How to size grid leaks and grid stoppers?

[Andy Evans]

I was just ruminating over the number of different values you see for grid
stoppers - anything from 330 ohms to 10K - and grid leaks. I'd like to size the
resistors on the front end of my Leak Stereo 20, presently with 6EW7 gain>DC
to>Concertina. I have a 20K stepped attenuator into the amp. Examples of
existing circuits (leak + stopper):
Morgan Jones: 1 meg + 330 ohms (into ECC88)
Van der Veen: 100k + 1k (into ECC82)

These are so different, how do you calculate these things?
=== Andy Evans ===
Visit our Website:- http://www.artsandmedia.com
Audio, music and health pages and interesting links.

[Yves Monmagnon]

Hello !
In a plate loaded triode, impedance from grid to ground affects the hi
frequency response, by virtue of Miller effect.
Consider the grid to plate internal capacitance (Cgp) wich reinject a
fraction of the output signal back to the grid.
It is nothing else that a local negative feedback whose effect increase with
frequency.
The lowest the grid leak resistor, the lowest the feedback, the wider the
bandwith.
The highest tube Rp and plate load value, the highest the feedback, the
lower frequency cut off !
So, stoppers are a way to reduce hi frequency gain, hence reducing risks of
parastics RF oscillations.
Value is heavily layout dependent.
Start with no stopper and check with a CRO for ultrasonic or RF parasitics.
Beware, sometimes the CRO probe itself may start or stop oscillations !!
Tubes with lo Rp and lo Cgp usually needs lower value stoppers.
To maintain wide bandwith, use the lowest possible leak resistor value
compatible with the output
impedance of the previus stage (or more generally, the impedance of the
signal source).
A 6DJ8 well designed stage always exhibits better hi frequency response than
a well designed 12AX7 one
because his lower Rp and Cgp.
At least 100Khz -3dB point is desirable, but this depend of the number of
stages and the amount of global
negative feedback.
I can't calculate ;>) Cut and try !

Cheers, Yves.

"Andy Evans" <aeatarts...@aol.comnohawker> wrote in message
news:20030703120418...@mb-m01.aol.com...

[Tom Schlangen]

Hi Yves,

> Start with no stopper and check with a CRO
> for ultrasonic or RF parasitics.

> Cut and try !

;-)

Regarding grid stoppers, I'd like to add the usual
recommendation: One should solder them as near as
possible to the socket pin.

Tom

--
Do not meddle in the affairs of wizards,
for you are crunchy and good with ketchup.

[TubeGarden]

>One should solder them as near as
>possible to the socket pin.
>

Hi RATs!

I just wire wrap one of the resistor leads on the tube pin, with the body of
the resistor at the glass and the lead leading down the pin.

Happy Ears!
Al

PS 100 to 200 ohms on VA, more on PT.
Alan J. Marcy
Phoenix, AZ

PWC/mystic/Earhead

[Tom Schlangen]

Hi Alan,

regarding grid stoppers:

>>One should solder them as near as
>>possible to the socket pin.

> I just wire wrap one of the resistor
> leads on the tube pin, [...]

Well, since you omit sockets, your positioning
of the grid stopper could hardly be closer
to the tube system except for smashing the
envelope and soldering the stopper directly
to the grid ;-)

> PS 100 to 200 ohms on VA, more on PT.

I don't have much experience, but a 220 ohms
resistor at the grid of the triode section (gain/driver
stage) of my ECL82 amp surely stopped all oscillation
whatsoever.

I added another 220 ohms resistor at the g1 of the
pentode section of ECL82 just for safety, but it
wasn't really necessary. Calculations show, that
it doesn't do no harm on (audible) bandwith either.

I am currently building (breadboard stage) two amps,
one 6SL7/EL34 SET, and one with 4 PCL86 PP/UL per
channel. The 6SL7/EL34 thing seems to be rock-stable
with no grid stoppers at all even with a breadboard
rats nest, while the multi-PCL86 thing seems to be
born to serve as a 20-30 watts longwave RF generator ;-)

No, not really: A microscopic little piece of earthed
iron sheet between pins 1-2-3 (triode section) and
pins 6-7-8-9 (pentode section) at all PCL86s stopped
all RF oscillation - without using any grid stoppers.

Tom

--
Life: All in all a bad game, but graphics are really good.

[Ian Iveson]

Using the usual filter equations. Take into account capacitance of
cable, Miller, coupling caps, output resistance of pre-amp, etc.

If you simulate you can save time on the maths.

Morgan has an excellent and detailed explanation of how to choose
grid leak and therefore input resistance. Also of course valves
have minimum values of grid leak stated in specs.

The difference there between van der Veen and Jones is possibly
because van der veen expects input from an SS source with its own
level control. Using a "passive preamp" with a 100k pot, I need
250k at least, 1M preferred, input resistance at the power amps.
There seems to be a tradition of a minimum 10:1 impedance ratio from
stage to stage throughout.

Van der veen has a good detailed explanation of how he uses the
power stage grid stoppers for removing a HF peak in an amp just like
yours.

I wonder if you repeat your toob rolling, altering resistors to give
same frequency response and optimum bias point for each stage and
each valve, whether the differences in sound will be so apparent?

cheers, Ian

"Andy Evans" <aeatarts...@aol.comnohawker> wrote in message
news:20030703120418...@mb-m01.aol.com...

[John Stewart]

Andy Evans wrote:

Grid stopper resistors lower the Q
(Quality Factor) of the circuit according to the formula

Q = (1/R) * SQR ( L/C )

As little as 100 ohms in a conductor interferes with the RF
to the extent that it is eliminated. The audio is left intact.
Helps to prevent hi-G tubes from oscillating & most of the time
keeps your friendly local radio station out of your amp front end!!!

That is the real reason why grid stoppers work.
For G1 anything from one to 10 K.
For G2 anything from ten to 100 ohms.
Sometimes 10 ohms in the plate circuit,
especially in an amp with parallel output tubes.
All 1/2 watt, non-inductive.

None of these have much effect on the HF response of the amp.
Take an example as follows-

Input to the grid of a 12AX7 whose gain is 50X.
Miller capacity would be 50*1.8 pf..........90 pf
Add 10 pf for strays, so the total is 100 pf.
Now connect that to the signal source thru a 50K resistor.
If you do the arithmetic you will see that it is down 3db at 31.8 KHz.
So you see, 10K will not do much to the audio.
Crowhurst used this to provide an HF corner in some of his designs.

Cheers, John Stewart