grid LED bias?

[fred76]

Hi,

Has anyone here tried grid LED bias for input stages? There would be very little if any current passing through the LED in this config. But I still wonder if anyone has tried it and want to share their findings. Thanks.

fred

[ErikdeBest]

Hi Fred

I don't have any experience, but how you want to achieve this? The grid must be negative to the cathode (that's why commonly the cathode is made positive, and the grid is grounded). I think you want to ground the cathode, and make the grid negative using the led...but how? With an additional power supply, maybe? Can you post a rough schematic?

Many thanks

Erik

[Brian Beck]

The only way that I can see using an LED for grid bias would be to use it as a shunt reference for a low voltage negative supply that then feeds the AC-coupled grid via a large resistor. If you had a -6.3V or -12.6V supply already, you’d connect a current limiting resistor (or even a CCS) to the negative end of an LED (or two…). Set it for 10mA or so. The positive end of the LED would be grounded. As others have reported, a single “cheap red LED” would give about -1.7 volts. For the majority of small triodes, you’d then connect a 1 Meg resistor to the grid from the LED. You could use two LEDs in series for -3.4 volts, and put a 10K pot across them both to tweak any bias voltage from 0 volts (careful!) to -3.4 volts. I don’t know how noisy LEDs might be in this application, but with a 5-ohm effective impedance, it can’t be too bad. PS noise rejection would be excellent with a CCS feeding the LED, so-so with just a resistor feeding the LED.

[rdf]

I plan on trying grid LED biasing myself for an EL84 output stage. At this point the model is a string of diodes with a 1000 ohm pot across the bottom 3-4 LEDs to provide just the required range, all fed by a solid state CCS. The arrangement has one big advantage over cathode placement, the variation in current through the LEDs, because the grid resistor is so large, is orders of magnitude less. Any voltage modulation from the device's non-linear impedance would be reduced as much as a consequence. The one disadvantage is the grid reference is now regulated and changes in the utility voltage have a bigger impact on operating point.

[Brian Beck]

That arrangement should work well, and as you say, there will be less modulation of the LED resistance by audio currents with grid bias. In the unregulated cases like your EL84 output stage, fixed LED bias will not prevent plate current changes due to plate voltage variations in either case: if the LED is in the cathode circuit or in the grid circuit. The amount of DC feedback from the small effective resistance of the LEDs (series plus parallel) will provide little DC NFB. Constant Vgk is constant Vgk. This is the disadvantage of "regulated" voltage biasing versus the traditional bypassed cathode resistor or even bypassed CCS. But a hard grounded cathode does have a lot going for it!

[rdf]

Originally posted by Brian Beck
But a hard grounded cathode does have a lot going for it!

That's my experience as well. Another advantage, somewhat at odds of that, of LED grid biasing (or of any fixed bias topology) is the option of OPT secondary feedback to the output tube. Having heard it I'm not sure I can live with an EL84SE without.

[kevinkr]

I am a firm proponent of fixed bias output stages, but regulating the grid voltage without regulating the plate supply in triode mode output stages can result in a significant variation in the operating point as a function of supply voltage. In the case of pentode operation regulating the screen supply would remove this dependence.

Leds make great little voltage references up to a few volts, in the case of the higher voltages required for a 6BQ5 I think a 12V - 15V zener would probably be a better choice. Adequate filtering can be applied in either case, and I'm not sure in any event that the average LED string is quieter or has a better tc than a zener. Very good psrr can be achieved as previously suggested with a ccs..

Also it is quite simple to implement cathode feedback even with cathode bias so the choice of one over the other should not hinge on whether or not cathode feedback will be employed.

Still I state my preference for fixed bias in output stages from both an efficiency standpoint (no energy wasted as heat in a cathode resistor and a lower effective B+ at the required current = some energy saved) and as always no capacitor is the best capacitor when a large electrolytic or film cap would be required.

[Giaime]

Hello Kevin,

what's the best way to implement cathode feedback in a fixed bias arrangment? I mean, with cathode bias, you could connect the negative end of the bypass cap to the OPT secondary, with the cathode grounded how could I do?

Let's assume I don't have a dedicated CFB winding, so I only have the OPT secondaries to have CFB, and I wouldn't have DC on my speakers.

[rdf]

Originally posted by kevinkr
Also it is quite simple to implement cathode feedback even with cathode bias so the choice of one over the other should not hinge on whether or not cathode feedback will be employed.

True but damping factor tests I've done on cathode-biased output stages with capacitor-coupled cathode feedback suggests it takes a very large value to maintain uniform DF to low frequencies. Then as you say we're back into cap issues, etc.. One of many design trade-offs in any circuit design I guess, for me fixed is far and away the better solution for cathode feedback.

Re: the power fluctuation issues, I'm fortunate to have stable local power and normally set quiescent power to around 80% the tube's nominal rating as a safety factor and for long life. Commercial vendors don't have that luxury, nor some of the DIYer's who've posted examples of their local utility's less 'stringent' standards. Most of my front end stages are CCS plate-loaded so it's not a factor there. Thanks for head's up on potential noise issues of LED strings vs. single zeners.

[Brian Beck]

Perhaps SY will chime in here. His Red Light District amp has its own thread. I believe he used a 7x7 array of cheap red LEDs to bias the cathodes of a pair of EL84s, thereby getting an equivalent dynamic resistance of about 5 ohms. Without looking at data sheets right now, I believe that 5 ohms might be a lower resistance than a typical 12 to 15 volt zener. And I'm guessing that noise will be lower as well. If you use zeners, and if it is possible to achieve the correct voltage in increments of 6.2 volts, it would be better to use two or three stacked 6.2 volt zeners since at that voltage tempco is low, as is noise, due to the transition from true zener to avalanche modes.