The Amp Bias Controversy

[Nowhere-Man]

I'll start with a quote here and maybe some will find this is an issue that deserves some debate. This was a comment made on an American radio engineer's book:
"There's a reason you will never see this author employed at a college or asked to give
lectures in the electronics community.Other mistakes such as biasing the grid in a triode to be positive is more evidence of someone who is poorly educated and has never spent even just a year in college. His definition of words are wrong and it's just bizarre to find someone this incompetent is actually writing books and making tutorial videos."

Over the last 3 weeks I've been finding gaps in my understanding of valve amplifier stages. In fact, I think this is not well explained at all. I spent some time tonight simplifying it as much as I can in plain English.
The guy slamming the book writer seems to be thinking of class A1 amp stages where the grids are always negative. Yet the writer probably meant that class B amp grids are driven into positive values. In basic English I summarised class A amps are voltage amplifiers so the signal voltage is amplified. Grid stays at negative values and plate current flows throughout the cycle. The class B amp in push pull splits power output between 2 triodes so one rests as the other works. The strong signals will drive each grid positive. So, I have no issue saying the triodes could be biased positive. Surely it's not such a major language issue? Oh, as many know they call class B a power amp with emphasis on watts.
The AB is supposed to be a bit of both. Best explanation I found online is class AB2 on push-pull is cut off on large negative signals while the other triode goes into positive on the grid.
Conclusion: it's a complicated matter that's hard to explain. Plus, I had a hard job sifting through various explanations. I don't think I agree class A1 amp valves keep the same plate (or anode) current regardless of signal value. Surely there must be rise and fall of current between say -5 to -3 on the grid. It's just the anode load resistor gets most of the voltage drop (hence "voltage amplifier".

[Murphyv310]

In simple terms in class A the bias should be arranged that it is near mid point of the characteristic curve, the resultant anode current will then be "x" it's mean DC figure, when a signal is applied it will be "x" + or - the signal appearing on the anode, run say a 1khz signal to the grid and the mean anode current will remain virtually the same, measure it with a scope and then you will see the AC component out of phase with input and obviously amplified. This is in simple terms.

In class B the grid bias is near to the bottom of the curve, and there should only be a small anode current, you cannot fully bias the valve off as you will land up with crossover distortion  so as the valve is fed with signal it will only conduct on the positive half cycles as they make the bias more positive and turn the valve on harder. A phase splitter is needed to reverse the signal so the lost lower part of the waveform that valve 1 cannot amplify on is then amplified by valve 2 fed from the phase splitter are combined in the output transformer after amplification from valve 1 & 2 the output pair.
The standing current or quiescent current is therefore low in a class B amp, the current is directly proportional to output, the louder the more HT current is required.

Finally there are some circuits that require a positive grid voltage, these can be in valve shaping or clipper devices, anode currents though are usually low.

[Terry]

For once, your American book is right - it is physically impossible to put a positive bias on a (control) grid.

However, it is important to measure it correctly. The bias is the voltage between the grid and cathode - not earth - although, in some circumstances it doesn't matter which you measure but this will only be when the cathode itself is connected to earth.

If you attempt to apply a positive bias to the grid, the grid effectively becomes the anode of a diode. This grid/cathode diode will conduct, as all diodes do, effectively clamping the grid and cathode together. If you attempt to increase this 'bias', all that will happen is that the current will increase. Continue with this action and you can expect to see the grid starting to get red hot - in fact, you could end up stripping the emissive materials from the cathode. Your valve will have a short life and not a happy one.

You can be fooled into thinking that a valve is positively biased if you don't examine the operating conditions very carefully. This is the well known Mullard 5-10 amplifier:


When you study it you will see that the anode of the EF86 is connected directly to the grids of the ECC83 without a DC blocking capacitor and, as the EF86 obviously has a positive voltage on its anode you could be fooled into thinking that the ECC83, therefore, is being positively biased. But you would be wrong!

If you look at this voltage table:

you will see that the voltage at the anode of the EF86 is 61V and, as expected, the voltage on the grids of the ECC83 is also 61V. However, if you look at the cathodes of the ECC83 you will see that they measure 63.5V, 2.5V higher than the grid voltage. This means that the grid bias is actually 2.5V negative wrt to the cathodes.

The ability of the grid and cathode to be used as a diode can be used very effectively in the right circumstances. The leaky grid detector has been used virtually from the beginning of receiver design.



Here the RF signal is applied directly to the unbiased grid of the valve so the valve promptly clamps the positive going half cycles of the incoming signal to the cathode, thus it functions as the detector diode, passing the modulation onto the anode where the 1nF capacitor shunts the remaining RF to ground so that only the recovered modulation appears at the output.

However, this isn't  positive bias! In fact, if you apply a very high impedance meter to the grid you will see that it has a negative  voltage on it which varies as you tune through different transmissions being more negative on stronger stations than weaker ones. If employed later in a multi-stage receiver, this varying negative voltage can be fed back to earlier stages as AGC.

[ppppenguin]

There may be a misunderstanding of what is meant by "grid bias". My understanding, which I think is correct, is that it is the DC potential, usually in the absence of signal**. As Terry has said, it is measured between the grid and cathode. This is never positive in any design I've ever seen. In some designs, notably high power output stages, the +ve peaks of the signal may drive the grid positive with respect to the cathode and hence cause grid current to flow for short periods.

**As Terry said, itis possible for the signal to be rectified and affect the bias.

[Nowhere-Man]

I'll comment later. By the way ARRL wasn't quoted in my first post - it was a review comment of a different book made on Amazon. As I haven't read the book that was reviewed, I can't judge if the writer was talking about AC positive signals driving the grid into positive values.
As ARRL was mentioned though, they mention operating angles in degrees.
What had never sunk in in the past was that voltage amplifiers use very high load resistors, to create large voltage drops across the load.
For power amps the operating angle maybe helps best with 180 degree operation of each triode at B class but 360 degree plate (or anode) current for A class.
As grid negative was mentioned the way I always viewed it as direct short of grid to chassis should cut off the valve. A resistor between grid and chassis should lessen the negative bias.

[Mark Hennessy]

Power amplifier classes are pretty clear-cut on the whole, and yes, it's best to think about the signal in degrees. This appears to be a reasonable summary: https://en.wikipedia.org/wiki/Power_amplifier_classes

AB1 and AB2 are for valves only. Outside of my knowledge and experience.

Most transistorised audio amps are class AB. Funnily enough, when you've eliminates all the other forms of distortion in a good power amp, class B is better than AB (most assume the opposite). The difficulty is holding the exact bias conditions for class B operation, so class AB operation is cheaper/easier/safer alternative to the risk of dipping into class C under certain conditions.

It can occasionally be difficult to be sure about which class an amp works in given a certain set of signal and load conditions, but it hardly seems worth worrying about IMHO. Certainly, "controversy" feels like far too strong a word. Borderline "clickbait", frankly.

[Nowhere-Man]

I'm a great believer in frequently going back to beginnings. When I first started the basic radio theory, I had Chas Miller's tutorial. Chas taught his students to measure grid bias with + meter probe on chassis and - on grid. As I progressed, I was taught that the grid of a triode is always in negative values. Then the alternating signal would add or subtract from this negative bias. Yet now the time has come to dig deeper. To fully understand modulators, voice amplifiers and drivers, it pays to understand the amplification stages. Sure, I've come across push-pull before but there is a bit more to it. There is the phase aspect and the effects on impedance (driver stages).
Class A stage was the one that had me most baffled. Rules so far are, the output waveform should be in phase with the input. The triode should be drawing anode current throughout the cycle (360 degrees) but the grid never goes into positive values. The emphasis is on voltage amplification so the anode load will be very high. Oh, and resistance capacity coupling is used.
The confusing bit too is when "grid current" is referred to. Not "anode current". When grid current per se is drawn, power is lost in the grid. In class B, each triode rests totally from any grid current draw while its partner triode draws into grid positive values.
Maybe a simple way of putting it is when the grid doesn't draw current, the valve is more power efficient. Really though I need a bit more time to digest it. It has no bearing on practical engineering at all but I like to progress in stages.

[Terry]

Class A stage was the one that had me most baffled. Rules so far are, the output waveform should be in phase with the input. 

No! That is impossible!

As the input signal becomes more positive, the anode current increases which causes a corresponding increase in the voltage drop across the load. Thus the anode voltage drops i.e.: goes negative.

Thus input and output voltages are in anti-phase.

You are obviously still confused by the concept of grid current. The whole point is that grid current should never be allowed to flow in virtually all applications. The leaky grid detector is an obvious exception but the current in that case would be infinitesimally small. 

As for the grid always being negative wrt the chassis, that is rarely the case with indirectly heated valves. If you look at the 5-10 circuit again, you will see that the grids of the EF86 and EL84s are connected to the chassis (via a resistor) but there is a resistor in the cathode circuit which develops a positive voltage on the cathode wrt the chassis. The grids are, therefore, negative  wrt  the cathodes.

[Nowhere-Man]

Not impossible. Output phase is the anode load. If signal voltage on grid is postive, anode current increases. Multiply that current by the anode load resistor and the sun is higher. Therefore the anode voltage falls and is out of phase with the grid cathode voltage. Yet the anode load is in phase. If the anode load was a bulb it would glow, consume watts and detract from the valve. If HT is 300 volts and fixed anode cathode voltage is 100 volts, you get 200 volts across the load resistor. Yet a varied signal on the grid can swing the current value and vary the load drop.

[Kalee20]

There may be a misunderstanding of what is meant by "grid bias". My understanding, which I think is correct, is that it is the DC potential, usually in the absence of signal**. As Terry has said, it is measured between the grid and cathode. This is never positive in any design I've ever seen. In some designs, notably high power output stages, the +ve peaks of the signal may drive the grid positive with respect to the cathode and hence cause grid current to flow for short periods.

That's what I would say, too. Bias is the steady voltage on which the signal is superimposed.

I've never seen any circuits which operate with a positive control-grid bias, but I have seen examples where the bias is zero (battery Class B double triodes). The signal can and does take the grid positive. Yes, grid current does flow and yes it is a nuisance, but to say it's physically impossible to take the grid positive, is wrong.

In Class AB2, and RF Class C, the grid is biased negatively but the signal is greater than the negative bias so again, the grid goes positive on signal peaks (and sometimes positive by several volts, not just fractionally).