I've found one instance in an old audio book that talks about the size
of the grid resistance. RCA indicates 50K max for fixed bias and 500K
max for cathode. Like most of the old audio handbooks, it generalizes
too much, leaving out technical details.

Essentially it states that positive charged ions from gassy tubes are
repelled by the plate and are attracted to the negative grid. As they
collect on the grid, one needs small grid resistance to bleed off the
charge, else it leads to a positive grid and increased bias current.
The handbook states that this is not a problem with cathode bias as
the increased plate current self-compensates with the biasing action
of the cathode resistor. So you can get away with the larger grid
resistance. Ok, I'm cool with that.

Question 1: Is this a really big problem? With modern production
tubes? I'm trying to run Sovtek 2A3's at about 20W static.

Question 2: Right now I'm strictly running fixed bias and want to try
to stay with it for adjustibility. Can one add a small cathode
resistance to compenstate for the grid buildup, if any, and keep the
fixed bias setup? I guess a small cathode resistor would have to be at
least big enough to develop a sizable drop across it for compensation,
but how big in general?

As indicated by the book, the lower grid resistance values load my
6SL7 gain stage badly and I want to kick up the grid resistance back
up to around 500k.

Any suggestion?

TIA
Gilbert

Gilbert Bates said:

I've found one instance in an old audio book that talks about the size
of the grid resistance. RCA indicates 50K max for fixed bias and 500K
max for cathode. Like most of the old audio handbooks, it generalizes
too much, leaving out technical details.

Essentially it states that positive charged ions from gassy tubes are
repelled by the plate and are attracted to the negative grid. As they
collect on the grid, one needs small grid resistance to bleed off the
charge, else it leads to a positive grid and increased bias current.
The handbook states that this is not a problem with cathode bias as
the increased plate current self-compensates with the biasing action
of the cathode resistor. So you can get away with the larger grid
resistance. Ok, I'm cool with that.


I was under the impression that the effect you describe is caused by
loose cathode particles that get attached to the grid, and are heated
by the adjacent cathode enough to start emitting electrons by
themselves, making the grid more positive.


Question 1: Is this a really big problem? With modern production
tubes? I'm trying to run Sovtek 2A3's at about 20W static.


I've observed the above with some pairs of Chinese KT66s and KT88s.
They were used with Rg of 330 kohm, which was enough to let them drift
away from their (cathode-biased!) working point.

Lowering Rg to 100 k solved the problem.


Question 2: Right now I'm strictly running fixed bias and want to try
to stay with it for adjustibility. Can one add a small cathode
resistance to compenstate for the grid buildup, if any, and keep the
fixed bias setup? I guess a small cathode resistor would have to be at
least big enough to develop a sizable drop across it for compensation,
but how big in general?


Dunno if a cathode resistor would entirely eliminate this effect, due
to what I said above.


As indicated by the book, the lower grid resistance values load my
6SL7 gain stage badly and I want to kick up the grid resistance back
up to around 500k.


Isn't 100...220 k enough?
How about a somewhat beefier driver configuration?
2A3 isn't that hard to drive I'd guess?

--

"Audio as a serious hobby is going down the tubes."
- Howard Ferstler, 25/4/2005

On Mon, 20 Jun 2005 21:43:57 +0200, Sander deWaal
wrote:

Gilbert Bates said:

I've found one instance in an old audio book that talks about the size
of the grid resistance. RCA indicates 50K max for fixed bias and 500K
max for cathode. Like most of the old audio handbooks, it generalizes
too much, leaving out technical details.

Essentially it states that positive charged ions from gassy tubes are
repelled by the plate and are attracted to the negative grid. As they
collect on the grid, one needs small grid resistance to bleed off the
charge, else it leads to a positive grid and increased bias current.
The handbook states that this is not a problem with cathode bias as
the increased plate current self-compensates with the biasing action
of the cathode resistor. So you can get away with the larger grid
resistance. Ok, I'm cool with that.


I was under the impression that the effect you describe is caused by
loose cathode particles that get attached to the grid, and are heated
by the adjacent cathode enough to start emitting electrons by
themselves, making the grid more positive.


Question 1: Is this a really big problem? With modern production
tubes? I'm trying to run Sovtek 2A3's at about 20W static.


I've observed the above with some pairs of Chinese KT66s and KT88s.
They were used with Rg of 330 kohm, which was enough to let them drift
away from their (cathode-biased!) working point.

Lowering Rg to 100 k solved the problem.


Question 2: Right now I'm strictly running fixed bias and want to try
to stay with it for adjustibility. Can one add a small cathode
resistance to compenstate for the grid buildup, if any, and keep the
fixed bias setup? I guess a small cathode resistor would have to be at
least big enough to develop a sizable drop across it for compensation,
but how big in general?


Dunno if a cathode resistor would entirely eliminate this effect, due
to what I said above.


As indicated by the book, the lower grid resistance values load my
6SL7 gain stage badly and I want to kick up the grid resistance back
up to around 500k.


Isn't 100...220 k enough?
How about a somewhat beefier driver configuration?
2A3 isn't that hard to drive I'd guess?

Please exscuse my ignorance on the matter. I've already configured my
chassis for one octal preamp tube while using 1/2 of a 6SL7 as a no
gain buffer for a sub-woofer output, which leaves me with only one
triode section for the 2A3's drive requirements. After I think about
it now, I could wire the sub-out directly without the triode stage
directly off the volume? The 2A3 needs a sizable current drive for the
large input capacitance. The 6sn7 would be a better drive candidate
but doesn't have enough voltage gain with just one stage and I'm
running the 2A3 at higher B+ voltages than the traditional
'sweet-spot' point at around 250V. So with the higher B+ I have more
negative bias voltage requirements which should leave room for higher
drive voltage swings. Initially I've been playing with high gain
configurations with the 6sl7 but results in low idle (drive) current
and high plate resistance in the drive stage. That was my reasoning
with the need to get the 2A3 grid resistance up and not load the plate
resistance of the 6sl7 so much.

I guess I'll see this through using just one triode section of the
6sl7 to drive the 2A3. I'll just have to accept reduced voltage swing
for increased drive current. At this point, I believe I'll move the
6sl7 plate resistor value down to 100k and live with the resultant
gain so that 100k - 220k grid resistor on the 2A3 won't load it so
badly.

So I still have to figure out an acceptable combination of
cathode/fixed biasing so that I can have increased grid resistance and
enough cathode resistance to keep the drift under control (if any)

Gilbert Bates said:

After I think about
it now, I could wire the sub-out directly without the triode stage
directly off the volume?


You could try that, and parrallel both halves of the 6SL7.
That allows you to halve the Ra and thus leave you with enough voltage
swing to drive the grid of the 2A3 and its grid resistor easily.

I'm building a 2A3 PP amp right now, and am using a 6SN7 phase
splitter and 2 x 6SN7 drivers, both halves in parallel feeding each a
grid of the 2A3.
With 300V B+ I reach over 80 Vp-p swing at the anodes, more than
enough for the 2A3s.

--

"Audio as a serious hobby is going down the tubes."
- Howard Ferstler, 25/4/2005

Gilbert Bates wrote:

Any suggestion?

Yes.

Don't operate gassy tubes.

Which you will recognize by their degraded performance, and check out,
long before they start going postive-grid.

Gilbert Bates wrote:

I've found one instance in an old audio book that talks about the size
of the grid resistance. RCA indicates 50K max for fixed bias and 500K
max for cathode. Like most of the old audio handbooks, it generalizes
too much, leaving out technical details.

Essentially it states that positive charged ions from gassy tubes are
repelled by the plate and are attracted to the negative grid. As they
collect on the grid, one needs small grid resistance to bleed off the
charge, else it leads to a positive grid and increased bias current.
The handbook states that this is not a problem with cathode bias as
the increased plate current self-compensates with the biasing action
of the cathode resistor. So you can get away with the larger grid
resistance. Ok, I'm cool with that.

Question 1: Is this a really big problem? With modern production
tubes? I'm trying to run Sovtek 2A3's at about 20W static.

By modern production do you mean tubes made in Russia or china
after 1993?

I have not had much problem with grids going positive in respect to 0V
with cathode bias or
with respect to the -ve fixed bias supply.

NOS GE6550A were quite prone to +ve grids, and the more Ia there is, the
larger the
rise in +ve voltage.

Its a weird phenomenon, becayse in a radio with a 6AV6 detector triode,
the cathode
is usually grounded, and the grid is biased perhaps with 10Mohms,
to get a grid bias of -1v at the grid.
A very small grid current = 0.1 µA flows; electrons are being collected by
the grid and flow
towards the bias supply, ie, 0V.
But with a power tube the grid starts off being -ve when new, then slowly
starts to go +ve. The dozen GE5550A I bought in 1995 all did this over
time.
The rise in grid voltage is limitable by Rg = 100k.
In one amp I used FB and Rg = 390k to begin with and there was up
to 4 volts across the 390k, and since the grid was slightly +ve, it meant
that electrons were being
emitted by the grid, like they are in grid current conditions, only not in
such great
quantities.
But perhaps the very +ve screen support rods in the mica and right beside
the grid1 support
rods allow a tiny leakage of electrons to the screen.


I am still not 100% sure why the g1 goes +ve though. Some say +ve ions,
some say leakage
across the mica inside the tube, whatever, but it happens, and its not a
condition to be confused with
leaking coupling caps, which exhibit a very similar sympton.





Question 2: Right now I'm strictly running fixed bias and want to try
to stay with it for adjustibility. Can one add a small cathode
resistance to compenstate for the grid buildup, if any, and keep the
fixed bias setup? I guess a small cathode resistor would have to be at
least big enough to develop a sizable drop across it for compensation,
but how big in general?

Using an Rk large enough to regulate the cathode current as the normal
Rk does in cathode bias amps would mean that Rk would have to be about
at least 1/2 the usual value of Rk for CB.

I do this myself in my 300 watt amps using 12 x 6550, but I use
Rk about 2/3 the size of the normal full value for cathode bias.
I apply some fixed bias which is 1/2 the value of the actual cathode bias
obtained.
This means the Rk dissipation in the amp is low,
since Pd = E x I and if E is 0.7 times normal, Pd will be substantially
lower,
and the effective Ea will be usefully higher.

it is essential to bypass any substantial Rk used to regulate the Ia,
since the Ra of the tube will effectively increase with Rk
unbypassed, so that Ra' ( effective Ra ) = Ra + ( µ x Rk ).
The power tubes unbypassed Rk would make the Ro of the
amp a lot higher than you'd want.


As indicated by the book, the lower grid resistance values load my
6SL7 gain stage badly and I want to kick up the grid resistance back
up to around 500k.

I wouldn't use a 6SL7 to drive any power tube grid.
I would use a 6SN7, in which can flow a much higher Ia,
so a large V swing across a lowish Rg is possible.

The load seen by the driver anode is its DC supply RL in parallel
with the cap coupled Rg of the following tube.
When you draw the loadline for the combined parallel
load at the anode of the driver, and passing through the selected
Q point for the anode, the combined loadline will intersect the
hrizontal axis below the Ea value for driver triode.
The Vswing can be thus limited by this, and that's why the RDH4
and other books recommend
that the cap coupled RL is should be greater than 3 times the DC RL,
especially for a driver tube driving an output triode.

So with a 6SN7, the DC RL might be 39k, and the 2A3 bias R could be
as low as say 120k, and you still should get a decent low distortion
drive signal.

The 6SL7 is incapable of doing the same thing with thse loads.

The 6SL7 is a very nice sounding input tube where a fair amount of gain is
to be used
and you want to use NFB around a 3 tube line up
like SL7, SN7, 2A3.
I use 6SL7/ECC35, ECC32/6SN7, 13E1 ( SEUL, 25 watts ) line up with 16 dB
of global NFB
and the owner is one very happy dude.
Each of the small input tubes has both halves paralleled.

This effectively lowers Ra for the stages, increases bandwidth and voltage
drive ability
and reduces thd.
The outcome is blameless sound devoid of high levels of artifacts, and
thus allows
the tubes to really sing for their supper well.

Another very nice driver tube for any SET output tube is a 6V6 in triode
with about 12 mA of Ia.......

Patrick Turner.





Any suggestion?

TIA
Gilbert

Patrick Turner wrote:

NOS GE6550A were quite prone to +ve grids, and the more Ia there is, the
larger the
rise in +ve voltage.

Yet the NOS GE 6550A has quite a robust G1 when intentionally driven +
in AB2 hi-pwr (not so hi-fi) applications, & there's hardly another
6550 made that holds up as well in that service. Some others melt,
horror show. But I admit that's venturing out of audiophile territory.

It's hard to beat the 8417 for unexpected G1-pos catastophies in
otherwise good gear. Whether the mechanism stated above as its cause
is accurate or fully understood, is up for grabs, but 8417's testing
gasfree on reliable equipment have done it.

Hi Gilbert !
I never tried modern Sovtek's 2A3 so I can't say.
But I got the same problem time ago with an amp I was working on :
http://www.arrakis.es/~igapop/6b4g_push_pull_amp.htm using
old Svetlana's 6B4G
I just soldered 120k grid resistors and had no problem at all.
But 500k - hmm, don't know...
Best regards,

--
Igor
http://www.arrakis.es/~igapop

"Gilbert Bates" escribió en el mensaje
...
I've found one instance in an old audio book that talks about the size
of the grid resistance. RCA indicates 50K max for fixed bias and 500K
max for cathode. Like most of the old audio handbooks, it generalizes
too much, leaving out technical details.

Essentially it states that positive charged ions from gassy tubes are
repelled by the plate and are attracted to the negative grid. As they
collect on the grid, one needs small grid resistance to bleed off the
charge, else it leads to a positive grid and increased bias current.
The handbook states that this is not a problem with cathode bias as
the increased plate current self-compensates with the biasing action
of the cathode resistor. So you can get away with the larger grid
resistance. Ok, I'm cool with that.

Question 1: Is this a really big problem? With modern production
tubes? I'm trying to run Sovtek 2A3's at about 20W static.

Question 2: Right now I'm strictly running fixed bias and want to try
to stay with it for adjustibility. Can one add a small cathode
resistance to compenstate for the grid buildup, if any, and keep the
fixed bias setup? I guess a small cathode resistor would have to be at
least big enough to develop a sizable drop across it for compensation,
but how big in general?

As indicated by the book, the lower grid resistance values load my
6SL7 gain stage badly and I want to kick up the grid resistance back
up to around 500k.

Any suggestion?

TIA
Gilbert