The idea of using an LED in a cathode is to maintain bias voltage in no dependance to anode current. See LED characteristics. LED is not a cure for heavy loads on B+. Well designed HV power supply is. What you can do is to calculate what is ΔVbias in regard to expected ΔB+. E.g.
B=+302V
Vbias = +2V
Ibias=10mA (one can take anode current change under the load onto account, i.e. ΔB influence on anode current)
ΔB +/- 5% or 30,2V (that is heavy I guess)
Rbias=(302V-2V)/10mA=30kΩ
Now,
30,2V/30kΩ=~1mA or 0,5mA around LED op point. Assume dynamic resistance of a LED is 5Ω
ΔVbias=1mA * 5Ω = 5mV
It is early in PL, so I may have made a mistake somwhere.
B=+302V
Vbias = +2V
Ibias=10mA (one can take anode current change under the load onto account, i.e. ΔB influence on anode current)
ΔB +/- 5% or 30,2V (that is heavy I guess)
Rbias=(302V-2V)/10mA=30kΩ
Now,
30,2V/30kΩ=~1mA or 0,5mA around LED op point. Assume dynamic resistance of a LED is 5Ω
ΔVbias=1mA * 5Ω = 5mV
It is early in PL, so I may have made a mistake somwhere.
Yagoolar, you didn't understand his question. He wants to avoid the extra 6mA current draw necessary to put more current through the LED than the tube is passing.
I was referring to post #19. The question was to use LV PSU and lower value resistor to feed LED. It is not THAT early 😀
My intention was to ask question "what for" if that works well.
My intention was to ask question "what for" if that works well.
Im curious, has anyone since found an LED for 12AX7 LED bias that works well enought at <1ma, so the B+ bias resistor can be avoided?
There are loads of low power LED's that work fine at around 1mA. Just check e.g. mouser. No idea how linear they are though, but they same pretty linear at the small amount of current variation the ECC83 exhibits.
If you have a CCS load in the plate circuit, that pretty much removes the nonlinear Z vs I characteristic of the LED. If not, the B+ resistor is still an easy option.
I have used low current red LED's for bias operation of ECC83's without having to use the resistor trick for added current.
Its funny, I just last night read a post of yours from a few years ago pointing this out. But I suspect I cant use a CCS plate load in this circuit. Its two sections stacked, bottom tube a mu follower. I was going to try it at some point.
Thanks for these responses, but I was kinda hoping someone would refer to specific part number. I have tried a number of Radio Shack LEDs and so far none of them sound quite right, though the current draw has been reasonably close to the resistor/cap combo I started with. I was hoping somebody made an led that had a linear Z vs I characteristic down to .5ma.
I have a drawer full of LED's but I think these are the ones I found to work good.
Invalid Request
Digikey part #1080-1067-ND
Invalid Request
Digikey part #1080-1067-ND
In my MM phono (6072, passive RIAA, 6072, ECC82 CF) a few years I use old 5mm green LEDs in the first and second stage.
Cathode current around 0.9mA.
Cathode current around 0.9mA.
Is there a good argument against using 2 or 3 Si diodes in series instead of the LED? 1N4148 x2 and 100k plate load looks better on even harmonics than LED.
Last edited:
The LED is a constant voltage device. It is unlikely any feedback voltage signal will develop across it.
Also, LED's introduce flicker noise into the non-inverting input for the vacuum tube whereas a good resistor would have very low to almost no flicker noise. Since the dynamic impedance of an LED is so low, it's nearly impossible to bypass the noise to ground also. A resistor, at least, would make it easier to bypass any noise to ground, since it would have a higher value than the dynamic impedance of an LED.
Last edited:
Think that statement through.
Funny that several of us have built incredibly quiet low level stages using LED bias and get no measurable noise contribution from them.
I didn't say the noise was high. I don't dispute your results. It remains a fact that LED's would be difficult to bypass (if you wanted to) because their dynamic impedance is so low that it would require a very large capacitor to do it to a low enough frequency to matter. I didn't say that you should bypass it. It is also a fact that LED's introduce flicker noise whereas a resistor may not, depending on the resistor type.
Please refrain from reading something into my statement that is not there.
Please refrain from reading something into my statement that is not there.
In general, it appears that an un-bypased LED has higher noise than a bypassed zener diode, but of course most LED's have lower forward voltage than most zener diodes, so the zener is not necessarily practical for anyone's particular application. I would also assume that a bypassed resistor has lower noise than an un-bypassed LED. Of course, you do not achieve constant voltage with a resistor. Also of note is the fact that a zener diode has far better linearity with respect to constant voltage with changes in current than does an LED.
Further, the OP's schematic can be improved by replacing the 47k resistor with a constant current device.
Further, the OP's schematic can be improved by replacing the 47k resistor with a constant current device.
Member
Joined 2009
Paid Member
I see a number of people reporting good results using LED cathode bias for 12AX7.
With fixed bias (i.e. LED) there is no self-correction of the idle current through the 12AX7. Swapping tubes or tube-ageing means small changes in bias current = large shifts in the plate voltage ??
With fixed bias (i.e. LED) there is no self-correction of the idle current through the 12AX7. Swapping tubes or tube-ageing means small changes in bias current = large shifts in the plate voltage ??
Sure, you just find the curve on the tube's I/V graph that has the grid voltage about equal to the LED voltage. Then plot the plate resistor's load line, and the intersection with that grid curve is the static plate voltage and current. (There's a slight error because the actual plate-cathode voltage is reduced by the LED voltage.
Last edited:
I think Bigun's point is that sample variation and age variation are partly self-compensated using resistor bias but not LED bias. The datasheet does not help, as it merely tells you what a typical valve will do not what your actual sample will do. Of course, in most cases a small bias shift does not matter too much - only simulation freaks worry about picking sweet spots where nth order distortion magically disappears.