Cheap Red LED bias on ecc83/12ax7

[soulmerchant]

Dear All

I have used LED cathode bias on a number of valves, but only recently tried it on some ecc83. I am pleasantly surprised with the improvement.

The problem with LED bias on ecc83 is that the current draw of this valve is too meager to induce any standard LED to conduct. I therefore bring the LED to continuously conduct by an additional connection, bypassing the valve (and annode resistor) entirely with a separate sufficient wattage higher value resistor directly to the HT.

The value of this resistor is chosen to drop almost all the HT and deliver approx. 6mA current to the cheap red LED. For my test circuit I am using HT of 280 Volts so chose a 5 watt 47K ohm resistor that I had handy. I pre-selected the red LED's and chose two that elevate the cathode by 1.67 volts on each channel.

My method of testing for linearity is a crude point-based method. Nevertheless, I measure distinct improvments that I believe can be heard with my own ears...

The by-passed LED bias on e83cc also seems to remove any last vestiges of power supply noise. Perhaps there is a bit of aikido-stype ps noise injection in this method. Since I by-pass the LED to make it continuously conduct, the effect is electrically similar to filament bias.

When I have used LED bias on other valves the change is distinct and subtle, but the changes I am getting with this method on e83cc seem even more significant. Maybe it is just my own biased hearing though. Has anyone else done this before?

Ian

[jane]

Originally Posted by soulmerchant

The by-passed LED bias on e83cc also seems to remove any last vestiges of power supply noise. Perhaps there is a bit of aikido ps noise injection in this method.

The dynamic resistance of a red LED is ~5 ohm and the resistor connected to the PS is 47000 ohm. This gives an injection ratio of ~5/47000 or 0.01%.
The PSRR of a commond catode stage is rp/(rp+Ra). Devide that ratio by the gain of the stage and you'll find what to inject to cancel out PS rippel..

[yagoolar]

Done exactly what? Do you mean injecting more current via additional resistor in series with LED? It must be done to set operating point for an LED dynamic resistance to be (more) linear and low(er).

[SY]

Yes, I've posted on this several times and the resistor "trick" was added to the components chapter of "Valve Amplifiers" 4th edition. It's a great way to run ECC83, a tube which is rarely used correctly. If you use a CCS on the plate, then buffer the output, the linearity is stunningly good.

[jane]

Looking at your schematic: the injection point is in phase with the ripple. No ripple cancellation in other words..

[yagoolar]

Originally Posted by jane

Looking at your schematic: the injection point is in phase with the ripple. No ripple cancellation in other words..

Voltage ripple is also at the plate, so you in general subtract those injected at the cathode. Yet, you should calculate if it is significant with the values from the schamatic.
See: Lowering SE Amplifier Noise (page 4)

[soulmerchant]

Thanks Sy, I always buffer the output on ecc83 - the output impedance for the above circuit is otherwise far too high... I only have the 3rd edition of MJ's book - my wife finally gave it to me for christmas 2011 so can't really justify buying the 4th.

For some of my testing I also used an annode choke instead of a resistor - which sounds fine, but the effect was at all so startling as that of the LED. I only wish I had tried this earlier.

Thanks for your input jane - it must just be my ears then. I have to push the circuit quite hard to get any audible ps noise anyway. Strange things is also this... it does seem to have greater amplification factor as well (at least to my ears). But I haven't yet bothered to try and measure this yet...

What I find is that this way of bias appears much better than standard LED bias - although it might only be possible on certain valves.

[the_manta]

I currently have a 12Ж1Л Pentode running at about 1mA with a CCS and LED bias. Cathode LED and Anode LEDs share the same resistor. See attachement. works well, of course with feedback due to gigantic amplification.

[kevinkr]

Originally Posted by soulmerchant

Thanks Sy, I always buffer the output on ecc83 - the output impedance for the above circuit is otherwise far too high... I only have the 3rd edition of MJ's book - my wife finally gave it to me for christmas 2011 so can't really justify buying the 4th.

<snip>

I have the first, third and fourth editions, there is sufficient new material in the fourth edition that I felt justified in getting a copy. (Well actually my wife ordered me a copy for last Christmas, which I finally got a couple of months ago.)

I use and like LED bias, but these days am also using battery powered grid bias. Lithium alkaline (1.8 - 1.9V new) and alkaline (1.6V new typically) have proven quite worthwhile in my recent designs. I've also used A23 12V types in certain applications. (See my Muscovite thread in analog sources for an example)

[DF96]

Originally Posted by yagoolar

Voltage ripple is also at the plate, so you in general subtract those injected at the cathode. Yet, you should calculate if it is significant with the values from the schamatic.
See: Lowering SE Amplifier Noise (page 4)

That only works for transformer coupling. For RC coupling the ripples add, as 'jane' said.