From reaike–(at)–x.netcom.com Tue Jan 2 20:19:33 CST 1996
Article: 4797 of rec.audio.tubes
From: reaike–(at)–x.netcom.com(Randall Aiken )
Subject: Re: 6L6 family
Date: 3 Jan 1996 00:27:58 GMT
References: <4cakqr$ea--(at)--ewsbf02.news.aol.com> <960101195059347--(at)--elcom.gen.nz> <960103041220348--(at)--elcom.gen.nz>
X-NETCOM-Date: Tue Jan 02 4:27:58 PM PST 1996
In <960103041220348--(at)--elcom.gen.nz> stephen.delf–(at)–elcom.gen.nz
(Stephen Delft) writes:
>Good question, but I did mean “about” 40mA. I was being a bit cautious
>about estimating worst-case dissipation. 35mA will do just fine. I
>found on some smaller amps that the overdrive went a bit more smoothly
>(smaller kinks in output wave) if the tubes were biassed to a
>”slightly” higher current above the minimum needed to remove obvious
>crossover distortion on a scope.
>BTW…about these kinks or notches which appear just beyond clipping
>level… What is the cause of this? …It is certainly load sensitive
>- eg for 2 x 6L6, a 4k plate load neds a bit more no-signal bias
>current to hide the kinks: with a 6k load, a bit less bias current.
>They will always appear eventually in any class AB output I have
>tried, but I like to be able to clip the signal, plus a little bit,
>before they appear. No technical reason – just a sound preference.
The “kinks” are crossover distortion caused by switching from one tube
(or parallel set of tubes) to the other in a push-pull amplifier. The
closer you are to class B operation, the sooner you will see these
kinks. Changing the plate impedance changes the AC load line the tube
is operating on and will change the point at which this problem occurs.
One factor that aggravates the situation is the shifting of the bias
point when an AC coupled push-pull stage is driven to clipping. The
grids of the output tubes are no longer biased in the negative region
where they present a high input impedance; instead, they are pushing
into the positive grid region where they act as a forward biased diode
in conjunction with the cathode.
This diode acts to “clamp” the tops of the input waveform to near the
cathode potential, and, since the driver tube is AC coupled to the
grids, increasing the amplitude of the input signal to the grids will
force the waveform downward. This is the equivalent of making the bias
more negative because the average value of the AC waveform (the zero
crossing) is now more negative.
This effect happens on both sides of the push-pull circuit, resulting
in more and more crossover distortion as you drive the tubes harder.
In extreme cases, particularly if the coupling caps and bias resistors
are of a high value resulting in a long RC time constant, the amp can
cut completely off for a small amount of time, especially on sharp
transient signals. This “blocking” causes a particularly harsh
“cutting in and out” sounding distortion. This is also why the “scope
crossover distortion at onset of clipping” method of biasing tubes is
not very accurate.
There are two solutions (that I have used, there may be others) to this
problem: one is to drive the grids with a low impedance source that is
DC coupled to the grids. The best approach is to use a cathode
follower with the cathode resistor returned to a voltage source more
negative than the lowest signal level. AC couple the driver and sum in
the bias voltage at the grid of the cathode follower, and DC couple the
cathode to the power tube grids. Alternatively, a step-down
transformer-coupling with the bias fed to a secondary center-tap should
also work, as this is commonly used in class AB2 power amps to allow
positive grid drive. The main disadvantages of this are the cost of a
transformer and possible hum pickup.
The second method is to add what is usually called a “grid stopper”
resistor in series with the power tube grids. However, in this case it
is not functioning to prevent parasitic oscillation, but rather to
limit grid current flow on large input swings that drive the grid
positive. A large value (around 47K – 100K) is required to completely
eliminate the problem. Beware, however, that most tubes have a
recommended maximum input resistance for fixed-bias operation to
prevent tube destruction caused by grid current flow due to
imperfections (gas) in the tube. This is generally specified at around
100K, although some older amps use up to 270K bias feed resistors. The
total value will be the value of the bias resistor in series with the
grid resistor. Also, the grid resistor in conjunction with the input
capacitance of the power tube will form a low-pass filter that will
attenuate the high frequencies.