When we bias a tube amp what are we measuring?

[ducati_EL34]

I like your Sinatra quote, made me smile.

What is a present wrench. did you mean preset? If so I would hand them a non adjustable wrench.

That should have been "crescent wrench".

I dont know of any single ended op amps, by definition op amps are differential otherwise they could not perform their "operations".

That is not their only function. While all op amps have two inputs, there are times when only one input is used and the other is grounded, hence single ended.

 

[sKiZo]

Also nice they say mA on them since that is what you are reading, the shunt is just elsewhere now.

Yup ... internal shunt snipped ...

... and replaced with an external VR that allowed me to tweak each lead to compensate for uneven length and resistance on the leads for each tube ...

(I really should clean that up - I went thru a few iterations before I got it right.)

 

[dcgillespie]

Assuming the discussion is limited to operation without grid current, once Eg1 = 0, a tube is saturated and conducting maximum current flow to the plate, screen voltage (if pentode), power supply, and load permitting. Whether the grid got to 0 from a quiescent bias point of -25 or -35 -- or from an even much greater negative number due to dynamic operation in Class AB1 or B1 mode, it is irrelevant. Once Eg1 = 0 is reached, the plate is conducting maximum current.

Now considering a push-pull Class AB1 amplifier, and using gross exaggeration to make a point, on the back side of things -- that is, where the quiescent bias point is actually set -- will in fact have an effect on power output -- again, using exaggerated conditions to make the point:

If the quiescent current is set so high as to nearly reach saturation in the tubes, then there won't be any ability for the tubes to conduct much more current to generate any output. This point has been well proven to anyone who has ever witnessed a bias failure of some sort in their amplifier: The tubes red-plate, and the sound becomes very weak and distorted since the tubes have precious little ability to conduct any "audio" current flow in this scenario. It's all being chewed up by the damagingly high quiescent current that of itself, produces no power output at all.

Now go the other direction. Apply (say) 3X the normal bias voltage to drive the tubes deep into cutoff under quiescent conditions (remember, this IS an exaggerated scenario for the purposes of illustration). And to keep it interesting, the driver is still capable of driving the output tube grids to Eg1 = 0 in this scenario. Here, the tubes will still reach saturation current on grid signal peaks. But there will be such a hole in the middle of the waveform, caused by there being absolutely no current flow well before the applied signal approaches the quiescent bias point. With such a distorted waveform, actual RMS power output delivered to the load is significantly reduced.

Under normal operating conditions however, the effect on power output in moving the standing current to either side of optimum (within a range of (say) above that which produces cross-over distortion and below that which good design practice would consider excessive quiescent plate dissipation) will be so small as to require sensitive measurements to determine. In other words, there will be no practical effect at all on the amount of power that the amplifier can deliver into the load.

In a Class A single ended design, maximum undistorted power output will be achieved with the quiescent operating point basically centered on the load line, since movement either way from this point will chop off the top or bottom of the waveform in this scenario, producing gross distortion, and requiring the drive to be reduced to such a level that the complete waveform is produced again. In terms of maximum undistorted power output then, a single ended Class A design is much more sensitive with regards to actual location of the operating point than the typical Class AB1 push-pull design is.

Dave

 

[Archguy]

That should have been "crescent wrench".

Whew, thanks for that. I was really mystified by that one!

And another upvote for this entire thread, which is civil and very educational--even for those of us who need a life preserver in this end of the swimming pool.

 

[Dandy]

RogerMod, would you consider starting a separate thread (or threads) outlining your thinking in the design of one (or more) of your amps? This would enable us to follow your thinking about the design decisions sequentially, rather than via the 'zig zag', however helpful, of a thread like this.

 

[maxhifi]

I went to the trouble of making the RM-200

I looked up specs on your amplifier, and I have to say speaking of semantics, there is one spec I am very impressed with. I really like how you specified damping factor also as output regulation, in dB. This is how RCA used to specify their amplifiers, and I believe it's a more useful and practical way to spec output impedance, from the perspective of the user. I wish all amps were specified this way.

 

[dcgillespie]

I would suggest that the fuss today is primarily about folks wanting to ensure that their tubes are operating within a safe zone that will ensure a combination of good audible characteristics combined with good tube life. Ergo, setting the bias to a percentage of rated plate dissipation "rule". Actually, there never was such a rule but rather, ensuring that the resulting operating conditions were were within good design practice relative to the Pd rating for the tubes used. In other words, it was just a parameter to observe in the design process. Today however, with tube life taking on foremost importance, and most equipment users not having the necessary equipment to adjust the bias to an optimum setting, a Pd rating percentage rule is an easy way to ensure achieving good tube life and acceptable performance with just a simple DVM. Assuming a Class AB1 push-pull design, varying the quiescent current within the current range suggested will have little impact on power output as discussed, but have a pronounced impact on the amount of distortion produced.

Dave

 

[dcgillespie]

Bias voltage and standing (quiescent) current are certainly (and obviously) interconnected. But bias voltage itself is simply a means to an end. It is a characteristic of given tube type, but for a given current draw under stated conditions, is subject to all the variables that tube condition and manufacturing tolerances create. And of course, the higher the Gm, the more those tolerances come into play. For example, a type 2A3 which typically uses about 65 volts of bias is not nearly so sensitive to changes in bias voltage, as a type 8417 or 7591. With the old 2A3 amplifiers then, it was acceptable enough to simply specify a bias voltage -- which most schematics for amplifiers using that tube in fact did -- rather than standing current. But understand that standing current is the end game that is sought to be achieved. As newer, and much higher Gm tubes were produced, specifying just a bias voltage can get such a tube into trouble real quick if it is off tolerance at all. Witness those with stock Fisher receivers (for example), who plug in modern day Tung Sol 7591 output tubes. Things can and often do get out of hand real quick in such a scenario if adjustment aren't made. Higher average AC line voltages then only add insult to injury in this case.

In the end, the salient point is that it is the standing current that matters, because it is the correct standing current that:

1. Ensures that in single ended Class A amplifiers that the operating point is centered on the load line, and

2. Ensures that in Class AB designs, the optimum hand-off conditions from one tube to the other are in place when Class B operation commences.

In both of these cases, whatever the grid bias voltage is that produces the necessary standing current is whatever the grid bias voltage is -- as stated earlier, it is simply a means to an end.

In terms of what characteristic determines the optimum setting, in a well designed Class AB1 output stage, the optimum setting is classically that setting that produces the lowest distortion when near full power is being produced -- say, at 1 db below the onset of clipping to ensure that the adjustment is made with the output level safely under clipping conditions.

All of this also speaks as to why for years I have sought to discourage relying on the results of Gm testing for power output tubes in an effort to determine condition or quality. The transconductance (Gm) of a power tube is a very useful tool in describing the characteristic of a given tube type, but as a measured parameter, it is rather poor for assessing the quality of a power tube. By the time the transconductance has lowered to an unacceptable level, the power output capability of the tube -- primarily a function of the cathode's ability to emit enough electrons to produce an acceptable output level, has long since been compromised -- this because Gm testing is done at very low levels (relatively), compared to power output testing (the very best quality indicator of a power tube), which is done at high levels. Tubes that fail a power output test, but pass a Gm test will no doubt still perform usefully in low demand applications, which is likely the case for most users today (we're rarely expecting full power output to be produced on a sustained level). But knowing the limitations of Gm testing with power tubes is still important when maximum performance capability is on the line -- all of which is to say that just because a given set of tubes can be biased to the proper quiescent current point doesn't mean that they are capable of producing full rated power output. A classic indicator of this condition is where the tubes in fact bias properly to the correct quiescent current, but require an unusually low amount of bias voltage to produce the needed current draw -- which bring us right back to the grid bias voltage/quiescent current draw discussion, and shows once again that it is the correct quiescent current draw that is the desired outcome. The grid bias voltage to achieve that end is just a characteristic means of producing it.

Dave

 

[dcgillespie]

Then the tubes are either grossly mismatched if by "channel" you mean one side of a Class AB1 push-pull pair, grossly out of specification for the tube type (supposed to have a gain of 4), or just shot. In any event, you are describing a tube problem, not a theory problem.

Dave

 

[dcgillespie]

And of course, this is what (in part) NFB is for in REproduction equipment -- to help iron out the real differences like you just described. Now in a guitar amp setting however -- which uses little to no NFB -- the tubes with a bias voltage of 11.75 volts will overload sooner relative to the volume control setting, and be said to develop more "crunch". The tube requiring 13.0 volts will be said to develop more cleaner sound, presumably because the amp can be turned up higher before distortion sets in. All else being equal however, the actual maximum power output between the two tubes will be the same.

 

[Selmerdave]

I'm still not clear on the bias-does-not-affect-maximum-power thing, so if someone is willing to indulge me to help me understand I'd appreciate it.

My understanding is that class of operation is largely a function of bias (in conjunction with B+ and primary Z), and for example with a pair of PP EL34s biased in class A it's not *possible* to get more than about 25 or 30W. In AB though, it seems easily possible to get 60W or more with low distortion from the same pair. So what am I missing with this equation?

Edit: is it that the lower quiescent current of the AB bias allows for higher B+ without exceeding plate dissipation? Thus only with the higher B+ is higher output obtained?

 

[dcgillespie]

6 -- No. I believe you're confusing Power with Gain -- easy to do, but two distinctly different things. In your scenario, both tubes still produce the same amount of power output -- but the tube requiring only 11.75 volts of bias will reach full power output sooner than the tube that requires 13 volts of bias. Now granted, if the input signal is advanced so as to drive the 13 volt tube to full power ouput, the tube requiring only 11.75 volts will now be over-driven. But again, once the tube saturates, maximum power is achieved. In your scenario, one tube is just achieving full power before the other because it is displaying more gain.

Dave -- In a Class A (single ended or push-pull) design, the tube or tubes by definition must conduct current at all times over the entire excursion of the input waveform, from quiescent to maximum power output. For the tubes to produce an equal maximum excursion on either side of the operating point then, while never being driven to the point of cutoff, that point must therefore basically be located in the center of the load line. Such operation requires a heavy quiescent current to be drawn, which therefore requires a reduction in B+ voltage so that the tube's plate dissipation will not be exceeded.

In a Class AB design, the operating point is moved off-center on the load line, and pushed close to but not up to the point of cutoff. Since this greatly reduces the quiescent plate dissipation, it allows the opportunity to raise the B+ to once again cause significant (but not excessive) plate dissipation to occur under quiescent conditions. With an off-center operating point however, it means that the output stage must now operate with two tubes in push-pull, with each tube conducting current during slightly greater than half, but much less than all of the complete input waveform applied to them, in an alternating fashion. This scenario allows for greater dynamic plate current and plate voltage to be developed from each tube (albeit for just a little more than half a cycle, each), and therefore represents a reduced output impedance which in turn, allows for the use of a lower impedance output transformer. A lower impedance output transformer has a reduced turns ratio between the primary and secondary winding (i.e., not stepping the voltage down as much), allowing more voltage to be delivered to the load, which represents an increase in power output over that produced by the Class A scenario.

To keep the thread on track then, a change in bias voltage that moves a design from Class A operation to Class AB operation, can in fact ultimately result in a significant increase in power output. But it is hardly created by the shift in bias voltage alone. As previously stated, the bias voltage that a given tube requires in a given application is merely a means to an end. It is no different in this case, either. The shift in bias voltage allows the use of greater B+ voltage and ultimately a lower load impedance OPT for the tubes to operate into, that together creates the increased power output developed.

Dave

 

[Selmerdave]

Many thanks Dave, that helps a lot.

 

[dcgillespie]

In your scenario, without any corrective measures in place (NFB), the two tubes are not producing the same power output for any given (assumed equal) drive level (up to 11.75 volts) as the two tubes are not displaying equal gain. However, they are both capable of producing the same power output, once their individual (unique) saturation drive has been achieved. Increased drive beyond that which saturates the tube with the highest gain -- in an effort to achieve full power output in the tube with lower gain -- will then be over-driving the higher gain tube.

Dave

 

[Selmerdave]

These two tubes are creating the bias voltage from the same plate voltage and cathode resistor value. They don't require those particular levels, they do obviously pass slightly different idle currents,.... but to carry your reasoning that they both produce the same output signal power suggests I could set bias, as a fixed bias scenario, on channel A to 9v and let it clip off 4v and they would still be producing equal power at 13v of drive.

I think the issue is that power is the product of voltage and current and you're only looking at the voltage. A smaller voltage swing at a higher current is the same power as a larger voltage swing at lower current.

 

[Selmerdave]

Current is part of the output power (as per Ohm's law), otherwise the 12ax7 would make a great power tube.

 

[RogerMod]

Lower the bias voltage from -35 to -25 and you have to limit the volume level which of course limits the swing and power.

Not at all. The bias point does not determine the gain nor the clipping point in a feedback amplifier. Are you considering that? Besides that large a change in bias voltage would typically create a 60 ma change in idle which is unreasonable. If already biased at 60 ma per tube now it is going to be 120 mA. HOT, HOT,HOT.

 

[RogerMod]

I looked up specs on your amplifier, and I have to say speaking of semantics, there is one spec I am very impressed with. I really like how you specified damping factor also as output regulation, in dB. This is how RCA used to specify their amplifiers, and I believe it's a more useful and practical way to spec output impedance, from the perspective of the user. I wish all amps were specified this way.

Thanks, I wish all makers did this also. It is then much easier to see what effect regulation has on frequency response with impedance variations. We have to realize that Damping a woofer isn't everything, in fact, its the minor effect of the two. As Paul Klipsch said. Who cares about a ohm or two when the voice coil is an 8 ohm resistor. There is really no way around that.

 

[RogerMod]

This is kind of a chicken or the egg puzzle. Which comes first, usable bias voltage range or perfect idle current.

The designer says... I need a 200v swing through the outputs, my tubes give me a gain of 4, I need +25v signal on the grid and a minimum of -25v bias. He throws in a set of tubes and biases to his paper design specs at the test point. 1v everywhere. Pefect idle current. All good. He cranks up the volume and about half way there it sounds like @#^%$@$%, he checks and finds -15v of bias on one channel.

If a tube biases at -15v and the target was -25 then you have a very weak tube. That is all. Throw it away.

 

[maxhifi]

Not at all. The bias point does not determine the gain nor the clipping point in a feedback amplifier. Are you considering that? Besides that large a change in bias voltage would typically create a 60 ma change in idle which is unreasonable. If already biased at 60 ma per tube now it is going to be 120 mA. HOT, HOT,HOT.

Ok Ok, I have a question... Of course after seeing you on this site before, I had to make myself familiar with some of your products. You can probably anticipate what I am going to ask, but I will ask it anyway. 35W from a pair of EL84? HOW? What are the operating conditions? It's incredible this is possible. Are you running them at 500V and barely any idle current? I've never seen anything like that before, it makes me wonder how much power one could get from 6L6GCs if the same sort of design was applied.