Background of the VDV E-CHOKE
This was true until recently. I performed measurements on my fantastic amp and saw something that really stroke me. Look at figure 1. There a 1kHz signal is fed through the amp and the spectrum of the loudspeaker signal is measured. What we see is the 1kHz, but also the higher harmonics at 2 and 3kHz. However, what to say about the intensive intermodulation products of 1kHz and 100Hz multiples of the rectified mains voltage?
Fig-1: Spectrum UL40 with 1kHz, 8Vtt in 4 Ohm, without a choke
I can't hear the intermodulation products around 1kHz, because they are masked by the stronger 1kHz signal. What I might hear is the residual 50 en 100Hz hum components, because they are above the masking level. Why then to care about this mesh of disturbance when you don't hear it. This was my thinking, and I was wrong. Study the following figure 2. Here we study the behavior of the UL40 in the time domain, after a sudden wide bandwidth burst signal. Look at the mesh after the burst has stopped. The does not look healthy at all.
Fig-2: UL40 without choke, mark the mesh in the signal floor.
I can tell you, this is what we can hear. Where we should notice silence, we now hear rubbish and noise and .... We do not only have to deal with the masking of our ear in the frequency domain, we also have to study any disturbance in the time domain, in order to understand what is going on. My wish was to recreate silence in the signal floor.
Lets suppose we can solve that with a standard choke. Next figure 3 shows the results. I applied a 10H choke after the rectification and buffering, plus an extra capacitor of 330uF, all in the very well known pi circuit. The results are striking.
Fig-3: UL40 with 10H choke and extra 330uF/450V capacitor
The 100Hz intermodulation products are almost gone. This looks much cleaner. Before we studied the results also in the time domain. Let us repeat that now, and study figure 4.
Fig-4: UL40 with 10H choke and extra 330uF capacitor
This does not look healthy at all! Look at the residual signals at the signal floor. The 50 and 100Hz components are smaller, but time wise they swing and oscillate over a decent amount of time. I know that 10H plus the capacitors of 330 uF each are not tuned optimally, but this example helps us to find out where to focus at.
The next figure shows the results when I apply my new VDV E-CHOKE in the UL40-S2 with an extra 330uF capacitor directly after the rectifiers. The other capacitor of 330uF is already on the PCB of the UL40. This new choke has two time settings (see later for more info). In this case the time is set OFF.
Fig-5: UL40 with VDV E-CHOKE, time modus at OFF.
What we see now is an even stronger rejection of 100Hz intermodulation products. Later I will say something about the 50Hz component, which seems not to change at all. These results are remarkably better than a standard huge iron 10H choke. The question now is: how does this choke behave in the time domain? See the next figure.
Fig-6: UL40 with VDV E-CHOKE, time modus at OFF
Apart from the 50Hz component (see later) the signal floor is almost completely clean. Only a tiny 50Hz resonance is visible, which has disappeared after 40 ms. In this time modus the new choke reacts very fast and this setting is meant for class AB amplifiers where the current demand can change rapidly.
But I can even do better. When I switch the time modus to ON, there is an extra improvement of intermodulation rejection. See the next figure.
Fig-7: UL40 with VDV E-CHOKE, time modus at ON
A normal bulky iron choke has no time modus switch. That's a new invention of me, intended for class A amplifiers with a constant current demand. Then large changes in current demand seldom occur and therefore you can tune the time behavior of the E-CHOKE very precise, without any concern about the currents drawn. This must be visible in the time domain. See the next figure.
Fig-8: UL40 with VDV E-CHOKE, time modus at ON
As expected any resonance of mains components has disappeared. The signal floor is optimal clean in the time domain. For those who like to measure: after a current change the mains voltage recovers in 10ms in a critical damped manner.
Conclusive: the VDV E-choke does not only remove hum components and intermodulation products in the frequency domain. It also makes the signal floor optimal clean after a signal burst. Now silence becomes silence, and this is what we can hear!
I would like to make a remark about the burst-decay measurements. It looks as if a 50Hz component stays present. This is caused by the computer measurement setup, this component is not present in reality.
With oscilloscope measurements the following results were shown. The amplifier was drawing a quiescent current of 4 x 60mA. The ripple voltage over C-in was 7Vpp. After the E-CHOKE with time modus at off the residual ripple was 50mVpp, equal to more than 40dB suppression. With the time modus set at on, the residual ripple was 3mVpp, more than 60dB suppression.
VDV E-CHOKE specifications
Application in the UL40-S2
Fig-9: Power supply of the UL40-S2
In the next figure the application of the E-CHOKE is shown.
Fig-10: UL40-S2 with VDV E-CHOKE
This figure shows that the plate voltage, after rectification by D1-4, directly goes into the input of the E-CHOKE where it is buffered by the input capacitor C-in = 330uF/450V. The E-CHOKE effectively is placed between D1-4 and the capacitor C12 = C-out = 330uF/450V, which is already present on the UL40 PCB.
How to place and wire the E-CHOKE is very simple, see the following figure.
Fig-11: Wiring E-CHOKE in UL40-S2
1) Place the E-CHOKE at the left side above the UL40 PCB. Use three glue spacers and not four, because the right upper hole is just above the mounting screw of the OPT. Therefore use the right middle hole. I always use extra silicon glue for the spacers. It deals well with the roughness of the metal case and can withstand the heat of the valves.
2) Unsolder the left sides (the cathodes) of the two upper diodes. Reconnect these left sides (in the air) and connect that joint to a red wire. Isolate this connection with heat shrinking tube. Solder this red wire to the IN-connection of the E-CHOKE.
3) Solder a wire (black) from GND on the E-CHOKE to the anode side (right side) of the lowest fourth diode.
4) Solder a wire (blue) from OUT on the E-CHOKE to the earlier PCB connection of the left side of the upper two diodes (where they were unsoldered in step 2).
That's it. Please check your work carefully.
a) TIME-bridge at OFF: UL40 in its original mode, with cathode resistor and cathode capacitor; UL40 in super triode mode; UL40 with the auto bias module for quiescent currents smaller than 60mA per tube.
b) TIME-bridge at ON: UL40 in all configurations with auto bias with quiescent current larger than 60mA per tube. There the amplifier almost completely operates in class A.
I have some hesitation writing down the following. I really hope that you will read it with the utmost attention, else you might destroy the E-CHOKE and that certainly is not my goal.
1) Switch the amplifier OFF and wait some minutes for total discharge of the large 330uF capacitors.
Suppose one has a different opinion about this procedure. Lets assume that the time was at ON. Or suppose you timely connected IN or OUT to GND. Well, you will end up with a dead E-CHOKE. This explains my hesitation. I have done my utmost to clearly tell you what to do, so mistakes are excluded. Is that true? Experience has taught me that ....
How to apply the E-CHOKE in other valve amps?