return to "index" page

published on Issue 1 of ‘i Quaderni di CHF’ (March 2007)



protection, cooling, damping vibrations


The magnificent electron tubes 211 and 845 are above average for construction, linearity, power and a special way of playing, due to the fact that they work with high voltage.

The multifunction structure described here is designed for the best usage of these excellent vacuum tubes, thanks to technical expedients that distill the musical quintessence and guarantee its protection at the same time.
The latter is obtained thanks to the design, whose expression of fine simplicity is fulfilled in the use of glass, which dematerializes the structure itself, sublimating the fascinating brilliance of the direct heated triode.



The adoption of a protecting structure in the use of these tubes has two purposes: to preserve the device and personal safety.

The will to protect the device is natural and understandable if we consider that these tubes, for example some 845 new old stock, are quoted at over $1,000 (eBay reference). Nevertheless there are also other reasons that justify it, such as the protection of particular good sounding paired and aged specimens.
The tubes 211 and 845 are built with fine glass to help the release of heat to the outside, and this makes them relatively fragile. Furthermore, when the filament works with bright-yellow color it is more vulnerable; strong vibrations caused by accidental impacts can damage it.

The cylinder used as protection is also made of glass, but it is thick Pyrex suitable for high temperatures, which guarantees high strength and offers optimal protection to the tube, also thanks to the three supporting metal stems.
The protection includes an optional grid (not shown) placed in the top part of the cylinder.



The protection provided by the glass cylinder has a dual purpose: it is not only *for* the tube, but also *from* the tube.

We must remember that these triodes work at very high temperatures and if we inadvertently touch them we can easily burn ourselves.
Furthermore, if there is an accidental impact that causes the breaking of the tube’s glass, there is the risk of cutting oneself as well as the danger of the high voltage.

All in all the presence of a solid Pyrex cylinder is quite reassuring.

Besides, we must consider the negative effects of high temperature on the components surrounding the tube, both in aesthetical terms, for premature yellowing, and qualitative terms, for the decay of several parameters, such as the dielectric strength of capacitors (especially if they are electrolytic) and transformers. On the outside the Pyrex cylinder keeps a moderate temperature, suitable for the layout of the amplifier, allowing the close arrangement of the components without suffering for the temperature.



Apart from being a protection, the glass cylinder has another very important function. By using the principle where hot air naturally goes up, it implements the ordinary Chimney Effect, where there is a lower pressure that draws in cooler air from the bottom and starts a cycle that helps the disposal of the heat produced by the tube, helping the cooling process.



If we observe, in the images, the bottom part of the glass cylinder, we can see the wide circular opening around the socket of the tube; towards the top this section becomes smaller as a consequence of the wider diameter of the tube.

In practice when the fresh air from the bottom goes up, it meets a narrowing. Because of the Venturi effect, the air is accelerated in the point where it is more useful, optimizing the efficiency of the cooling system.

The more the section around the tube is reduced, the more the speed of the air passing into it increases; nevertheless we must consider that the tube needs a sufficient margin to irradiate heat, otherwise a reflection can occur, with effects opposite to the desired ones.

In conclusion, there is an optimal internal diameter for the glass cylinder able to guarantee the best performance in all conditions.


I still keep the piece of paper, now gone yellow, with the first hand-drawn sketch made around fifteen years ago. Initially the structure was conceived to integrate to the side of the chassis, the supporting stems of the glass cylinder are shown in a 5mm diameter, but later 4mm stems were used. All the remaining dimensions are the same. In particular the inside diameter of the glass cylinder is 77.5mm; this dimension is a result of experimentation and it is very important for its efficency.

With this natural cooling method the tubes 211/845 old stock can dissipate between 120W and 130W depending on the Ambient Temperature, causing only a light reddening of the central area of the plate. Such conditions can be kept for a long time without jeopardizing the performance of the tube, but of course its lifespan will be reduced.

Even if we limit the dissipation of the valve to 100W, the guaranteed cooling of the glass chimney is advantageous, because it guarantees a lower emission of internal gases, from the plate, the metal and glass parts. The consequence is a benefit in the musical quality, in relation to the less internal reflections and the reduced grid current which, in the best specimens, will correspond to a few µA.

Basically the cooling allows better performance which can be capitalized according to different criteria. Generally we aim to squeeze out more power, also thanks to the possibility of increasing the bias current, but it is also possible to achieve a more calibrated orientation, aimed at minimizing the distortion rate.

In order to better evaluate the various possibilities, it is useful to make some considerations on a graphical basis, using the tube 211 as an example.



In order to simplify the calculation process I’ve used software, where I’ve initially provided a mapping with points of the anodic curves of the 211, obtained from the diagram in the picture.

In the data-sheet of the tube 211, in correspondence to the ‘classic’ point of work at 1250V with 60mA, the power reported is 19.7W with 5% second harmonic.
The software completely reproduces the situation represented in the data-sheet, calculating the same power and distortion values; it is therefore suitable to supply valid indications.

The supplementary cooling provided by the glass chimney and by the Venturi effect, allows the tube to operate with dissipation in the order of 110W in conditions of extreme safety for the tube; the plate doesn’t become red and the grid current doesn’t increase significantly.
Now we will verify the values given back by the software by increasing the bias current of a quantity corresponding to the new parameter of dissipation. 



In these new conditions the available power in class A is 18.2W, which is lower than the ‘classic’ point of work proposed by the data-sheet; in return the distortion of the second harmonic is three times lower: 1.64%!

This aspect is already definitely interesting, but that’s not all. We must highlight how the 19.7W used as a reference are obtained in extreme conditions of end test (the 5% of second harmonic shows it clearly), whilst in the condition we are considering it is still possible to extend the swing in A2.



The second diagram completes the picture of the situation: in A2 we obtain 35W with 3% of THD!
Resuming… not only we will have a much better quality in the 18W range in class A, but we will also have the possibility to manage peaks of signal with double power and reduced distortion, all to the advantage of a more energetic and dynamic musical reproduction.

Now let’s examine a hypothesis of particular operating point.
By personal experience I know that the 211 old stock can safely and permanently work at 1350V; this characteristic has found confirmation in the testimony of many other enthusiasts and this has been consolidated over many years.
Let’s put the bias point, then, to 1350V without any fear, reducing the current to 81mA in order to maintain the dissipation of the tube at 110W.
Furthermore let’s use a different ratio of transformation, connecting the load not to the 8ohm terminal of the secondary any more, but to the 4ohm terminal. This maneuver turns out to be interesting only with very good quality transformers, with many henry at the primary and limited loss, such as the transformer ISO-Tango X-10SF which, with its 80H, responds to the needs very well. In this instance we don’t consider the theoretical impedance value reflected in the primary, but a lower value, 16Kohm (by chance the same impedance value of the output transformer in the Ongaku of the great Kondo-San). Notice that such a value equates well in proportion with the 50H corresponding to 10Kohm of impedance of Tamura F-2013.



The diagram shows the new point of work at 1350V, 81mA, with 16Kohm of charge. The power obtainable in class A1 is again 18.2W, but the distortion is further lowered consolidating at 0.9%!

The new situation has the following characteristics: 1) definitely reduced distortion, 2) lower continuous current in the output transformer, 3) higher damping towards the charge.
In other words… very good 18W!



The quality of the 18W is corroborated by the possibility of extending the functioning of the amplifier in class A2. As shown in the diagram we can easily reach 40Wrms at 2% of THD!
This graphic representation finds confirmation in reality, with equivalent instrumental measures obtained at 1 KHz. We find more distortion at very low frequency, because of the limits reached in the power of the exit transformer.
Notice how in this case the positive swing doesn’t lead the final tube to delivery high peaks of current, maintaining less than 160mA, as security for a functioning condition of the tube safer than what the abundant dissipation makes us assume.
In particular, thanks to this, the performance supplied by the tube will be more homogeneous during the whole lifespan. 

To drive the 211 in positive grid is an arduous task for many signal tubes, to the point that, in medium frequencies, there is a very high probability that the driver tube is the weak link in the circuit



The image of the oscilloscope’s screen shows a circumstance of this kind. The widest sinusoid, 1 KHz, corresponds to full power at the output of the amplifier, whilst the smaller sinusoid shows the current supplied by an EC8020 in the grid circuit of the 211, obtained by a probe Tektronix A6302 + AM503. The reached limit in the ability to supply current to the grid of the final can be clearly seen when this becomes extremely positive.

The points of work examined and the relative considerations represent simple starting points for reflection on alternative approaches.
But independently of the bias point preferences, it is a fact that cooling and good ventilation optimizes the performance of the tube, both for signal and power, even when it is underused.
On the contrary, to smother a tube with frills can worsen certain parameters; in some cases it can be fatal. Indeed, it has happened that such practices have caused damage.
When using accessories we must carefully weigh pros and cons; a simple rubber ring on the body of a tube can disturb the convective transfer and reduce the dissipation of heat. This is an important disadvantage especially for power tubes.


Vibrations are a problem for the tubes 211 / 845, not only for the typical problem of a microphonic susceptible grid, but also in relation to another more deceitful aspect concerning filaments.

To supply the filaments of the direct heated triodes I often use DC voltage, in order to avoid possible hums. Unfortunately this method has some drawbacks; in particular, it inhibits the highest musical quality as a consequence of asymmetry in the electrodes potential.
In order to have the highest quality the only solution is to supply the filaments in AC voltage and the best way to avoid hums is to use a frequency out of the audio range.

On this subject I would like to point out the solution proposed by Peter Millet; his circuit to supply the filaments at 100 KHz is definitely brilliant ( I used a different method for the tests that I did a while ago, using an ordinary 100+100Wrms audio amplifier, piloted by an audio oscillator, to obtain sinusoidal tensions at 40 KHz in output. In actual fact I also tried higher frequencies, up to 60 KHz, but the geometrical perfection of the sinusoid obtained, in such a way allows an optimal reduction by cancellation, averting the possibility of intermodulations; in other words with this pure analogical system there is no reason for preferring extremely high frequencies.



The various tests also included verifications with excursion in frequency from 50 Hz to 60 KHz; I tried different tubes, in particular VT-52, and when I used 211 / 845 tubes… worrying poltergeists started showing. The instruments revealed strange frequencies, which appeared suddenly in the most random fashion, overlapping the test frequency.
In actual fact there was a random link and it was me. By handling the measuring set I occasionally caused the tube to vibrate.
Basically, the long and fine filaments of these tubes are susceptible when they are excited by energy sufficient to cause their motion, manifesting prolonged resonances. The difference from the grid’s microphonic susceptibility is in the longer duration, as a consequence of the longer filament span.
It is impossible to forecast the level and frequency of these resonances, because they depend on the entity of the exciting force and on the small mechanical differences between the different specimens of tube, in particular on the tension of the springs supporting the filament.
The methods used to annul the effects of Alternating Current reduce its effect on the signal, but not enough, because they are not optimally tuned in.

In other words the best option is to knock down the transmission of energy to the tube, annulling the vibrations given to the grid and the filament, because they translate into light background noises and degrade micro-details; in this way we obtain an unmasking in the musical reproduction.



The energy to be knocked down reaches the tube by direct irradiation of sonorous waves, or through the supporting mechanical structure.

The glass cylinder annihilates the sonorous energy by direct irradiation, typically coming from the speakers, surrounding and protecting the tube in the part where it is more sensitive to this stress. The tube remains exposed on the top, but the dome part is less sensitive; furthermore at this point there is the flow of hot and rarefied air ascending, therefore the transmission of energy is quite reduced.

In order to dispose of the energy brought to the tube through the support mechanics, we have used a technique of fractioning and reduction.
The whole structure refers to a main supporting disc (on which the glass cylinder rests), from it, through the return of three columns at 120°, we obtain a counter-plane that, through three rubber M3 vibrostops, supports the small disc for the decoupled fastening of the ceramic socket E.F.Johnson model 211.



The energy dissipation operated by the structure is completed in the assembling to the top chassis of the amplifier. The structure is in actual fact separate, therefore it can be elastically hung through three M3 or M4 vibrostops at 120° fixed to the main disc.



Alternatively, it can be placed on a soft polyurethane gasket that guarantees airtight condition along the whole circumference of the main disc. In this instance we want to use a fan (undersupplied and decoupled in the fastening) to allow the airflow inside the device, keeping the only outlet the chimney to increase at will the flow rate along the chimney itself.



I have tested a version with the three columns interrupted by M4 vibrostop, the sensitivity to mechanically induced vibrations is further reduced and this solution can be used to increase the elastic decoupling of the chassis.

On the whole the solutions adopted are sufficient to still the conditions of work of the tube.               
The vibrations typically found in a domestic environment, are not of such energy to carry out the whole path fixed by the structure, and reach the tube with sufficient potential to excite it in a significant manner in relation to the function performed.

We must highlight that the annulment of vibrations is obtained inside a situation that protects the tube and improves its electrical performance at the same time.



The lower discs have wide holes in the middle, for three reasons: 1) reduce the resonances of the discs, 2) allow aeration to the pin and the base of the tube (the ceramic socket has a wide hole at the bottom as well), 3) allow the optional setting of a mass-damper for oscillations.

With regards to the point 3), we must specify that the ceramic gasket is fixed elastically only on two points (as it is clearly shown by the images), through three soft rubber washer plus two plastic jacks, that catch the center of the hole and prevent the bolt from touching the ceramic of the socket.
With this arrangement the oscillation of the tube draws an arc and this situation allows the implementation of a mass-damper with a flexible plate staff as a spring.



In the picture the mass-damper is represented in red, the flexible plate staff is seen in profile. The length and flexibility of the plate, in relation to the weight used for the mass, are to be set with reference to the elastic reaction of the group socket-tube. I have successfully tried fiberglass plates, where the flexibility is very easy to set, simply by changing the width. This material is preferable also for the crossbar under the socket, because it increases the flexibility and it is a guarantee in terms of electrical isolation.


In conclusion…exceptional tubes like 211 and 845 are able to play wonderfully with a conventional installation, as many great amplifiers demonstrate, nevertheless, in order to improve their performance and musicality, within a situation that ensures protection and safety to the level of CE marking, the concepts and solutions above described are very convenient.