Modify the UL40-S2 into a Super-Triode amplifier Ir. Menno van der Veen Introduction about modifications: The UL40-S2 is already some years on the market and meanwhile I have received several requests for modification of the original circuit. I tested these and in most cases I was not convinced that they offered a real improvement. I got questions about upgrading with other brands of capacitors or resistors or connectors. Again I implemented those, but noticed that they did not offer a well balanced improvement of quality versus money. When I designed the UL40-S2 one of my goals was to design an amplifier that would function well with standard quality components, thus preventing extreme high costs. Up till now this original design has proven to be stable and resistant to modifications. However, one modification inside the toroidal output transformers showed a tremendous improvement in sound quality. Instead of standard copper magnet wire, I used high purity silver magnet wire with a little amount of gold doting (wire from Siltech). I did not offer such a silver output transformer as a standard component in the UL40-S2. The customer was free to decide to upgrade the amplifier with silver output transformers. I must admit that this upgrade is a costly one. However, the improvement is of such a high caliber that I spent extra research to understand the nature of this improvement (which is intended to be published as an AES paper). I also did tests with this silver wire as an internal interlink between volume potentiometer and switchboard. Again this upgrade worked very fine. Up till now this was the only officially published modification. New research leading to new opportunities: Recently the situation has changed because new research in my lab offered me an outstanding improvement, which I following will explain.
I always used the UL40-S2 in its Ultra-Linear mode, because I needed the large output power of 29 Watt for my loudspeakers. In the Triode mode the sound quality was better, but the output power went down to a mere 1 Watt, which was too little in my applications. What I really would like to have is a smart invention that would offer me 29 Watt Ultra-Linear output power with Triode mode quality. But we all know that inventions only come when their time is right. Meanwhile I was busy in my lab with a new research about a large general system, that contains all possible topologies and concepts of valve amplifiers. The system not only covers Single-Ended and Push-Pull amplifiers, but also Paralleled-Push-Pull plus balanced circuits like the Circlotron and Unity-Coupled plus many other smart vacuum tube amplifier inventions. This new approach clearly showed that the different valve amp topologies all use a certain amount of local negative feedback around the output transformer and the power tubes. Following I tested many topologies in a universal amplifier (almost equal to the UL40-S2). These tests clearly taught me that there are two winners: the pure triode circuit (SE or PP, both are excellent) and the Super-Triode circuit which I had invented and published in 1998. Comparing these two winners, it became clear that their sound quality was equal, but the output power of the Super-Triode was twice the power of the Triode circuit. Knowing this, I was close to the new invention which I will discuss. The only issue I had to solve was the implementation of the Super-Triode concept into the original UL40-S2. From Triode to Super-Triode: The major difference between Triode and Super-Triode circuitry is that the Triode only uses one control grid, while the Super-Triode uses a control grid and a screen grid connected to the Ultra- Linear taps on the primary of the output transformer. See the figures 1 and 2 for more information. 1 3 Vht inverse Fig.1 : Basic Triode circuit. Copyright 2004 Vanderveen
1 2 3 Vht 4 inverse Fig. 2 : Basic Super-Triode circuit. Copyright 2004 Vanderveen The standard output transformer (VDV6040PP or PAT4002) inside the UL40-S2 does have Ultra- Linear taps (taps 2 and 4) on the primary. However, this output transformer does not have a separate cathode feedback winding. Would this mean that the Super-Triode circuit is not possible in the UL40-S2? By experience I know that the words not possible ask for closer attention, and therefore I focused on finding an alternative solution. Fortunately my new system showed me the way how to solve this problem, and a new invention was the result. I now will discuss this new modification that will change the UL40-S2 into a Super-Triode amplifier. The price of this modification is little, only a few Euros, while the net result gives a tremendous improvement in sound quality. Background of the modification: When we convert the UL40-S2 amplifier into a Triode amplifier, we connect the screen grid to the anode in each EL34 pentode. This is how a pentode is converted into a triode. This simple action delivers two results. Firstly, the lines in the Ia-Vak-Vgk characteristics get closer toward each others and move to the right side, resulting in a smaller maximum output power. Secondly, the lines in the characteristics get a steeper slope. This means that the effective plate resistance of each EL34 has become smaller, resulting in a larger and better damping of the loudspeaker (higher damping factor). How can a change in screen grid connection have such a clear influence? Because the screen grid functions like a weak control grid on which a certain amount of local feedback is applied. The screen grid receives as a local feedback signal the same alternating voltage as present on the anode. However, when we connect the screen gird to the Ultra-Linear tap (tap 2 or 4) on the output transformer, then only 1/3 of the alternating voltage at the anode is sent to this screen grid as a local feedback signal. This is because the UL-tap is placed on 1/3 of each primary halve. Compared to the Triode situation we now have less (1/3 part) local feedback. The lines in the Ia-Vak-Vgk characteristics do hardly move to the right side of these characteristics, which means that the output power stays large. Their slope is a little less steep as compared to triode, but much steeper as compared to pentode, resulting in a reasonable amount (although less as compared to triode) of damping of the loudspeaker. Compared to Ultra-Linear, the essence of the Super-Triode circuitry is that I apply extra local feedback at the cathode. I connect each cathode to a separate winding on the output transformer. This gives me the extra local feedback to make the power tube function as a Triode, while the output power stays large (the position of the lines in the Ia-Vak-Vgk characteristics is only determined by the local feedback on the screen grids and not by the local feedback on the cathode).
How to make Super-Triode without a separate cathode winding?: As said before, the output transformer inside the UL40-S2 contains no separate cathode winding. I do need extra local feedback to change the Ultra-Linear tube into Triode behavior without loosing output power. Fortunately there exists another technique of local feedback. This technique applies local feedback at the control grid of the power tube instead of at the cathode. See for details figure 3. R1 R2 C 1 2 3 Vht 4 inverse R1 R2 C Fig. 3 : New Super-Triode circuit with control grid feedback. Copyright 2004 Vanderveen A certain part of the alternating voltage at the anode is sent to the control grid through a voltage divider (created by R1 and R2. Forget C for this moment. Its function only is to stop the DC voltage at the anode to enter the control grid). When I make the voltage division through R1 and R2 equal to the turns ratio of the cathode winding to the primary winding, then the control grid will receive the same amount of local feedback as in the Super-Triode circuit. So, only with two resistors (and a capacitor as DC stopper) I can reach the same goal as with a separate cathode winding. Is this way of thinking correct? Yes, else I would not have published. Also, there is nothing new under the sun. See my book Modern High-End Valve Amplifiers based on toroidal output transformers, page 240, figure 20.1. There the resistors R and R6 perform the same trick. What resistance should the resistors R1 and R2 have in the UL40-S2? Look into the schematics of this amplifier, to see that R1 in figure 3 equals R12 and R22 in the UL40-S2, while R2 and C in figure 3 have to be added as new components in the UL40-S2. R12 (in UL40-S2) = 82k (or series circuit of 47k resistor + 0k trim pot, see later) R22 (in UL40-S2) = 100k In parallel to R12 and R22 a Styroflex capacitor of (see remarks 28-11-04) Rnew (in UL40-S2) = (This Rnew equals R2 in figure 3) Cnew (In UL40-S2) = 10nF/1000V (This Cnew equals C in figure 3). Why do R12 and R22 not have the same resistance? The output impedance of the upper halve of the phase splitter (6N1P) has an effective plate resistance of about 18k. This resistance is in series with R12, meaning that R12 + 18k = R22. The output impedance of the lower halve of the phase splitter is negligible due to the large amount of local feedback around this lower halve circuit. Therefore R22 does not need extra compensation. The resistors R12 and R22 should be ¼ Watt power resistors. The capacitor C can be ordered at Farnell, order number 106-364 (order per pieces).
How to implement this modification? Remove R12 and R22 (both were 1k) in the UL40-S2 and place there the new resistors R12 = 82k and R22 = 100K. Now add above the PCB the series circuit of Rnew = and Cnew = 10nF. See for details figure 4. R12 = 82 k R12 = 82 k f U+ S2 GND 10 nf 10 nf R12 P4 UL40S2 COPYRIGHT 2000 IR BURO VANDERVEEN PLUS HIER COMPONENTENKANT Plaats hier R-C Solderen, zie manual R22 C1 - C1 - R22 + + P3 P2 P4 P3 R12 10 nf 10 nf R22 = 100k 40s2-4 Copyright 2004 Ir. Bureau Vanderveen : 17-11-04 R22 = 100k Figure 4: The modification of the UL40-S2 into a Super-Triode amplifier (28-11-04) What are the results of this modification?: Let me start with some measurements. The output power equals 29 Watt into a 4 Ohm loudspeaker. This is the same as with the former Ultra-Linear circuit. The output impedance at the loudspeaker terminals equals 2. Ohm. This is equal to a damping factor of DF = 8 / 2. = 3.2 and this is real Triode behavior, as expected. The subjective sound character of the amp really is pure Triode. The sound is very clean, with deep depth in the soundstage and an open detailed resolution of every voice and instrument. Compared to this excellent result, the Ultra-Linear mode sounded more muddy. The dynamic character (output power) of the Ultra-Linear mode was good, and with the Super-Triode it again is good and powerful (as expected). The bass sounds tight, loud and well controlled. But., there is always a but. Every advantage comes with a disadvantage at another spot. So, where to look for the disadvantage? The frequency range gets more restricted. Before it was 3dB at 80kHz, now it equals 3dB at 40kHz. The original Super-Triode circuit with cathode winding did not show such a disadvantage. Did I do something wrong? No, because in this new circuit the signal from phase splitter to EL34 tubes first has to pass rather large resistors (approx.100k). These resistors create a low pass filter with the capacitance (a few picofarad) at the control grid of the EL34, thus restricting the 3dB frequency range to 40kHz. I decided to leave this disadvantage as is, because with CD s (they are restricted to 22kHz) I have not been able to hear any disadvantage in this respect. I only hear clear improvement. I know that when I would try to solve this disadvantage as well, it would create other problems and no further improvement. So, let it be.
An unexpected extra advantage: While building and testing the modification in my amplifier, I did not directly implement R12 = 82k, but first used a series test circuit of R12 is 47k plus a trim potentiometer of 0k. I now call this 0k trim pot P. See figure and the pictures. P = 0k 47k + P = R12 P = 0k 10 nf 10 nf 47k R22 f U+ S2 GND P P P4 P3 UL40S2 COPYRIGHT 2000 IR BURO VANDERVEEN C1 - + PLUS HIER P2 COMPONENTENKANT Plaats hier R-C Solderen, zie manual C1 - + P4 P3 47k R22 10 nf 10 nf R22 = 100k 40s2- Copyright 2004 Ir. Bureau Vanderveen; 1-11-04 R22 = 100k Fig. : The modification with R12 = 47k plus a series trim pot P (28-11-04)
With this trim pot P I could compensate very careful for the difference in output impedances of both halves of the phase splitter. While doing so, I noticed something unexpected to happen. I was measuring the output signal at the speaker terminals with an oscilloscope (mv/div.) There was no signal present at the amplifiers. I noticed that the hum-level at the output changed while changing the resistance of P. Before, with trim pot P4, I carefully had made the quiescent currents equal of both EL34 tubes per channel with an accuracy of 0.1 ma. This can be checked by measuring equal DC voltage drop over the resistors R14 and R21 (or zero Volt between the two EL34 cathodes). Consequently there was no unbalance in quiescent currents and this could not cause this hum. This effect made me aware of something very important. The supply voltage Vo always has a ripple voltage on top of the DC voltage Vo. This ripple voltage only balances to zero in the balanced output transformer when both EL34 tubes have exactly the same effective plate resistance. From the former given description, it is understandable now that this effective plate resistance can be changed with the resistance of trim pot P. Why is this so important? Because I have never seen any valve circuit in which you can independently change three parameters of a power vacuum tube. See the original UL40-S2 schematics: Trim pot P4 makes the quiescent currents of both power tubes equal Trim pot P makes the plate resistances of both power tubes equal Trim pot P3 makes the effective amplification of both power tubes equal This means that we can trim the power tubes to exactly equal behavior on three independent parameters. Such a feature I have never seen before. Therefore I now give a very precise procedure how to bring the Super-Triode UL40-S2 amplifier to optimal balanced behavior. 1) Connect a DC-voltmeter (200mV-setting) to both cathodes of the two EL34 tubes. Trim with P4 for zero voltmeter reading. Then both quiescent currents are equal. 2) Connect an oscilloscope (mv/div.) to the output of the amplifier. Trim with P for minimal humlevel (100 or 120 Hz, hum smaller than 1 mv effective). Make sure that P1 (at amplifiers ) is closed to prevent any hum blurring this measurement. Now both plate resistances are equal.
3) Connect an oscillator (1kHz sinus) to the amplifier. Load the amplifier output with a 4 Ohm dummy load. Check the output with an oscilloscope. Slowly turn the potentiometer (volume control) P1 to higher setting up till the output voltages is at the edge of clipping. Now trim with P3 for symmetrical clipping. (It is interesting, during this procedure, to listen to the soft sound generated by the output transformer. It is noticeable that the second harmonics of 1kHz disappears at correct setting of P3). An earlier given test method for trimming P3 with a 100Hz square wave now will give a very stable result without any overshoot because the EL34 tubes behave exactly equal. 4) Repeat the above given procedure for both channels of the stereo amplifier. Connect dummy loads of 4 Ohm to the outputs of the two channels and connect a two channel oscilloscope to the amplifier outputs. Set the volume potentiometer P1 at 3.00 o clock. Apply an equal of 1kHz sinus to both channels up till the output level at both outputs equals 8 Volt peak to peak. Trim P2 for equal output voltage. With this procedure the left-right balance has been corrected. (Why P1 at 3 o clock? Because the Alps volume controls have some deviation between its two channels above 3 o clock. The mostly used volume setting is below 3 o clock (I mostly listen at 12 o clock), so we should make the left-right balance optimal in that range of P1). Summary: The Super-Triode modification gives the UL40-S2 amp maximum output power of 29 Watt per channel, with the damping and clarity of a pure superb sounding Triode amplifier. The modification destroys nothing in the amp. Only a few inexpensive components have to be added or changed in value. Especially trim pot P offers new ways to balance the power tubes of each channel on three independent parameters. After this balancing the distortion is truly minimal, resulting in an extreme open and clean and powerful sound reproduction. I wish you much success with this simple and very good sounding modification. Zwolle; 12-october 2004; Ir. Menno van der Veen, Ir. Bureau Vanderveen This modification and publication: Copyright 2004 Ir. Bureau Vanderveen; Super-Triode: Copyright & Registered1998 Ir. Bureau Vanderveen
Remark (Nov-28-2004): A customer told me that his amplifier started to oscillate (at 640 khz) with the modification. I researched his amp and found the cause. There was a capacitive feedback between the screen grid G2 and the control grid G1, which was not damped because R12 and R22 have much larger resistances in this modification (about 100kOhm) compared to the old situation (where R12 and R22 were 1kOhm). I stopped this oscillation by placing capacitors of (Styroflex) in parallel with R12 and R22. To be sure that such an oscillation does not occur (it did not happen in my amplifier which I used to test the modification), I placed these new capacitors in all the relevant drawings and schematics. The capacitors have the extra advantage that they widen the 3dB bandwidth again to 70kHz, because they compensate the EL34 s capacitance between control grid and cathode. UL40-S modification; See "Modern High-End Valve Amplifiers..." page 199 Cnew UL40-S modification R10 82 k Rnew 100pF Rnew Cnew R16 100 k 100pF R16 100 k Rnew Cnew 100pF 100pF Cnew Rnew R10 82 k