DECWARE / High Fidelity Engineering Co.

A U D I O... P A P E R


 Oct 2009
by Steve Deckert


In Part I  we introduced the Christmas comes early mod (aka CCE mod) but didn't get into  specifics  about what is actually happening inside the tube to create this most apparent sonic improvement. The mod improves clarity, dynamics and seems to show a significant removal of typical edginess or grain in the sound.  This paper will show what is known so far and hopefully further perpetuate the discussions going on in various amp building forums around the Internet.


For starters it should be understood what context this mod is having this result in.  The context is Decware amplifiers, put simply, and therefor this mod may not apply to or even be wise to implement in other amplifiers.


As mentioned in part one, this mod is only applicable on tubes where the suppressor grid is not internally connected to the cathode.  The EL34 tube  as well as the SV83 are both examples  of this type of tube and  are the two main output tubes that all  Decware amplifiers have been based on since the original Zen Triode amp in 1996.


As predicted some basic measurements don't seem to show an effect or explain the resulting improvement in clarity.  This is why in the interest of the greater good I have been inviting discussion on the topic and writing these papers. 

If we look at a triode with respect to the electrons that become dislodged from the positively charged plate and have nowhere to go but back to the plate we have to ask the following question: When the stray electrons re-attached themselves to the plate where was their new location with respect to where they left?  It's unlikely they bounced back to fill the same location they occupied before they were dissloged.  And why would this even matter?

In a pentode where the suppressor grid is kept at the same negative DC voltage as the cathode, the dissloged electrons are also pushed back to some location on the plate before they hit the screen grid and cause problems with that.  My focus is on the dislodged electrons coming off the plate and being pushed back to the plate which is happening in both the directly heated triodes and pentodes.

I can see the path in a large bottle triode that the dislodged electrons take before they return to the plate could be a longer path than a pentode where the suppressor grid greatly reduces the gap from the plate.  You might think that if this makes any difference at all, pentodes would sound better than triodes, especially if they are triode wired, but they usually don't.  Again supporting the assumption that the distance the secondary electrons travel back to the plate and where they  land on the plate is inconsequential.  However, the way I'm looking at it is that the mere fact there are secondary electrons that have to be dealt with is a less than ideal thing. 

Due to the fact that the suppressor grid deals with the dissloged electrons from the plate, and that modifying it's operation changes the sound in the way it does leads one to believe that the amount of or behavior of the dissloged electrons have changed.

Because the change is for the better, you can only assume that dissloged electrons are probably a bad thing for clarity and dimensionality.

I've listened to a fair numbers of triodes and triode wired EL34's over the past 20 years and each has it's strengths.  The triode is not categorically superior to the triode wired EL34 as some would think.  However when the suppressor grid of the EL34 is modified with the cap there is a huge difference in the presentation of the amplifier for the better, despite the fact that typical measurements show no substantial change has occurred.   I think one of two events are taking place:  A) the phase angle of the AC component on the suppressor is modified resulting in a deceleration of the electrons as they pass through this sparsely wound suppressor grid on their way to the plate resulting in less dislodged electrons in the first place.  B) the electrons that are being dislodged from the plate are being absorbed by the suppressor grid.  In case A, there are less electrons being dislodged from the plate so there are less secondary electrons being re-assembled on the plate.  In case B, there are the same amount of electrons being dislodged from the plate, but less secondary electrons being re-assembled on the plate, giving us the same result as case A. 



Decware amplifiers  are self biasing designs that  use a cathode resistor that  is often bypassed  with a cap .   For the purpose of this article we will focus on  one particular single ended triode design that uses a triode wired EL34.  This is the amp that this mod was discovered on, so it seems a fitting place to start.

Below is the output stage of this amp (or model SE34I.2)

output stage schematic

NOTE:   Suppressor Cap 0.1uf DC Voltage reading may start at 27 VDC and then drop to 13.5 VDC when the VOM is used to measure it.  The voltage will then rise back to 25 VDC leaving 2 Volts DC from the suppressor to ground.   With the 20uf cap lifted the voltage drops to 1 VDC.


Below is an approx 1 volt signal on the control grid at various frequencies in a working amplifier with an 8 ohm loudspeaker connected to it's output and set to a normal listening level (about 1/2 watt). 

The chart below shows what the suppressor grid (G3) is doing at different frequencies with and without the 20uf cathode bypass cap installed on the cathode resistor.  Also in 3 locations starting at 200Hz the 0.1uf cap was jumped out with a wire, effectively removing the CCE mod from the circuit.  Also notice that in this frequency bandwidth the DC voltage between the suppressor Grid G3 and ground is not only there, but steady throughout the spectrum.

AC Signal at G1
 DC at G3  G3 AC
G3 AC w/ 20uf cap lifted
Wire w/ cap lifted
DC at G3
20Hz 890 mv
262 mv
263 mv
907 mv
26.4 535 mv
327 mv

1.004 v
26.4 613 mv
239 mv

999 mv
26.4 400 mv
60 mv
406 mv
60 mv
994 mv
26.4 443 mv
40 mv

982 mv
26.4 397 mv
31 mv

991 mv
26.4 331 mv
28 mv
335 mv
28 mv
3kHz 985 mv
26.4 287 mv
31 mv

5kHz 987 mv
26.4 220 mv
31 mv
199 mv
28 mv
10kHz 985 mv
26.4 128 mv
30 mv

975 mv
28 mv


It would be hard to interpret much meaningful data from looking at the voltages, in fact when compared to a piece of wire there is little difference made by the suppressor cap.

I also did some other measurements comparing the cap to a piece of wire in this same amplifier:

Frequency response from 20Hz to 20kHz +0.42 dB
+0.43 dB

-2.27 dB
-2.31 dB
-76.1 dB
-76.0 dB
Dynamic Range
75.9 dB
75.8 dB
Phase angle was virtually the same both ways.

Again, everything actually improved slightly except IMD which got worse, however these values are so close they become pretty much meaningless with respect to understanding how the CCE mod makes the tube sound better.

You'd almost have to capture it in real time reacting to the sudden dynamics of music vs. a simple AC sign wave if any of this data is to become  meaningful.   The kind of things I would like to see in a series of real time snapshots are beyond my resources and knowledge to actually measure so naturally no way to prove my theory. 

I came to the theory by process of elimination and driven by the fact that my ears are telling me something profound is happening even though I can't find it.  I'll follow the discussions on various forums around the Internet regarding the CCE mod or Hazen Grid Mod as it's also called, with great interest. 

-Steve Deckert

From the RCA transmitting tube manual on air-cooled tubes from 1938: 

In all radio tubes. electrons striking a positive electrode may, if moving at sufficient speed, dislodge or "splash out" other or secondary electrons. In diodes and triodes, such secondary electrons produced at the plate usually do not cause any trouble because no positive electrode other than the plate itself is present to attract them. These electrons, therefore, are eventually drawn back to the plate. In tetrodes, the screen (operating at a positive potential) offers a strong attraction to secondary electrons when the plate voltage swings lower than the screen voltage. This effect limits the permissible plate swing for tetrodes because the major portion of the space current then goes to the screen rather than to the plate. The plate swing limitation can be substantially removed when a fifth electrode, known as the suppressor, is placed in the tube between the screen and the plate. Such five-electrode types are called pentodes.

The suppressor in a pentode is usually connected to the cathode, or to a low positive or negative voltage, depending on the tube application. Because of its negative potential (in any case) with respect to the plate, the suppressor retards the flight of secondary electrons and diverts them back to the plate, where they cause no undesirable effects. Thus, in pentodes, the plate voltage may swing below the screen voltage."


Decware is a trademark of High Fidelity Engineering Co.
Copyright 1996 1997 1998 1999 2000 2001 2002 2003 2004  2005 2006 2007 2008 2009
by Steve Deckert