una lapicera y un tester.

Procedí a dibujar lo que veía, reordenar y volver a dibujar en limpio. Por último, tomé algunas medidas de tensiones y corrientes. Hugo y sus hijos se encargaron de hacer un chasis con las mismas perforaciones y medidas y yo procedí al montaje del circuito, unos días después. El original venía en una base de pertinax con pilares en doble fila. Todos los componentes se disponían en paralelo y diferentes cables hacían el resto de las conexiones. Algunos resistores estaban sobre los zócalos de las válvulas.

Cuando llegó el momento de la prueba, Hugo se mostró muy conforme con el sonido. Según dijo era igual al del equipo clonado. Usamos todos los componentes que debíamos, pero el transformador de salida era de 6.600 ohms placa a placa y 35 W RMS, en lugar de los 100 W requeridos. Esto se hizo provisoriamente hasta que Hugo encargara un transformador de salida a Alamtec, que en aquel entonces bobinaba salidas de audio con siete particiones del primario. Ni el transformador que usé, ni el que fabricaba Alamtec eran copia del original de Marshall, que ignoro si era de la misma impedancia. Sucede que el manual de Mullard indica 4.000 ó 5.000 para polarizaciones muy próximas a las de este circuito. Sin embargo, el músico dijo que daba el mismo sonido heavy inglés que él quería.

Para no achicharrar al transformador de salida de 35 W, el potenciómetro de realce, que varía la realimentación negativa en el extremo alto de la banda pasante, fue llevado de 500 ohms a 5.000, sin cambiar el capacitor. Daba así mucha menos potencia y, sin embargo, mantuvo el sonido característico.

Las tres imágenes siguientes muestran el preamplificador, la etapa de potencia y la fuente de alimentación. Queda una duda con respecto al inductor de filtro. Hugo dijo que era de 1,3 H y trajo uno, junto con el transformador de poder. Yo lo conecté y funcionó bien, sin zumbidos ni ruidos.

Además de la prueba con el bajo de Racca, un amigo, que era primer guitarrista del conjunto, ensayó con el mismo amplificador y le agradó su sonido. Aunque es un cabezal para bajos, tiene suficiente respuesta de agudos como para reproducir bien una guitarra.

English translation (Sorry for my English):

For at least twenty years ago, Hugo Racca, a Heavy Metal instrumental musician, called me to copy a Marshall head of his property.

He wanted an amp that sounded like, but make it less expensive and could perform the tests instead of imported equipment. Thereby minimizing the risk that the English amp was stolen, too expensive for the average purchasing power of Argentines.

Since we were at his home, all I had at hand was graph paper, a pen and a tester.

I proceeded to draw what I saw, rearrange and redraw in clean. Finally, I took some measurements of voltages and currents. Hugo and his sons were ordered for making a chassis with the same drilling and measures , and I proceeded to assemble the circuit , a few days later . The original was built in a base of pertinax with pillars in two rows. All components were located in parallel and different wires did the rest of the connections. Some resistors were on the valve sockets.

When it came time to test, Hugo was very happy with the sound. Said was the same as the cloned amp. We use all the components that the circuit needed, but the output transformer was of 6,600 ohms plate to plate and 35 W RMS, instead of the 100 W required. This was provisionally until Hugo commission to Alamtec an output transformer, which was then wound audio outputs with seven primary partitions. The transformer I used wasn't the copy of the original Marshall's output, neither it was made by Alamtec. I don't know if it was the same impedance. It happens that the Mullard's manual indicates 4,000 or 5,000 for polarizations close to those of this circuit. However, the musician said it was the same English heavy sound he wanted.

For not to burn the transformer output of 35 W, potentiometer enhancement, which varies the negative feedback in the high end of the bandwidth, was taken from 500 ohms to 5,000, without changing the capacitor. Was thus much less power, however, retained the characteristic sound.

The next three images show the preamplifier, power amplifier and power supply. There remains a doubt as to the filter inductor. Hugo said that was 1.3 H and brought one along with the power transformer. I plugged it and it worked fine, no hum or noise.

In addition to the test with the bass of Racca, a friend who was lead guitarist of the group, tested with the same amp and liked the sound. Although a head for bass, treble response is sufficient to reproduce well a guitar.






*: Chosen equal to the best possible accuracy.



Activate the switch 2 30 seconds after the switch 1. When turning off, first turn off the switch 2.

Otro amplificador con 6BX7.

Los capacitores de filtrado y desacoplamiento están dibujados de manera genérica, pues es posible colocar electrolíticos de buena calidad, capacitores de papel al aceite, de polipropileno impregnado o de polipropileno.

La válvula rectificadora es una GZ32 o alguna similar que sirva de reemplazo. Entregará algo más de 100 mA, por lo que sirve para alimentar un solo canal. De cualquier forma es mejor que tengan fuentes independientes para aumentar la diafonía.

Un amplificador simétrico con 6BX7.

Este amplificador simétrico en paralelo con dos válvulas 6BX7 como salidas no es un circuito muy elaborado, pero siempre se cuenta con la bondad de los tríodos. Su potencia de salida estará próxima a los 7 W RMS o un poco más; no más de 10 W.

Como siempre se dice y se repite hasta el hartazgo, la calidad final dependerá fundamentalmente de su transformador de salida. Después entran en consideración los componentes pasivos utilizados, el armado y la fuente de alimentación, que también es muy importante.

Los resistores de buena calidad, como los Sofcor, Allen Bradley, Caddock o Rodenstein, y los capacitores, Wima, ITT, Elna, etc. juegan un papel nada despreciable, si primero se consiguen buenos transformadores.

Hace un rato largo que estoy pensando en dar precisiones acerca de los transformadores de salida. Este caso en especial sería ideal, porque el circuito se alimenta con 300 V y no puede haber más que 260 V de audio sobre el primario. Trabajar con bajas tensiones facilita el bobinado, porque no hay demasiadas posibilidades de que salten arcos ni exigencias extremas de aislación. Por sobre los 450 V la cosa cambia y mucho. Pero hay otro factor que influye, además de lo experto que hay que ser para trabajar en bobinados con tensiones altas: es la configuración partida de las bobinas la que viene a complicar también el trabajo. En un bobinado lineal común se calcula un 20% más de espacio de ventana, para compensar las cosas que suceden en la práctica. Esto con bobinadoras eléctricas y con tensores. Si el bobinado es a mano, es casi imposible mantener una tensión pareja en todo el carrete. El factor de seguridad para la ventana pasa al 50% si se bobina sin tensores. Esto y la inexperiencia pueden causar la pérdida de algunos billetes, aunque siempre se gana algo en aprendizaje.

De cualquier forma, hay gente que sabe -y mucho- y puede llegar a ser conveniente comprar un buen transformador a un costo alto, antes que perder una cantidad de dinero en materiales que no son reutilizables.

Un amplificador para auriculares.

Este amplificador lo vi en un sitio que no puedo recordar y que me fue imposible ubicar ahora. No puedo decidirme si estaba situado en Suecia, Dinamarca o en Holanda. El hecho es que está en una página que ofrece el kit completo para armarlo o ya armado, si lo prefiere.

En su momento el circuito estaba dibujado a mano en un papel cuadriculado. Estuve mucho tiempo buscando un CAD que dibujara válvulas. No logré encontrar uno. Por eso postergué por bastante tiempo la edición de este blog. Casi desanimado por completo, un día se me ocurrió guardar algunos sectores de circuitos que ya estaban dibujados de alguna manera. Por ejemplo, con el ACD See 8 recorté un pentodo y lo corregí con el editor y la función clonación. También pude borrar el interior del pentodo y redibujar un tríodo. Insertando esas imágenes en Microsoft Office PowerPoint 2007, fue posible rotarlas y ajustarlas a distintos tamaños para adecuarlas a las diferentes necesidades. Surgió, así un pequeño banco de imágenes, que uso en este blog. Los dibujos no son perfectos, especialmente en lo que respecta a simetría; no siempre las imágenes invertidas coinciden completamente cuando hay que interconectarlas. El ajuste procede por saltos y las imágenes no son equidistantes a los bordes de los recortes. De eso me di cuenta cuando empecé a hacer coincidir imágenes en conjuntos más o menos complejos.

Para armar un circuito abro el Office PowerPoint 2007, pongo diapositiva en blanco y comienzo a insertar imágenes. Las corro y hasta redimensiono según la necesidad. Cuando tengo todas las piezas posibles. Grabo el archivo .ppt y lo transformo a .pps. Luego lo abro y capturo la pantalla con la tecla "Impr Pant PetSis". Abro el ACD See 8 y pego la imagen. Después la abro con el mismo programa y lo pongo en edición. Por clonación voy dibujando las líneas que unen los componentes, los capacitores, los saltos entre líneas que se cruzan, los terminales y otras imágenes que no tengo grabadas en el banco. Dibujo punto por punto. Un trabajo de preso, pero que hago cada vez más rápido y con mayor facilidad. Manteniendo el botón izquierdo del mouse apretado va clonando el punto anterior y, con un poco de pulso, uno puede arrastrar el mouse y dibujar una línea recta bastante rápido.

Así fue como redibujé este circuito y se lo envié al sitio para que lo publicaran, si querían, en lugar del dibujado a mano. La dirección del correo es: AquaBlue Diyparadiso Hi End "sales(arroba)diyparadiso.com". Ignoro por qué no puedo encontrar el circuito ahora.

Publico este circuito porque me parece que es una solución barata para escuchar música con cierta calidad de sonido. No hay que gastar en transformadores de salida, ni en gabinetes acústicos. Además, las válvulas son baratas; especialmente si se cambia la 6SN7 por una 6CG7 o una 6FQ7. La válvula de salida vertical 6BX7 no es cara y con ella se pueden construir amplificadores push-pull de 7 W RMS, aunque ya se necesitan transformadores de salida, que si son buenos, cuestan. Hay auriculares de muy buena calidad y cuestan mucho menos que un gabinete acústico bien construido y con altavoces de calidad. Cuide sus oídos, no exagere con el volumen.