An application of the AV2A3
By Marco Ferretti About the AV2A3
The AV2A3 is built by AVVT as a direct replacement of the classical single plate 2A3. I wouldn't say it is. It think it's more than that. Let’s say it’s fully compatible, but it can give us much more, electrically and from a listening point of view. It's plate dissipation and maximum voltage are higher than in the original tube, and as a consequence we get more power fom the AV2A3 which improves the versatility of this medium power tube.
The plate resistance and the amplification factor are very close to the parameters of the common 2A3, to ensure the full compatibility, but it's design and construction technique are completely different. The AV2A3 is much larger than the obsolete 2A3. The tube is about 15cm high and the largest diameter of the glass bulb is 7cm. You may notice inside the glass the large plate (yes, it’s a mono-plate tube!!), the anti-microphonic filament suspension and the typical shape of the plate. Both the grid and fiament are longer than the plate itself. This feature will allow strict linearity and decreased distortion, because in the region where the fiament is mounted to the suspension, the temperature is lower, and the electron emission is not constant over all the filament length. In the AV2A3, the plate will only collect the eletrons emitted by the filament where its temperature is constant.
Something similar happens with the grid: at the extremity of the grid winding, the electron flow is not deviated constantly, but there will be a some kind of turbulence because the grid will end bluntly. Also here, the plate will only collect electrons in the region where the grid performs constantly.
On the AVVT’s datasheet, the suggested load for the AV2A3 ranges from 2400 to 5500 ohms. To use the tube at it's maximum capability, a typical 2A3 output transformer is not suitable here, because of the higher power the AV2A3 will handle in this circuit. I have designed a dedicated output transformer for it, since it has to opererate with a higher bias current and output power. For the AV2A3, I have choosen a 3400 ohm load which was for me the best compromise between output power, distortion and listening performace.
Let’s start with the power stage. I have chosen 3400 ohm for the AV2A3 and a bias point that gives 24,8 Watt plate dissipation. The power stage operated at 270Volt, and 90 mA plate current. This will allow for 5 Watts of output power. I believe that it's useless to push the plate dissipation of a tube to the maximum value to get only some extra milli-watts. I have looked for a bias point and a load where the tube is still very linear. Driving: the grid of the power tube is connetced to a cathode follower that will both drive and bias it. I'm perfectly aware that the AV2A3 does not draw much grid current, but this does not mean that it doesn't draw gird current at all. The situation is different when the operation switches to class A2 where the grid will attract the electrons since it becomes positive, and here a large amount of current is absorbed. It is true that the AV2A3 was not designed to operate in class A2, but we can use this class of operation as well to get more punch and improve speaker driving during large transients and peaks. The AV2A3 can handle a 600mA impulse current, so it becomes very interesting to allow an occasional A2 operation, but in this case the driver must be able to supply current to the grid of the power tube. The cathode follower and it's low output impedance have another strong advantage... it allows the grid to be biased and to keep this voltage very stable. The cathode follower as a driver excists also because the 6SL7 is as a voltage amplifier. This tube has a very high gain, and a very high plate impedanc, so that would not allow direct driving of a power tube. A high gain was needed in order to avoid two voltage amplifiers in cascade, but in this case gain was even too high, so the cathode resistor was left un-bypassed and on the other hand, a local feedback will reduce the amount of distortion of this stage.
Finally, and as a consequence, we have avoided the cathode capacitor that is effectively on the signal path. The voltage amplifier can generate a 160V signal swing with a very soft clipping; the overall gain of this stage is about 20...26dB. A peak voltage of 2.7V is needed to push the power tube up to class A2. May be the sensitivity of this amplifier is not very high, but we are dealing with a power amplifier, not integrated equipment. The low gain brings some extra advantages for the the overall distortion and noise.
Here things get a bit more complicated. Let’s see what kind of power supply this circuit needs. The 6SL7 has a high plate resistance and a high gain, this makes me think that it will need a very stabile, clean and noiseless supply voltage. If not, it would amplify all the residual noise coming from the power supply. The best solution is to dedicate to this stage a stand-alone power supply for this reason. When we deal with eletronic circuitly, we are alvays very far from the ideal operation, and what makes a power supply "non-ideal" is its internal resistance and the noise generated by the rectifier. So, even if we connect all stages to the same power supply, and use plenty of RC filters to decouple the input stage from the others, at the moment the power tube will demand for current, there will be a change in supply voltage because of the internal resistance of the power supply. Well, this effect will reach the input stage, and will be amplified by the power stage that will drive your speakers. You will listen to something that has nothing to do with the music program. All decoupling networks are there to avoid this, but a music signal is not a sine wave, but it looks like a series of pulses, and these will pass through the decoupling network because it will not be able to attenuate them fully.
The input stage has its own dedicated power supply with a dedicated secondary on the power transformer. This power supply is made of a solid state regulator with no error amplifier that controls the output voltage. If feedback is used to regulate the supply voltage, it will take some time for the error amplifier to correct the voltage rating because of it's hysteresis. This will generate additonal noise components to the supply line. Now imagine what the input stage will do: it will amplify all of this too!
There is nothing to say about the power stage power supply, since I believe you have seen something similar in many other designs. The driver stage gets its supply voltage form the power stage power supply because it is effectively part of the power stage itself.
Components and construction
I am strongly convinced that a circuit is not fully optimized when it is sensitive to the quality of the passive components.
Passive components are very important to get the best listening performance, but first we need the circuit to sound good with standard parts. A high quality part is just an improvement; the place where it must be used has to be located carefuflly and there it can really make the difference.
I had very good results with Holco resistors on the signal path only, please look for a good potentiometer to be used on the bias network. The only coupling capacitor is a paper in oil unit. All power supply components are standard parts, but please look for good capacitors!
The 6SL7 is a Sovtek tube, its quality is very good and it sounds great. Do not buy any 6SN7s if you have one left in your laboratory, when used as a cathode follower it will not influence sound at all.
Do not use a standard 2A3 power transformer here. The AV2A3 is able to provide higher power and is operated with a higher bias current. The output transformer has to be able to handle a higher power and current without core saturation. I have designed a dedicated output transformer for this application. It has 13 section windings and bi-filar secondary. There are two power transformers here, one to supply the high voltage and another for the filaments. They have been both designed for very low magnetic induction in order to reduce hum effects. The line stage is very simple, a point-to-point mounting technique will do, but for the power supply circuitry you will need a PCB.
Once you have fininshed mounting, you should check that the power supplies do their job correctly without the installing the tubes, then turn both of the bias potentiometers in order to supply to the 6SN7 grids the largest negative voltage. Now switch the power off and allow the capacitors to discharge before going any further. Now connect a voltmeter across the 1 ohm resistor on the virtual cathode of the AV2A3 and set the bias until you read 0,09Volt. Doing this you have set a 90mA bias current. Connect the amplifier to your speakers and turn the two potentiometers on the power tube’s filaments until every trace of hum disapears. Leave the tubes in this state some time so that they get electrically stable and then fine-tune the bias on both channels.
It was not easy to give the listening impressions for an amplifier made by ourselves, but I will try to give a fair judgement about it. The amlpifier was first connected directly to the source, and then to two different preamplifiers, the first was a hybrid design, and the second one a very simple tube preamp. We have tested this amplfier on different speakers. One small two way system equipped with a 13cm mid-woofer, a Lowther DX2 loaded in a TWQT cabinet, and the second system was a pair of Magneplanar MG 1.6.
Even if the output power was low, the amplifier was reacting very well in every situation thanks to the power tubes and the dedicated output transformers. The amount of sound pressure was not high with the Magneplanars, but driving was excellent, and while listening, it seemed as if we had much more power than we really had. With all speakers we used for testings, the attention was on the mid-bass that was very strong and present but well damped. The acoustical imagae was very stabile and well focused. It was amazingly easy to locate each instrument on the virtual scene, thanks to the mid-range realism that we found very present, and with plenty of dynamics.
What amazed us was the stability of the deep bass, up to the highest range where sound begins to be a little softer and graceful. During all test, the amplifier left us with no words, and it was not easy at all to collect the impressions from the other persons that were present that day.
I was really amazed by the performance of the AV2A3. This is why I am here to propose to all of you this work even if not finished in the layout and engineering.
The price of this tube is very interesting if we look at the quotation of a Chinese 300B, that has less technical contents, or innovations and doesn not have the carefull construction of AVVT's AV2A3. This amplifier has the great advantage to be able to drive mid-low sensitivity speakers, in addition it's output power is suffucient for almost any domestic use. If you dwould ecide to build it, you will be surprised by it's listening performance and by it's versatility in use. The overall quality is very high, and I believe it’s even better than with a good 300B amplifier.
Even if the results I have obtained with this amp are excellent, this is still an open design, there are some points where improvements can be done. In any case, I believe that the version presented here is very complete from a technical view, but it can be seen also as a starting point for other personal interpretations.
The tubes and transformers as decribed in the article, can be purchased at:
di Bruno Ceccarelli,
Via Adda 7,
63031 Castel di Lama (AP),
tel/fax: 07188.8.131.52, tel: 03184.108.40.2069
About the author
Marco Ferretti Marco Ferretti is a student of electronic engineering at the Turin Polytechnic in Italy. He became involved with tube amplification in highschool, and then was attracted to audio reproduction. After practising with push-pull amplifiers, he focused his research on single ended designs. His dream was to turn his hobby into a profession: Sotis Audio was recently fouded with the help of two friends. He is now developing his product line that will be soon presented.