Tube Amps / Music Electronics
|For current discussions, please visit Music Electronics Forum.||New: view Recent Searches.
New: visit Schematic Hell!
The sunn still shines online!
|Listen to great tunes streaming live right now!|
|9/1/2000 3:09 PM|
||Re: "Crossover notch" biasing - my final answer!|
Yeah, PSpice is cheating! Actually, I'll guarantee the Line6 chips were designed using a simulator - when a company designs a massive integrated circuit they will simulate it first to make sure it has a good chance of working before they fork out the money for the wafer fab. Fortunately for us, Spice can be used for good or evil, and when properly applied, the result can be a better grasp of the workings of tube circuits. Just be sure to resist the dark side of the force, and ignore all those BJT and MOSFET models that in there.
When I was in college, I took a course in IC design, and we had to start by designing our own tiny MOSFETs of all different sizes, then we had to design a library of basic multi-input NAND gates and other logic functions, and finally, we had to pick any MSI IC and duplicate it's function using our gate library. Everything from that first transistor on up to the full IC was simulated in Spice. It was quite fun, but very time-consuming. Unfortunately, the professor didn't consider a Plexi to be an MSI IC.
Once you get a small distance away from class B operation, the crossover notch is no longer visible. It doesn't take too much bias current to render the notch invisible, particularly if the tubes are well matched. This is because the non-linearities in the individual curves cancel each other out in the composite characteristic. If you have ever seen a graph of a push-pull composite characteristic, where the second tube is inverted and graphed below the first, you'll see that the "bending" of the curves down towards the zero bias line will become straight when you generate a composite bias line from the two tubes. This "linearizing" is one reason why you can get more power out of two tubes in push-pull for a given amount amount of distortion.
If you look at the curves in my simulations, you'll see the crossover notch goes away at fairly low bias currents (another reason this method gives you a cold bias point). I did some simulations on the AC-30 output stage and found that each output tube was in cutoff for around 30% of the time at full undistorted output. This is quite a bit higher duty-cycle than your 6L6 output section at 480V, so the crossover notch is not visible, even though each tube is completely shut off 30% of the time. That's the beauty of push-pull operation - when one tube gives up the other is just nicely getting going, so it takes over. You never see any visible indication of the cutoff time on the scope, and if you didn't know better, you'd think that both sides were on all the time. You get more output power because the average dissipation is lower in each tube due to the amount of time it spends in the off state and the lower required bias current.
The reason the crossover distortion becomes apparent when you clip the output stage is because of the grid clamping effect I mentioned previously. The grid normally is a very high impedance, until you try to drive it positive. When that happens, it acts like a forward-biased diode, and becomes a very low impedance, "clamping" the peaks of the grid waveform to a level near the cathode voltage. Since the peak is clamped, and the output tube grids are capacitively-coupled to the phase inverter outputs (which are relatively high-impedance drive), the grid voltage has no choice but to get "pushed down" as the drive increases. Since the average level of the sine wave is at a more negative voltage when it gets pushed down, this is exactly the same effect as if the grid bias voltage was made larger. It effectively pushes the amp farther towards class B operation, with the resultant increase in crossover distortion.
This is why your AC30 clone shows a crossover notch when you go into clipping. If it were a true class A amp, you'd have to drive it much harder at the grids to see the crossover notch, because the grid voltage would have to be pushed down to almost half it's full-drive amplitude before the average bias current is low enough to see visibile crossover distortion. In your amp with the high plate voltage, you are already biased quite near cutoff, so it doesn't take much grid drive to show a large increase in crossover distortion.
|9/1/2000 5:39 PM|
What about rectification effects (average current increases at high levels if it's not class ..er and then charges the cathode bypass cap, making the grid more negative)?
|9/1/2000 5:42 PM|
that's certainly easy enough to measure... just connect your DC meter across the cathode resistance, and measure the increase as you drive the hell out of the amp.
i suspect it's quite big enough to be a factor. i think it is also one of the reasons you can't reasonably run a class B amp with cathode bias.
|9/2/2000 1:22 AM|
Yes, GFR, you are correct. If the time constant at the cathode is short enough in relation to the input signal, you will get a shift in the average DC level, which will change the bias in cathode-biased amps. This is on top of any shift at the grid due to the AC-coupling clamping effects.
|9/4/2000 11:21 AM|
Thanks for the explanation. Electronics was a hobby that ended up paying off my mortgage. I’m still learning all the time. If I can avoid electrocuting myself one of these days I may actually know what I’m talking about. This thread sure turned "interesting" over the weekend. When you said "I suggest you take your scope probe.." I was wondering what was coming next
|9/2/2000 12:50 AM|
||Re: "Crossover notch" biasing - my final answer! (Long-winded ramble: Part 1)|
OK, this is going to be Part 1 of a very long-winded ramble here. I apologize now for boring the crap put of anyone who reads this. If you have the patience to read through it ALL, you will have much food for thought, I promise.
Firstly, R.A.'s technical paper is well researched and well written. I admit I was entrapped by his question of how I would bias up an EL34 amplifier. His paper does not cover aspects I measure that are not part of the bias procedure, but a part of weeding out bad EL34s. This is important to remember because without counting bad EL34s, tubes are a very forgiving thing.
I met an elderly Dutch fellow a few years back that actually worked as an engineer at the Philips tube factory during their golden years of tube manufacturing. His exact words to me ring so true the more I learn about tube circuits. To paraphrase (use your best 'Sven' accent):
'The numbers in your tube manual are just a suggestion on how to run the tube conservatively. We would torture test a new tube to see what it could do, and then publish the data telling you how we would like you to use the tube.' Obviously, Fender knew this as well. This is why every Deluxe Reverb has about 415VDC on the plates of a 6V6, which is rated at 385VDC MAXIMUM on its' plate(s). But enough chatter, let's read R.A.'s paper.
-Firstly, I could never see how biasing up an output stage would be construed as being potentially hazardous if '…at best… it biases the amplifier to a very cold point of operation.'
A FEW REASONS TO AVOID THIS METHOS OF BIASING
-It is not an inaccurate method. Let's look at the hit-the-nail-on-the-head accuracy of the current draw method. How about Weber's current draw 'magic number' suggestion: 'Not less than 10mA and not more than 40mA per tube. Any bias setting that gives you tone you like is correct.' To be fair, he had this propaganda teat fed to him by Ken Fischer. And to further refine this highly accurate method of biasing, we have different schools of thought about plate dissipation. Was it 50% or 60%? Or 70%? Oh, well, why be wrong but definitive when you can be vague and always 'sort of' right?
-'Because of this clamping effect, the amount of crossover distortion that you are trying to 'bias out' will change depending upon how far into clipping you set the grid drive.' Maybe, maybe not. First, you don't set the grid driving voltage that high to begin with, and secondly you may/may not what you are seeing on the 'scope. From the book 'Audio Design Handbook' (H.A. Hartley, 1958):
'…if the amplifier under test is driven into overload (t)he increased harmonic distortion will then be readily seen and its nature determined.' After showing us a multitude of traces displaying 2nd, 3rd, 4th, 5th, 2nd + 3rd, and 3rd + 5th order harmonic distortion at various phase relationships to the original sine wave, we are shown that a crossover notch looks 100% EXACTLY like the sine wave with:
a)30% 3rd harmonic distortion, -90 degrees out of phase.
b)5% 5th harmonic distortion, 180 degrees out of phase.
c)5% 2nd harmonic and 30% 3rd harmonic distortion, in phase.
d)5% 3rd harmonic distortion 180 degrees out of phase and 5% 5th harmonic distortion in phase.
This may not be understood by 99% of people who own 'scopes. Also, keep in mind the 'scope 'quality'. Poorer 'scopes with an anode accelerating potential of only 2kV won't show the waveform as nicely as a 'scope with an anode accelerating potential of 12kV. So these people blame the method when they can't 'see' when the notch disappears. Perhaps I wasn't so rough in suggesting someone learn how to read a 'scope. This is also why I don't bother with the asinine suggestion about the effects of global negative feedback. Of course the extra driving voltage will drive the amplifier into overload. But does the amplifier have FB or not? There's no in between here.
-'If you try to bias a push-pull class A amplifier using this method…'. Hold your horses right there. The sentence should conclude… 'you haven't read your tube theory books enough'. Class A amplifiers draw maximum plate current 100% of the time, so when could you have a crossover notch to try to bias out with this method? To bow to the current draw Gods, this is the ONLY time I would bias an amplifier ENTIRELY by current draw (still comparing plate dissipation).
Wow, I'll stop there for now and catch my breath. Later.
|9/2/2000 3:42 AM|
I think it's the worst case scenario you should be worried about.
Surely you've dimed an older Fender. It doesn't take too much overdrive to create this effect. Surely you can hear it, can't you?
This is just out of curiosity, but if it looks 100% EXACTLY like a sine wave, how can you see the distortion characteristics? Further, if it looks 100% EXACTLY like a sine wave, there is no crossover notch. Think about it for a second.
The issue is the effect of the NFB. There comes a point when the amp just can't do any more and the NFB ceases to have an effect.
I'm just commenting here so someone learning about this doesn't get confused. A better way of saying what I think you're trying to say, is that all the tubes in a class A power amp conduct for 100% of the time. When mentioning the class of an amp, there is no reason to mention anything about plate current. It has nothing to do with the class of an amp.
|<<First Page||<Prev||Page 2 of 7||Next>||Last Page>>|