Grid leak bias, how does it work?

[nigelwright7557]

Seems not the best way to save on a 1 penny cathode resistor !

Using high ohmic values isn't always good.
One electronics company I worked for banned resistors over 1 meg-ohm due to problems in high humidity countries. Apparently someone designed a telephone exchange with high value resistors and it worked fine. When it reached its final destination (somewhere much hotter) it failed to work. It was all down to the resistors. The business had to pay to install air conditioning !

 

[DF96]

The higher distortion that you got for this mode puzzled me since I previously had seen measurements showing the opposite under certain conditions. (That was w a y back for me; I no longer have those figures.) From memory, at the time, the author wrote that working in optimal class A all the time, i.e. as close to grid-drawing-current as is healthy, was a better condition than with a small signal deep into negative bias for g1 - something like that.

I have a paper from the 1950s which shows a reduction in second-order distortion due to cancellation between grid current distortion and anode curve distortion. Back then they didn't realise (or didn't care) that this type of bias can lead to larger amounts of higher order distortion; THD goes down but it sounds worse.

 

[Wavebourn]

What distortions are you talking about?!
The result of a grid leak bias for the same anode current and voltage would be THE SAME BIAS VOLTAGE as obtained from cahtode resistor bias. The only difference is, obtained from grid current rather than from cathode current. The same voltage, the same distortions, no difference. However, feedback by current provided by cathode resistor decreases distortions, but it is a completely different case, unrelated to bias voltage.

"Cancellation of distortions" by making them more symmetric compensates even order distortions increasing percentage of odd order distortions. The question is, do we need such reduction of distortions, or not. If we want to fool measurtement tools "decreasing THD", we want that. If we want to fool imagination decreasing audibility of distortions, we don't.

 

[Johan Potgieter]

What distortions are you talking about?!
If we want to fool measurtement tools "decreasing THD", we want that. If we want to fool imagination decreasing audibility of distortions, we don't.

! ! ! Exactly!


Still reading up here - you and your interesting points!

Just recalled somewhere during said search: The data tables I was thinking of earlier showed an increase in distortion over conventional (cathode bias) use when the grid is fed from a high impedance source; decrease over conventional when fed from a low impedance source. And sadly as I recall, figures were for t.h.d. Still ....

The above also follows from some previous posts here. [My own use of grid leak bias: Feeding impedance always low (< a few K.ohm)].

Still canvassing internet for more specific data. Re previous data about ECF80 pentode grid leak bias: I use 20M with some trepidation. I might perhaps use 10M and restore design anode current by increasing the screen resistor. As said demands for shielding around pin 2 from whatever else roamed about the chassis became severe despite using an alu. chassis. Even an external steel tube shield helped somewhat. (I did determine that the interference was of magnetic nature.)

Apology; back to topic!

 

[Wavebourn]

When I used 4M7 grid bias with 6J32P, I had to solder a wire to 40U-9 47 NF input cap to shield it.

 

[DF96]

What distortions are you talking about?!
The result of a grid leak bias for the same anode current and voltage would be THE SAME BIAS VOLTAGE as obtained from cahtode resistor bias. The only difference is, obtained from grid current rather than from cathode current. The same voltage, the same distortions, no difference. However, feedback by current provided by cathode resistor decreases distortions, but it is a completely different case, unrelated to bias voltage.

True, but in many cases grid leak bias uses a smaller grid voltage and smaller anode voltage than would be typical for cathode bias. Hence the anode response could be more nonlinear, and the compensating distortion created by grid current from a high impedance source was regarded as a good thing. The paper I have somewhere actually showed graphs of higher order distortions rising under these conditions, but this was not regarded as a problem back then by most people.

I am not knocking grid leak bias. It is a useful tool, especially for low level signals from a low impedance source or where some distortion is not a problem.