The small signal pentode / settings?!.

[ChopSauce]

A common mod for Vox amps with an EF86 as preamp tube is to alter the value of the screen grid resistor Rg2

pentode1.jpg

(picture taken from http://www.valvewizard.co.uk/pentode.html )

based on pure "copy and paste from other models" considerations...

So I wanted to do the maths, to understand what happens when you change the Rg2 value, according to Merlin's book.

Merlin however states that "As a rule of thumb, screen current is a fixed ratio (see datasheet) of the anode current. " and that Rg2 should be chosed accordingly, so my question is:

- what happens when Rg2 departs much from the value corresponding to the above statement...

(With apologies if I missed the obvious)

EDIT: changed title according to current development

[roberto]

Yes, for the EF86 the ratio is generally Ia = 4 Ig2.
So if you have Ik = 1 mA, Ia will be 0,8 mA and Ig2 will be 0,2.

So you can drop the voltage seen by the screen by implementing the simple V = I x R formula, and adapt it to the curves you want to fit.

But this is true up to when secondary emission takes place, then thew ratio is no more constant and Ig2 increases, so the voltage on the screen drops significantly and the screen resistor acts as a current limiter.
Look at this plot taken from http://www.tube-and-valve-electronics.c ... n_Tube.asp , even if it's for other tube:

[roberto]

Valvewizard says the ratio between Ia and Ig2 is 5 instead of 4, but you can test both and see.
If you want to see how loadlines change by changing the Vg2, you can check this web app:
https://www.vtadiy.com/loadline-calcula ... alculator/

- select the EF86 from the list;
- keep it in pentode, Single Ended, Resistive Load;
- change the Screen Voltage from 100 V to 150V and see how the loadlines go up (extended);
- change the Screen Voltage from 150 V to 50V and see how the loadlines go down (compressed);

This effect is also the reason behind the VVRs designed to supply power amp screens: you get less amplification from the same input signal.

[ChopSauce]

(With apologies if I missed the obvious)

Thats' is:

- it doesn't change the current, its changes the voltage - right?



Thanks for the links & the explanations...

[ChopSauce]

Well, I still feel much like a dummy but...

If I analyse the AC4 circuit as per the voltages indicated on the schematic

VoxAC4.jpg

Ik = 2.7 / 1500 = 1.8 mA

and Ig2 ~= Ik/6 = 0.3 mA.

Hence Ug2 voltage drop equals: 5600000 * 0.0003 = 5600 * 0.3 = 3360 V...

Does that mean that the EF86 is almost always abused in this amp?

At least, from the link to the online plotter linked above, it seems that the 220k resistor is a real overload for the plate of the EF86...

[roberto]

Ik = 2.7 / 1500 = 1.8 mA

No. On that cathode you have also the current of V3's CF through R9's 100k.
Do the calculation based on Ia: (260 - 215) V / 220 kOhm = 0,2 mA

Ig2 ~= Ik/6 = 0.3 mA.

Using Valvewizard's values for Ig2: 0,2 / 5 = 40 uA

Hence Ug2 voltage drop equals: 5600000 * 0.0003 = 5600 * 0.3 = 3360 V...

5,6 MOhm x 40 uA = 224 V

So Vg2 is 260 V - 224 V = 36 V.

[roberto]

I've found a pdf with EF86's Ia and Ig2 vs Va curves here:
https://guitar-gear.ru/forum/index.php? ... h_id=15957

it is in russian, but images are images.

[martin manning]

Looks like a Russian version of Blencowe. It doesn't match exactly, but the images look like they are copies.

[ChopSauce]

I'll have a look at Merlin's book again, yes.

Ik = 2.7 / 1500 = 1.8 mA

No. On that cathode you have also the current of V3's CF through R9's 100k...

... of course!!!

Thanks again...

[ChopSauce]

So, if I summarize well:

- the 220k plate resistor (Rp, or Ra on the above fig.) somewhat overwhelmes the EF86 - settled for a gain of order 187 [Blencowe, p.49] ;

- the 5M6 grid resistor (Rg2) strongly limits the current - and the real gain should be limited beyond the above estimation ;

right?

If that is, why not rather settle the EF86 to something more balanced, for example (Rp, Rk, Rg2) = (47k, 680R, 330k) [Blencowe, p.54] than what's used on the AC4 circuit: (Rp, Rk, Rg2) = (220k, 2k2, 5M6)...

That would somewhat be consistent with the informations below, yet - one more time - the suggested (Rp, Rk) pair does not match much of Merlin's book...

(from https://ampgarage.com/forum/viewtopic.php?t=1132)
Here's an Ampage thread I saved. It wavers on and off topic.

[...]

The 'best' circuit I came up with (kg provided most of the details for this, I still have the post if anyone is interested) was this: 100k plate, 2k2/22uF cathode, the screen was connected to a voltage divider of 270k to B+ and 820K to ground and the .1uF screen cap was connected to the cathode (not ground).
(Matt)

[...]

a few hints:

a) use a fixed voltage divider from b+ to gnd to generate Vg2. i never liked the datasheet method of a simple series resistor from b+ and a cap to ground... any variation in tube characteristics and the Vg2 is all over the place, plus it tends to wander around as you drive it hard. a lower impedance divider will give you a bit less sag, and a bit less compression as the stages are
driven hard. a higher impedance will give a slightly spongier sound. bypass the divider with a cap. a smaller cap will give you a quicker recovery from the compression/sponginess, a larger will give a slower recovery. if you use cathode bias then tie the
bypass cap to the CATHODE, not to ground. this will help with preserving the transients.
(ken)

[...]

Has anyone ever tried any such things?