Using LED bias for 4.5 mA cathode current, is it a good idea?

[rtsang]

I read from Morgan Jones' book that using LED bias for < 10mA may not be a good idea as the LED may not be linear.

My amp has a pentode driver, using EF37A running a plate current of 3.6mA and cathode current of around 4.5mA. Previously it was biased with a bypassed cathode resistor.

I have changed it to a green LED, with around 1.9V Vgk now.
I have changed the screen from a voltage dropping resistor to a 114V zener diode chain bypassed to ground with a 0.22u cap.

I found the amp has quite a bit of white noise.

Could it be due to the LED?

Raymond

[SY]

Could it be due to the LED?

Unlikely, but you never know.

Yes, LEDs are somewhat nonlinear at low currents, but as a practical matter, I couldn't measure the difference when I ran them at 2mA versus 10mA. YMMV, so if you want to confirm, run a resistor from B+ to the LED sized to pump another 6 or 7mA through it.

[wrenchone]

If you starve the high voltage zener, you could be injecting noise into the screen despite the bypass cap (which is a bit small). Starved zeners (low bias current) can be really noisy. Remember also that the screen will be drawing some current, so you might not have as much bias current through your zener as you thought...
There is nothing wrong with adding some extra bias current to the LED using a resistor or a current source of some sort. Thats what I'll be doing with my LED biased phono preamp, though I'm using an infrared rather than a green LED (1.2V vs. 2V drop).

As SY points out, some LEDs are better (or crappier) than others. Since no one but us weirdos cares how much noise they make, you could possibly get a noisy one that passed all the emissions criteria. I'd still look at that screen zener, though.

[rtsang]

Thanks SY ad wrench clone.

I was suspecting noises from the zeners but I am not sure how the noise in the screen would affect the output noise.

I will try injecting some more current to both the LED and the zeners.

Also I noticed that with some EF37's the noise is less, with some, it is more noisy.
I am running the tubes pretty hard, probably close to 0.9W. The max . dissipation for EF37A is 1W. Would running the tubes hard bring in more noises?

My other plan was to change to a 717A pentode as driver (both are octal, EF37A uses grid cap).
The 717A can dissipate 4.5W.

I try the first plan of injecting more current first and report back.

Raymond.

[rtsang]

As suggested by wrench clone, I increased the current to the zener by paralleling a 75k resistor to the 100k screen voltage dropping resistor (this was what I had in hand). Effective resistance now around 43k.

Calculated the current passing through the resistor to be about 5.7mA
The screen current should be about 0.9mA, so about 4.8mA passing through the zener chain.
Previously, there was around 1.56mA passing through the zeners. My zeners are rated 2W.

I did not have a oscilloscope in hand, but by listening through the speakers, a 95dB/W/m sensitivity one, the noise is much decreased.

Next up to experiment is to increase the current passing through the LED. But right now, listen to some music first.

Thank you once again to SY and Wrench Clone for the rescue.

Raymond.

[kenpeter]

Was there a previous thread where a schematic was posted?
I'm not sure what feat of mind reading let us know there was
a Zenier to be found on his screen?

I understand the process of forward bias, how the depletion
region is flooded with carriers. I don't know how the reverse
situation works, in avalanche???

[rtsang]

Hi Kenpeter,

here is the schematic.
Mainly a WE91 style using British 4V tubes.

Raymond

[rtsang]

This is the previous version, where the driver is running in a lower current.

Also you may notice the PX4 in the new version is running a bit over spec,16.5W plate dissipation. Just try to imitate tubelab.com
The new KR PX4 can tolerate up to 20W.

To use PX25,I just change the cathode resistor to 500R (paralleling 2 1K resistors with a switch).

I changed the driver operating point because I worried that the driver will be current limited if I use it to drive a PX4.

Raymond

[rdf]

Originally posted by rtsang
Hi Kenpeter,

here is the schematic.
Mainly a WE91 style using British 4V tubes.

Raymond

Have you tried returning the zener string through the green LED? Might be all the current required.

[rtsang]

Dear rdf,

Yes, that was what I initally did. The zener string and the bypass caps all through the LED.
Initial impression was the noise was less. But I did not listen to it seriously.
Obviously the screen voltage lifted by 2 Vs.

Later, my worry was the the effect of changing current in the LED would affect the screen via the zeners. But the LED and zeners should clamp the voltage tight and the changing cathode current should not affect the operation, if the current can push the LED to work in the linear zone.

Am my reasoning correct. I'll change it to your suggestion and post the results.

Raymond.