Self biasing without a circuit board?

[primosounds]

I do not think so. For nearly twenty years, Tentlabs has been selling similar
auto-bias modules, which work on the same principle ...

Well, a cathode bias resistor after 60+ years is working perfectly fine. Even fixed bias circuits still work although they need to be rebuilt. If in 30 or 40 years the auto bias circuit is still working without repairs then that would be considered a bulletproof design. I have already fixed 2 auto bias circuits in modern amps so not really up to the same level of reliability. Perhaps different circuit with tent lab, can't say i worked on one of those.

 

[Retrovert]

Another technique—long fallen into disfavor so people have forgotten it—used in those old radios was "grid-leak bias" which relies upon a very large grid-leak resistor to implement a form of fixed bias, where contact electrons (cathode electrons which hit the grid) automatically creates the bias voltage as with fixed bias. This saves a bias resistor and bypass capacitor at the cathode.

Numerous problems, however, preclude its use: it isn't self adjusting, varies by the tube type and even the manufacturer, and doesn't have built-in safety factor so if the grid leak resistor drifts high the tube biases to fully on and melts down into slag.

 

[gadget73]

It also only works with extremely high impedance circuits. I have an amp that came with a 10M grid leak up front, no coupling caps though. It would only work if I added an input coupling cap. I fixed it with a reasonable value resistor on the grid and a cathode bias circuit to allow for sane bias and feedback.

 

[Retrovert]

It also only works with extremely high impedance circuits. I have an amp that came with a 10M grid leak up front, no coupling caps though. It would only work if I added an input coupling cap. I fixed it with a reasonable value resistor on the grid and a cathode bias circuit to allow for sane bias and feedback.

Sure, the rate of leakage must be low enough to ensure that the grid-leak resistor controls the bias point. But since input was being fed by the output of a tube with low current-drive, that tended to be the case. Remember,. those tubes were had low voltage gain and low current drive.

You thereby built the circuit the designer would have made had resistors not cost $5 apiece and capacitors $15 apiece.

 

[gadget73]

This was an EF86 pentode, so it had a whole lot of voltage gain. Between that and a lack of a complete feedback loop the amp was kind of useless as-built. I suspect the feedback tied into the preamp somehow since it had wire in the umbilical off the output trafo, but I didn't have that half. I just made it work with what I had, which ends up being a pentode driver, a triode split-phase inverter, and a pair of EL84 tubes at the back end. To make that into a properly useful amp I think it took adding 4 resistors and 2 caps per channel. I deleted one cap and one resistor from the stock config though. The only thing that sucked about it, this amp is built on a 1960s circuit board. It required some relatively minor trace hacking to use cathode bias, and just working with those old boards in general is not a picnic.

 

[Retrovert]

This was an EF86 pentode, so it had a whole lot of voltage gain. Between that and a lack of a complete feedback loop the amp was kind of useless as-built. I suspect the feedback tied into the preamp somehow since it had wire in the umbilical off the output trafo... I deleted one cap and one resistor from the stock config though.

Your description, particularly what I take to be the feedback, sounds like the Mullard 3-3.

See: http://www.r-type.org/articles/art-003h.htm

Look about right? I was just studying that circuit a few weeks ago trying to understand why anyone would use it in a modern amplifier. Well, modern being a relative term, of course,m but 1955 was certainly modern enough that grid-leak bias was unusual.

I believe that grid-leak bias was used in that design because the input was driven by a low-output crystal cartridge. A high-impedance input would consequently be required, and grid-leak bias was perhaps the easiest way to obtain that.

The pentode noise plus the Johnson-Nyquist noise of the large-value resistor would likely have been considerable.

The only thing that sucked about it, this amp is built on a 1960s circuit board. It required some relatively minor trace hacking to use cathode bias, and just working with those old boards in general is not a picnic.

Yeah, the formaldehyde phenolic was never stable with heat or age, brittle and fragile, and the traces peel off from the heat of a candle.

 

[gadget73]

Push-pull EL84 on this one. Circuit-wise its now actually very close to a Dyna SCA-35 or any number of stereo push-pull EL84 amps running a 7199, 6U8 or 6GH8 driver with the exception of being fixed bias instead of cathode bias. I initially bought it for the tubes and transformers, but decided to tinker before taking it apart. It ended up coming out well enough that I left it together. The transformers saturate under 40hz, but above that its very flat and the tubes are a well matched healthy set of Valvo.