• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Cathode Bias questions

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I'm new here, so hello there, I'm glad I found this great forum.

I'm hoping someone can help me with a question about cathode biasing.

I've seen three simple ways of biasing a grounded grid, plate-loaded triode stage. They're in this illustration...

cathode_bias_schemes.jpg


1) Of course the standard way is to put a suitable value resistor in the cathode circuilt so that the cathode is positive in regards to the grid. I showed the cathode resistor unbypassed, for simplicity.

2) Putting a rechargeable AAA or AA battery in the cathode can also put the tube's cathode 1.2V above the grid. I tried it and it worked. I'm not sure how much I liked the sound, but you know how that is...

3) I picked this up from a Van Alstine SuperPAS 3 I built from a kit, back in the '80s. A 1N4148 fast switching diode goes in the tube's cathode, although I never understood why the diode's anode would be sent to ground.

Anyone have experience with battery bias, or diode bias? Opinions?

I've got a phono preamp where the first two stages are diode-biased, and it seems to work well. Any horrible downsides to this that I don't know about? I wonder why I never see this...

Thanks for any info.
--
 
Look up LED bias here. An LED, other than emitting light, is just a diode. You'll find lots of examples. An LED or an ordinary diode has to be oriented as shown, or no current will flow. In this case the forward diode drop provides the bias voltage. For higher voltage differential, a zener could be used in the opposite orientation.

LED's are often preferred, because their change in resistance with current variation is small (low dynamic resistance - or impedance). Dynamic resistance will generate distortion. An LED, in it's linear range has a dynamic resistance of 10R or sometimes less. Putting an unbypassed resistor of, say, 100R in series with the diode will reduce the distortion in this case (10R impedance) by a factor of 10.

Sheldon
 
More than that, LEDs have instantaneous recovery from overload- none of the blocking that RC bias can have.

The single diode bias has one major problem- it's too low. Most tubes are starting to draw grid current with only 0.7V grid-to-cathode, reflected in increased distortion.
 
rdf said:
He reports interesting measured dynamic impedance numbers. Does 1.6 ohm sound reasonable?

Yes. But the real figure of merit is Z/Vf- one needs to series them to get a useful voltage value. With two in series, the forward voltage is about the same as a red LED and so is the impedance. How does their price compare to (say) one of the Agilent red LEDs?
 
dsavitsk: Maybe add this to your list as well.

As I see it, you're putting one leg of the heater to the cathode of the tube. So one leg would be lifted above ground by the voltage developed across the cathode bias resistor. With AC heaters, wouldn't that put 6.3VAC at the cathode? Or if with DC heaters, wouldn't you have the heater DC voltage at the cathode? I can't see how that makes sense. But you know, I'm not any kind of expert at this.

Is there some reason why an LED would be preferable to something like a 1N4148?

(Possible trolling question) - Any preference for LED color? Red? Green? Amber? Blue? Infrared? Other?

I've seen cathode bias as low as- 0.5V in the first stage of a phono preamp, using 12AX7 or 6DJ8. Are these drawing grid current with the 47kOhm load of an MM cart? What about 1kOhm for a high-output MC cart?
 
The advantage of the diode or LED over a resistor is that the stage will have higher gain. The resistor, assuming it is not bypassed with a capacitor, reduces the stage gain, but also provides feedback to reduce the distortion of the stage. With a small voltage swing in a preamplifer, the distortion reduction is likely to be insignificant (as is the gain reduction too). The diode or LED would be lower noise than a carbon composition resistor, but at line levels it too would be insignificant.

The reason to use one LED over another is for the voltage drop, with red typically being the lowest and blue the highest. You can accomplish the same effect by placing diodes in series. I would not use LEDs in low current stages (less than 5 mA).

The LED or diodes will set the cathode/grid voltage which will determine the operating point for the tube. You need to consult the average plate characteristic curves for the tube you wish to use to pick a plate voltage and current that correspond to the cathode/grid voltage you selected. For example, using the curves for a 12AX7, and three 1N4148 diode in series (for a 2 volt bias voltage) you have 0.5 mA bias at a 200 volt plate voltage and 1.5 mA at a 275 volt plate voltage. You choose your palte resistor based on your supply voltage. The higher the plate resistor, the higher the gain (but you have to consider the loading from the following stage if there is another stage).

If you use a 6SN7, the cathode/grid voltage tends to be 5 to 8 volts so you'll need a lot of diodes or LEDs. Unless you really need the higher gain, I think most times the resistor is the better choice.

Actually, I prefer current sourced long tailed pair stages to resistor or diode bias. The current source will handel the bias voltage for you automatically and I have found it gives MUCH less variation from tube to tube in operating point. I'm sure that statement is not supported universally, but it has been my experience.
 
No troll at all! Different color LEDs have different voltage drops and different internal impedances. So they each have their place. In Morgan Jones's "Valve Amplifiers," 3rd edition, there's some data for a variety of LEDs and diodes. It's somewhat dated (I await the 4th edition!), but a very useful guide to this circuit trick.

As to the distortion issue, you will indeed start seeing some not-good changes in the distortion spectrum as you drop the bias voltage. How much and at what point depends on the particular tube and the circuit operating points. Having a 0.5V bias is asking for trouble, IMO. However, schemes that increase distortion are often popular, for what reason I do not understand.
 
rongon said:
As I see it, you're putting one leg of the heater to the cathode of the tube. So one leg would be lifted above ground by the voltage developed across the cathode bias resistor. With AC heaters, wouldn't that put 6.3VAC at the cathode? Or if with DC heaters, wouldn't you have the heater DC voltage at the cathode?

Of course, it must be DC. And yes, voltage would be that of the filament. With a small signal tube, the tube current would have little affect on the filament voltage, so it would act as a bypassed cathode resistor. Or think of it as a different way to do fixed bias, limited to the filament voltage, of course.

rongon said:
Is there some reason why an LED would be preferable to something like a 1N4148?

(Possible trolling question) - Any preference for LED color? Red? Green? Amber? Blue? Infrared? Other?

You'd need three small signal diodes to equal one red LED, with a slight increase in noise. Plus LED's emit light. Color depends on voltage drop needed, though noise and impedance differs somewhat between types. I think Christer did a comparison. Search under LED Noise.

rongon said:
I've seen cathode bias as low as- 0.5V in the first stage of a phono preamp, using 12AX7 or 6DJ8. Are these drawing grid current with the 47kOhm load of an MM cart? What about 1kOhm for a high-output MC cart?

They will draw some grid current, and the amount depends on plate voltage (higher current at lower plate voltage). May not be a problem with low grid resistor values. If the cartridge is not cap coupled, would want to make sure current is low enough not to upset the cartridge.

Sheldon
 
A good trick is to bias LED by constant current about ten times higher than a tube draws. Such a way you keep it's dynamic resistance low and less dependent on signal, i.e. higher amplification factor and lower distortions.

However, the obvious disadvantage of a LED load is the same as an obvious disadvantage of any fixed bias: absence of feedback by current that helps to keep plate voltage less dependent on tube's emission (brand, age, etc...) But it is good for tube rollers: with fixed bias they have a toy to play with tubes listening to their sounds.
 
A good trick is to bias LED by constant current about ten times higher than a tube draws.

You have to be careful about that and limit this current to something that over the "knee" of the Vf vs. I curve but well short of Imax. With a CCS plate load, I found that increasing LED current had about zero effect on gain and distortion.

It's easy to see why. Let's take an extreme example- an ECC83 running at 0.75mA. With a common red LED, this will increase the dynamic impedance from 4 ohms (at 10mA) to 8 ohms (at 0.75mA). The effect on the plate resistance is to increase from about 70k to 70.8k. Pretty negligible considering the tube current swing is going to be maybe one fifth of the idle current with a resistor load.
 
rongon said:


As I see it, you're putting one leg of the heater to the cathode of the tube. So one leg would be lifted above ground by the voltage developed across the cathode bias resistor. With AC heaters, wouldn't that put 6.3VAC at the cathode? Or if with DC heaters, wouldn't you have the heater DC voltage at the cathode? I can't see how that makes sense. But you know, I'm not any kind of expert at this.

I actually prefer with a CCS as it is a little easier. But, the idea is that you run the heater's current through the biasing resistor. Obviously you need a heater per channel meaning either a CT heater for dual triodes, or single triode for this to work.

So, for instance, a 5687 needs 450mA @6.3V per heater side. If you run this current plus the plate current (~20mA) through a biasing resistor, then, for a -4V grid bias, you need a 8.5 ohm resistor which adds about the same amount of impedance to the plate as a red LED. Actually, less, as you'd need 2 LEDs in series for the same bias.
 
rongon said:


Any preference for LED color? Red? Green? Amber? Blue? Infrared? Other?

The basic red led will give 1.7V drop and offers the best dynamic properties; an exotic red one around 2.2V: If that isn't enough then series a couple up. This is better than a green one.

If you examine some pentode video tube plots in triode mode, EF184; 12BY7 ete (Tom Schlangens site): one arrives at a bunch of curves which for a given anode current, g1 volts lies very nicely between -1 and -4V. My favourite is a 2.2V red led running at 4-7mA.
Once youv'e got this bug it's hard to shake it off.


richy
 
jonusgrumby said:
......................... I would not use LEDs in low current stages (less than 5 mA)...................

This would preclude using LED bias for 12a*7 tubes in preamp circuits. The idea of only ~1ma going thru the diode did seem strange to me which is why I got a “low current” red LED for my 12ax7 oscillator circuit. (Got the idea from the valvewizard site.) I didn’t see any spec differences between this “low current” LED and the others so I don’t know what “low” is supposed to mean.

The LED is working nicely with 1.7v of bias, matching the bias previously obtained with a 1.5k (and a 100u cap.) It blinks with the oscillation providing a visual speed indicator on the front of the (guitar) amp. It also appears to be stronger than with the traditional bias approach.

What would be the concern with using LED bias in a low current tube?
 
jjman said:
This would preclude using LED bias for 12a*7 tubes in preamp circuits. The idea of only ~1ma going thru the diode did seem strange to me which is why I got a “low current” red LED for my 12ax7 oscillator circuit. (Got the idea from the valvewizard site.) I didn’t see any spec differences between this “low current” LED and the others so I don’t know what “low” is supposed to mean.

The LED is working nicely with 1.7v of bias, matching the bias previously obtained with a 1.5k (and a 100u cap.) It blinks with the oscillation providing a visual speed indicator on the front of the (guitar) amp. It also appears to be stronger than with the traditional bias approach.

What would be the concern with using LED bias in a low current tube?

Slight misquote of a 60's saying: "If it sounds good, do it." If you like it, leave it.

I don't think that there is any inherent advantage to low current LED's. For lowest distortion, the number you want to look at is V/I. The less, the better, and the straighter the curve, the better. But, you can add some current to the the biasing LED with a simple dropping resistor from B+. Just calculate the amount of B+ you want to add. But this will eliminate your blinking effect. For your application, I wouldn't bother. I don't think distortion should be a big issue.

Sheldon
 
jjman said:


This would preclude using LED bias for 12a*7 tubes in preamp circuits. The idea of only ~1ma going thru the diode did seem strange to me which is why I got a “low current” red LED for my 12ax7 oscillator circuit.

You need to be careful to to confuse hifi applications with guitar applications, and with non-audio (e.g., oscillator) applications.

For hifi you want to minimise the dynamic resistance of the LED, and keep what little resistance which does exist as linear possible, and keep the forward voltage of the LED as constant as possible. That means running some healthy current through the LED. You can still do that with a 12AX7 by feeding extra current into the LED from elsewhere, via a resistor from the HT, or even from a CCS.

For guitar use these above concerns are negligible, since getting absolute minimal distortion is no longer important. Even at <1mA, LED bias is still waaay more "advanced" than the crude RC biasing used in guitar amps, and entirely swamped by all the other distortions within the amp.

For the trem oscillator in a guitar amp the same is true. Absolute linearity isn't required, and it'll still be better and more maintenence-free than the old RC biasing. And it flashes, which is cool!
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.