Bullet Proofing The EFB Cathode Bias Regulator -- Including November 2025 Update

[dcgillespie]

That will solve it. The four signal diodes specified in the original update did a perfect job of protecting the regulator against the conditions identified that were damaging some versions of the regulator, but also brought with them the issue of thermal sensitivity. Changing them out for the single Zener diode recommended will still provide all the protections that the four signal diodes did, but without their sensitivity to changing environmental conditions that was impacting the operation of the regulator.

Dave

 

[spantou01]

Dave, I have the following EFB implementation Knight KN735 amplifier, minus the per-channel bias adjustment pots. I also see cathode current creeping with heat. Would the same zener mode apply here? Replace the 4 diodes with your specified zener ( cathode banded end towards ADJ terminal)?
Thanks

View attachment 3659615

Thanks for posting this. I’m intrigued by your individual tube grid adjustments. Does it work well to adjust individual EL84 bias currents and have enough range to use quads that are not perfectly matched? I assume you have a sensing resistor in each EL84 circuit to measure the individual currents.

I might try to rig that up with the @dcgillespie EFT/PS boards for my ST35 and SCA35. These boards are so convenient, but I miss the ability to do fine adjustment. I find it very difficult to obtain quads of EL84s that are (and stay) perfectly matched.

I hope Dave will consider adding these features to next generation EFB boards for ST35 and SCA35.

 

[grindfix]

As drawn, it's not per tube but rather per channel pair. Another potentiometer can be installed to balance between tubes within the channel.
I have over 100 soviet 6P14P tubes, so matching is not a problem for me, and I never installed those potentiometers. Grid resistors go to ground.

 

[wparks]

These boards are so convenient, but I miss the ability to do fine adjustment. I find it very difficult to obtain quads of EL84s that are (and stay) perfectly matched.

I built my 6AQ5 based push-pull Mirror Mite modeled closely on Dave G's schematic for the upgraded Magnavox 8800 (below), which incorporates both bias per channel and DC balance adjustment between the two outputs of each channel. These two controls allow a complete balancing of the bias current for all tubes. The balance adjustment range is fairly limited, as it was intended to match two already fairly close tubes, but I can attest to how nice this setup works. The intent here was to minimize the bias current imbalance within each output transformer primary, to minimize the baseline magnetic saturation that would steal bass response, particularly in the small transformers on the 8800. I'm actually in the process of doing experiments on my Mirror Mite right now, which also uses small transformers, to characterize low end response between perfectly balanced, and very imbalanced tubes to discover how much impact this has.

I would be very interested to hear Dave G's thoughts on how current balance could be incorporated into an EFB implementation. I suspect the resistor from LM337 OUT to GND(IN) could simply be broken out into separate resistors/current paths with pots to derive the small positive grid voltages for each tube from the regulated bias voltage.

But then I'm also curious how in a non-EFB design like the 8800, how the changing channel bias current in class B operation and it's effect on the grid voltages- would that interaction tend to increase or reduce distortion? If a current balance were to be implemented with EFB, would the variable grid voltage benefit from being regulated, or variable with the changing B+ voltage as would happen if derived from the screen voltage.

 

[gadget73]

There was a discussion about EFB in the 8800 thread, I do not remember the fine points but I want to say it was determined to not be beneficial because it basically ran class A and the power supply voltages didn't shift by much so there wasn't really anything for the regulator to do.

There is a per-channel EFB bias adjust setup in the original post about EFB on the SCA35. It tickles in some positive voltage in a similar manner to how assorted amps that ran a string of tube heaters for a cathode resistor made it work.

New Modifications for the Dynaco SCA35 and ST-35

thats per-channel but you could expand that to 4 pots for per-tube easily enough. R35,R36 are the grid to ground resistors at the output tubes. The 5K Bias Range pot would move the current on all 4 tubes at once, the other pots move it per-tube or per-channel depending on your config.

I don't think this should affect how the regulator does its work either. The pots will shift the grid to ground voltage but since its derived from the same voltage source as the regulator the grids should move by the same percentage as the screens and the cathode to keep it all tracking relative to each other.

 

[Tube Radio]

So if one were to go per tube bias, the 2k pots would need to be 5k given there's four pots, right?

 

[gadget73]

If you want to maintain the same adjustment range, yes.