[Alex]

Prompted by the "flipper's" thread, I am wondering is it generally better to
bypass screen grids of small signal pentode stages to ground or to the
cathode. (It is more robust to solder a heavy cap to an earthed lug than to
a flimsy lug of a valve socket, hi-hi!)

I have never thought about it, and always was grounding. Of course, if the
cathode is bypassed to ground, it makes no difference. But if not? If the
global NFB is applied to an unbypassed cathode resistor?

Now in the wake of the "reinjection" thread I might assume that bypassing
screen to cathode is better, creating a virtual pentode with less pronounced
screen current intercept.

What do you think?

Regards,
Alex

[Patrick Turner]

Alex wrote:

Prompted by the "flipper's" thread, I am wondering is it generally better to
bypass screen grids of small signal pentode stages to ground or to the
cathode. (It is more robust to solder a heavy cap to an earthed lug than to
a flimsy lug of a valve socket, hi-hi!)

The main purpose of the Rk and its bypass cap is to provide good self
regulating bias conditions.

If you try to use battery bias or adjustable fixed bias, Ea moves a lot
without any self regulation.



I have never thought about it, and always was grounding. Of course, if the
cathode is bypassed to ground, it makes no difference. But if not? If the
global NFB is applied to an unbypassed cathode resistor?

In many amps this is what is done; leave out the Rk bypass cap and allow
local current FB and add the global NFB to it.
meanwhile the screen is always bypassed to the cathode, never to 0V,
lest this cause a los of gain.

See the circuit for the classic Mulard 520 etc with EF86 input pentode.



Now in the wake of the "reinjection" thread I might assume that bypassing
screen to cathode is better, creating a virtual pentode with less pronounced
screen current intercept.

Most ppl think this is best. Yo can bypass the screen to 0V with GNF
taken to the cathode, and then you have say 1V of cathode signal
effectively applied to the screen of the pentode and a resulting effect
of the screen gain. I leave you to work out the benefits if any,
but please breadboard and measure it all to find the real truth!!!!



What do you think?

See above.

Patrick Turner.

Regards,
Alex

[Ian Iveson]

Patrick Turner wrote

Prompted by the "flipper's" thread, I am wondering is it
generally better to
bypass screen grids of small signal pentode stages to
ground or to the
cathode. (It is more robust to solder a heavy cap to an
earthed lug than to
a flimsy lug of a valve socket, hi-hi!)

The main purpose of the Rk and its bypass cap is to
provide good self
regulating bias conditions.

If you try to use battery bias or adjustable fixed bias,
Ea moves a lot
without any self regulation.

I have never thought about it, and always was grounding.
Of course, if the
cathode is bypassed to ground, it makes no difference.
But if not? If the
global NFB is applied to an unbypassed cathode resistor?

In many amps this is what is done; leave out the Rk bypass
cap and allow
local current FB and add the global NFB to it.
meanwhile the screen is always bypassed to the cathode,
never to 0V,
lest this cause a los of gain.

See the circuit for the classic Mulard 520 etc with EF86
input pentode.

Now in the wake of the "reinjection" thread I might
assume that bypassing
screen to cathode is better, creating a virtual pentode
with less pronounced
screen current intercept.

Most ppl think this is best. Yo can bypass the screen to
0V with GNF
taken to the cathode, and then you have say 1V of cathode
signal
effectively applied to the screen of the pentode and a
resulting effect
of the screen gain. I leave you to work out the benefits
if any,
but please breadboard and measure it all to find the real
truth!!!!

For this purpose, Patrick, simulation would be much easier,
quicker, cheaper and safer. Quite likely more precise, too,
depending on what real measuring instruments are available
for breadboard testing.

Breadboarding is only useful to check the exact mix of
distortion, assuming you can measure it exactly. More
generally, breadboarding is invaluable where a precise model
of a component is not available, as is often the case with
transformers or uncommon valves when the creation of a model
is not worthwhile.

Ian

[Patrick Turner]

Ian Iveson wrote:

Patrick Turner wrote

Prompted by the "flipper's" thread, I am wondering is it
generally better to
bypass screen grids of small signal pentode stages to
ground or to the
cathode. (It is more robust to solder a heavy cap to an
earthed lug than to
a flimsy lug of a valve socket, hi-hi!)

The main purpose of the Rk and its bypass cap is to
provide good self
regulating bias conditions.

If you try to use battery bias or adjustable fixed bias,
Ea moves a lot
without any self regulation.

I have never thought about it, and always was grounding.
Of course, if the
cathode is bypassed to ground, it makes no difference.
But if not? If the
global NFB is applied to an unbypassed cathode resistor?

In many amps this is what is done; leave out the Rk bypass
cap and allow
local current FB and add the global NFB to it.
meanwhile the screen is always bypassed to the cathode,
never to 0V,
lest this cause a los of gain.

See the circuit for the classic Mulard 520 etc with EF86
input pentode.

Now in the wake of the "reinjection" thread I might
assume that bypassing
screen to cathode is better, creating a virtual pentode
with less pronounced
screen current intercept.

Most ppl think this is best. Yo can bypass the screen to
0V with GNF
taken to the cathode, and then you have say 1V of cathode
signal
effectively applied to the screen of the pentode and a
resulting effect
of the screen gain. I leave you to work out the benefits
if any,
but please breadboard and measure it all to find the real
truth!!!!

For this purpose, Patrick, simulation would be much easier,
quicker, cheaper and safer. Quite likely more precise, too,
depending on what real measuring instruments are available
for breadboard testing.

Breadboarding is only useful to check the exact mix of
distortion, assuming you can measure it exactly. More
generally, breadboarding is invaluable where a precise model
of a component is not available, as is often the case with
transformers or uncommon valves when the creation of a model
is not worthwhile.

You always take the laziest way. I don't.

Simulation measurements of THD etc are NEVER as precise as measurements
of a real breadboarded trial of the simulated circuit. When on earth was
a PC model going to give you more precision than using real tubes in a
real circuit?

Many ppl in the past designed far more circuits than you ever have just
used their mind, a slide rule and an exercize book to maintain a hard
copy of a proposed circuit, then they built it and found out the
reality.

Moon shots or Mars shots are far too complex to breadboard. But not a
circuit with one pentode.

With 3 pentodes and several overlapping loops of NFB and PFB to make a
bandpass filter or bandstop filter it soon becomes impossible to just
estimate and trim values to get the wanted Q of the filter, so a model
is required because the math is far too difficult, and the model allows
you to suck the brains of someone who wrote the model because you cannot
work it out. I rarely need such filter circuits but I do have and
bandpass filter using 3 opamps which I copied from a book which did
actually work as the author said it would to give a constant Q of 50
between 1kHz and 10Khz to allow measurements of multiple harmonics of a
1kHz tone and IMD products. So plenty of specific circuits are available
for copying so simulation isn't needed.

But one pentode in a linear circuit? only daft buggers like you who
rarely ever design or build anything might spend all day ****ing around
in front of a PC before spending time in the workshop.

Have fun being a do-little , but I have to earn money from my time spent
doing real things.

Patrick Turner.



Ian

[Ian Iveson]

Patrick Turner wrote

Simulation measurements of THD etc are NEVER as precise as
measurements
of a real breadboarded trial of the simulated circuit.
When on earth was
a PC model going to give you more precision than using
real tubes in a
real circuit?

As I said, it depends on how accurate your real measuring
insruments are. Also on how correct are the assumptions you
make in using the measurements to come to valid conclusions,
on the ranges of those instruments, and on your ability to
test the real circuit properly without electrocution or
setting yourself on fire.

Simulation usually gives a good approximation for imperfect
models, such as valves and complicated inductors, and a
perfect result for perfect models, such as perfect resistors
and capacitors. Direct measurements of voltages and currents
are perfect in simulations, of course. So it's swings and
roundabouts.

Your distortion measuring equipment has a narrow range, for
example. A virtual spectrum analyser has any range you want,
and any resolution, at any test frequency, but may suffer
from spurious effects due to the simulation process.
Likewise a virtual sig gen: it suffers from the usual
problems of digital generators. So we interpret the results
appropriately, whether we breadboard or simulate. Once we
compared a few simulations with breadboard results, we got
bored with soldering. For ppl without workshops, it's a real
pain getting a heap of real equipment, tools and components
together.

For the rest of what you say, "flipper" makes the
appropriate points. I would add that you contradict yourself
in suggesting that breadboarding is less fuss, and also that
simulating is lazy. Perhaps you have a phobia...the same one
that prevents you from designing your own filters, perhaps?

It's never too late to get an education, and to start
simulating.

Ian



in message ...


Ian Iveson wrote:

Patrick Turner wrote

Prompted by the "flipper's" thread, I am wondering is
it
generally better to
bypass screen grids of small signal pentode stages to
ground or to the
cathode. (It is more robust to solder a heavy cap to
an
earthed lug than to
a flimsy lug of a valve socket, hi-hi!)

The main purpose of the Rk and its bypass cap is to
provide good self
regulating bias conditions.

If you try to use battery bias or adjustable fixed
bias,
Ea moves a lot
without any self regulation.

I have never thought about it, and always was
grounding.
Of course, if the
cathode is bypassed to ground, it makes no difference.
But if not? If the
global NFB is applied to an unbypassed cathode
resistor?

In many amps this is what is done; leave out the Rk
bypass
cap and allow
local current FB and add the global NFB to it.
meanwhile the screen is always bypassed to the cathode,
never to 0V,
lest this cause a los of gain.

See the circuit for the classic Mulard 520 etc with
EF86
input pentode.

Now in the wake of the "reinjection" thread I might
assume that bypassing
screen to cathode is better, creating a virtual
pentode
with less pronounced
screen current intercept.

Most ppl think this is best. Yo can bypass the screen
to
0V with GNF
taken to the cathode, and then you have say 1V of
cathode
signal
effectively applied to the screen of the pentode and a
resulting effect
of the screen gain. I leave you to work out the
benefits
if any,
but please breadboard and measure it all to find the
real
truth!!!!

For this purpose, Patrick, simulation would be much
easier,
quicker, cheaper and safer. Quite likely more precise,
too,
depending on what real measuring instruments are
available
for breadboard testing.

Breadboarding is only useful to check the exact mix of
distortion, assuming you can measure it exactly. More
generally, breadboarding is invaluable where a precise
model
of a component is not available, as is often the case
with
transformers or uncommon valves when the creation of a
model
is not worthwhile.

You always take the laziest way. I don't.


Many ppl in the past designed far more circuits than you
ever have just
used their mind, a slide rule and an exercize book to
maintain a hard
copy of a proposed circuit, then they built it and found
out the
reality.

Moon shots or Mars shots are far too complex to
breadboard. But not a
circuit with one pentode.

With 3 pentodes and several overlapping loops of NFB and
PFB to make a
bandpass filter or bandstop filter it soon becomes
impossible to just
estimate and trim values to get the wanted Q of the
filter, so a model
is required because the math is far too difficult, and the
model allows
you to suck the brains of someone who wrote the model
because you cannot
work it out. I rarely need such filter circuits but I do
have and
bandpass filter using 3 opamps which I copied from a book
which did
actually work as the author said it would to give a
constant Q of 50
between 1kHz and 10Khz to allow measurements of multiple
harmonics of a
1kHz tone and IMD products. So plenty of specific circuits
are available
for copying so simulation isn't needed.

But one pentode in a linear circuit? only daft buggers
like you who
rarely ever design or build anything might spend all day
****ing around
in front of a PC before spending time in the workshop.

Have fun being a do-little , but I have to earn money from
my time spent
doing real things.

Patrick Turner.



Ian