• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Neutralizing An Amplifier

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi all guy´s. Some years ago I saw a circuit in where a push pull driver around a 12AX7, and neutralizing them crossing 1.5pF between their grid and plate of different output circuits, like it is done in RF circuits. Did any of yours try this method in audio? What is your opinion or taste about it?

Osvaldo.
 
Some methods do, some don't. It's noteworthy that Miller effect is itself frequency-dependent (it's a capacitor), so you need a frequency-dependent method (in the same direction!) to cancel it for all frequencies.

Three common methods are:
1. Resonating Cmiller with a neutralizing inductor
2. Inductive loop from input to output coupling circuits (negative feedback)
3. Cross-connected capacitors

Of these, 1. is inverse frequency dependent, so it truely works only in narrowband applications. I would guess, in a circuit which requires heavy neutralization, you wouldn't get more than 20% bandwidth or so (i.e., +/-5MHz out of a 50MHz VHF amp). 2. is frequency independent, so it works over a wide bandwidth (with good coupling, it'll work over the full bandwidth of the input/output coupling coils/transformers). Unfortunately, miller is not, so it only needs a certain amount of neutralization over part of that range, and more elsewhere. So the bandwidth would be higher, but not wideband.
3. has the potential for wideband operation, as long as the phase shift between phases is maintained well. This requires tight control over parasitics and very good transformers.

Now, for audio purposes, capacitance is capacitance, and adding more won't magically fix anything. You can get some PP enhancement with a small miller-canceling cap, but you can't reduce the effective capacitance below, I would guess, about double the interelectrode capacitance. Attempting to do more will load it down more, and make it more prone to weird things like oscillation.

Tim
 
I did try netralisation caps (grid to opposite anode) on a differential splitter using 12AX7.
I could'nt get it stable, always ended up with an oscillator. I think I was using a cap of around 10pF which was probably too big.

Try putting a short piece of wire on each of the anode and opposite grid and then twist the wires around each other, more twists => more capacitance, there was a "rule of thumb" of pF per twist for a particular type of wire but I can't remember it. For 1.5pF you would probably only need 2 or 3 twists.

Cheers,
Ian
 
gingertube said:
I think I was using a cap of around 10pF which was probably too big.
2pF might be nearer the mark. IIRC the valve itself has 1.6pF, plus strays. 10pF will tip the balance too far the other way.

As this technique balances capacitance with capacitance it is inherently wideband. The issue with RF neutralisation is that the output or input circuits are often narrowband so they introduce phase shifts. That will not normally arise with audio, at least not to the same extent.
 
IIRC, Llyn Olson tried this out with the Karna amplifier - and gave up due to parasitic oscillations, finding it produced more problems than it solved.

Counterculture - seems you took some objection to the discussion? it always amazes me how easy it is to read into somebody else's messages what wasn't intended and I suspect you took offence at something not intended. I value all the contributions I see on the forum, always something to learn from folk such as yourself and from others.
 
Last edited:
Note, BTW, that coupling capacitance to the opposite side generally makes a multivibrator. This will be the dominant action in audio applications, with resistive coupling.

You're adding positive feedback, which acts to oppose internal negative feedback, but the gain depends on operating point. When adding PFB (in static or dynamic applications), you can only use as much as the transfer curve's peak gain point. More and the gain at that point "goes past" infinite, i.e. it skips past hysteretically, turning the transfer curve from a curved line, to a sigmoidal curve, to a discontinuous loop.

At RF as well, gain and dynamics change with bias point and power output, meaning neutralization must as well. This is especially pronounced in semiconductors due to their voltage-dependent capacitances.

Tim
 
It depends on if it is a modern short plate or classic long plate the miller cap of each is widely different . The short plates have less and have greater bandwidth all 12ax7 are not close to being small range referenced to Miller cap. Variable air cap may work the best . I willing to say you may need to tune it with every 12ax7 change to stay out of the range that caused Llyn Olson's amp problems.
 
When you begin to add the size of the neutralizing capacitor the high frequency response will rise. This change can be easily observed with audio generator and AC-voltage meter.
In such case that the gain will be rising above 100 kHz (compared to 10 kHz), then the neutralizing capacitor is already too big and the possibility to instability is obvious.
Assume that you have a circuit that has frequency response, say -10 dB at 100 kHz
and you want to have flat response up to 100 kHz. Then simply take the reference level at 10 kHz, move the frequency of the input signal to 100 kHz and measure the output voltage. Then adjust neutralizing cap(s) until the level in increased to -1 dB...0 dB.
That's all. A change of 0.5 pF has huge impact when adjusting.
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.