Local NFB

[Tobacco Soup]

Hi folks!

IIRC, (in a triode, preamp is only of interest here for now) the plate voltage is out of phase with the grid signal and by placing a high value resistor from plate to grid we get some local NFB.

IIRC, this local NFB can also be achieved simply by using a small value capacitor from the plate to grid. Now, does this capacitor simply parallel the grid-to-plate internal resistance forming a filter network?

What are some of the 'tried and true' values for local NFB in a preamp? Can anyone offer a simplistic explanation of the theory involved with this technique? I would be very grateful for any contribution concerning this. Regards,
TS

[Don Moose]

Well, you need a DC blocking cap AND a high-value resistor to do the local NFB right. See the schematic for an old tube hi-fi preamp (suggestions?) for some examples.

Leaving out the DC blocking cap will yank the grid up to Vplate jamming the triode full ON.

Leaving out the resistor will give you 100% NFB, and no gain.

How to calculate the values? See if you can find a Radiotron Designer's Handbook (RDH) on line - I know there are a few, just don't recall where.

Hope this helps!

[Old Tele man]

two versions of RDH are posted over at Pete Millett's Technical Books website at URL: http://www.pmillett.com/technical_books_online.htm

[Tobacco Soup]

Thanks folks!

I'll digest that for spell.

[Loogie]

You can also find information here:

http://www.aikenamps.com/

Tech Info/Advanced/Designing single-stage inverting feedback amplifiers

And here:

http://www.freewebs.com/valvewizard/localfeedback.html

[isaac]

Try 22M resistor in series with a .047 uf cap. When I did it on a bandmaster preamp, I had it switchable, and the only problem i had is that without the cap in the circuit, it had a pop when I hit the switch, I Don't recall hearing much difference with or without it, though. If the sound is too much for you, use a 44M for a more subtle effect.

[jazbo8]

Adding the plate-to-grid capacitor, increases the Miller capacitance of the the input stage as the capacitance is added to the Cgp of the tube, together with the input grid resistor and Cgk, they form a LPF as shown on the image below, you can plug in some values to the equation and get an idea of the corner frequency. Note that the plate-to-grid capacitance and Cgp are multiplied by the gain of the stage, so you probably don't want to use a value larger than 10pF, unless you are using a pentode, then the value could be higher due to its inherently low Cgp. Then again, there are other ways to accomplish the same thing, so increasing the Miller capacitance is not often used...



The plate-to-grid resistor case is already well covered on Merlin's site.

Jaz