Thanks for the reply, 1. I have a couple 12dw7s and I was just trying to figure out something interesting to do with them and coax a little more power out.
2. I haven't seen any data sheet charts that list positive grid voltage example but
3. I suspect the grids would be fine as long as I didn't push them too hard but I don't really know for sure.
2. I haven't seen any data sheet charts that list positive grid voltage example but
3. I suspect the grids would be fine as long as I didn't push them too hard but I don't really know for sure.
The tubes have no problem handling control grid current, and do continuously when used in Class C r.f. power amplifiers in transmitting equipment. With the small current flowing only towards the tip of positive peaks in audio service, the average grid power is negligible. As in pentode operation, it is the screen grid dissipation that needs to be watched. Many beam pentodes specify doubled screen ratings on music/speech peaks in class AB or B (1 or 2) as seen with ultra-linear operation, but with the high bias currents many people run tube life is severely impaired. There is also a compounding heating effect where the screen rating should not be pushed when the plate is run near rated value. Bias current down a bit, triode operation is actually not as tough on the screen as the worst case pentode operation which is seen when the plate voltage falls below the screen voltage. That's particularly a problem when their is heavy bass content and the speaker system has a impedance peak due to resonance at those frequencies because the higher load resistance makes it more likely that the available plate current results in a bigger plate voltage swing for a given current. If your driver is capable of producing much larger plate currents than needed and you try pentode mode, a series grid resistor can soft-limit the drive. Some slight compression of drive voltage as grid current rises can actually reduce distortion since it would tend to cancel the effects of transconductance rising with current. Also, being progressive compression it makes for a more pleasant sounding overload characteristic. The Ampeg SVT series AB2 amps have particularly large drive-limiting resistors. What to limit drive to depends on a number of circuit conditions including your choice for the plate load. Screen overload is much less of a problem with lower plate circuit impedance values.
Last edited:
So correct me if I am wrong but when a when a tube works into a resistor for the load the energy is converted to heat in the resistor, and the idea is that same energy could be conserved and passed on by driving and interacting with the grid of the next tube as part of it's load instead. Is this right? What kind of a load does swinging the next tubes screen positive present to the previous stage?
Well if you're thinking of where the power that gets to the grid came from and changes in the power dissipated in the resistor between the driver plate and the power supply when there is a load, the power lost in the resistor actually increases when load current is drawn, because the current through the resistor rises. When the output goes positive, the load current goes through the resistor. When the output goes towards ground, the tube current is higher too, since it sees both current from the supply resistor and the load current coming the other way (using charge from the capacitor). It is easiest to visualize that if you think of the capacitor as acting like a low-capacity battery that's charged to match the resting voltages in the surrounding circuit and can keep the voltage across itself constant in the short term. If the tube switched off, the rising voltage on the plate would cause charging current into through the capacitor that also flows through the load. With the tube switching to ground, the discharge current flows through the load in the other direction. The problem with that circuit for driving grid current is that current the other direction can't flow through the grid, it acting much like an ideal diode and resistor in series only passes current in one way. With more current flowing through the capacitor when the output goes positive, and only a very tiny current back through the grid bias supply resistor when it goes negative, over time the capacitor would charge up to a higher voltage. At that point the average voltage at the grid would be lower than the average voltage at the driver plate by a greater amount, the voltage across the cap. The effect is that the bias voltage becomes increasingly negative, and the grid current obtained falls to match the small amount of that can flow back through the high value bias resistor, typically hundreds of K Ohms. If the amp is over driven, that increased negative bias can go beyond cutoff, with only a portion of the positive audio peaks breaking through to where the output tube conducts. Those spikes of audio represent very severe distortion, more than what symmetrical clipping would give. Also, dips in the audio level would have the lowest level sounds muted, and the bias wouldn't go back to normal right away as the excess voltage across the cap could only be discharged by the small current that could flow through the large grid bias resistor. Ironically, when some people modify an amp working to get better low frequency response with a larger coupling cap, they make it the recovery time when overdriven even longer. That overload distortion with a bias shift and breakup/muting is called blocking distortion, and is most often seen by musicians. If the distortion event wasn't intentional, it can particularly disruptive when the amp needs a long time to recover. Usually the caps are smaller in that application. Then the distortion still occurs, available as an effect when levels are cranked, but the recovery to undistorted operation on backing off the levels is faster. Looking at things another way, grid current has an average dc value at a point in time, and you can't maintain DC through a capacitor. The current can only match what goes through the bias resistor. So for class 2 operation, either another path must be added for current flow on negative excursions, or a different circuit with no capacitor is needed where current can flow without disrupting the bias voltage. Some like the vintage method with a driver transformer (see 4-65A datasheet), but good transformers are very hard to find and expensive, and complicate life for any considering negative feedback. Sometimes driving with an op-amp can work very well and provide some excellent current sampling feedback opportunities, but especially with the lower gain triodes, getting enough voltage swing requires higher voltage chips or added buffers. With op-amps it may beneficial to run mismatched positive and negative supply voltages to get the most optimal output swing while staying within the maximum total supply voltage limit. Direct coupled drive from a cathode or transistor follower is common. A classic example of class AB2 drive and biasing with a follower is the Heathkit W6A or W6M amp. Inserting a MOSFET follower into an existing tube amp between the grid and existing circuit is very simple, but the the negative supply for the source resistor returns to needs to be beefed up compared to the usual bias supply. Many simple negative bias supplies were made for extremely light loads. Another trick is to use a low impedance adjustable negative bias supply in the normal capacitor coupled circuit, but add a fast diode or even tube diode (6AL5, 6H6, 6JU8 etc) sometimes with series resistance, across the bias resistor, cathode of diode to the grid drive. The clamping action of that diode, can either force the bias to go more positive, or at least keep it from going more negative when grid current flows. That method usually introduces some distortion since the higher output impedance of an ordinary driver sags some when added current starts to flow, but it is a way to avoid blocking distortion and see more power than class 1 under heavy drive conditions. Fixed bias generally works better than cathode bias for class 2 stages. With time for cooling at low bias currents, and the low average screen power being sufficient when the control grid has more drive, the ratio of possible music/speech output power to plate dissipation is increased. For instance in class 2 a pair of 4-65As can deliver 270 Watts. That's 95 Watts more than in AB1. The Heath amp in AB2 delivers 70 Watts with ultra-linear connected 6550's even though ultra-linear usually gives only half the power of straight pentode operation. Of course it takes some added work to get very low distortion in high power amps with very low idle currents, but when pushing power an SET does too. Class 2 will get you the peak currents to deliver more power from triodes too. To get maximum power you'll also want drive capable of swinging all the way to cutoff. Some semi-local negative feedback, like tying the secondary output tap to the output cathode, can help linearize the amp against transconductance changing with current. That's increasingly needed when trying to operate over a very wide current range for more power.
The 6BQ5/EL84 is a pretty high perveance valve, so won't benefit much in class A2, maybe a dB or two at most. Still won't hurt to try it - you might gain another dB or two lowering the load impedance if it really matters.
And be sure to think through the turn on and turn off sequences; they're critical in a direct-coupled amp.
All good fortune,
Chris
And be sure to think through the turn on and turn off sequences; they're critical in a direct-coupled amp.
All good fortune,
Chris
The 6BQ5 delivers higher currents than the 6CM6, also a 9-pin 12 Watt pentode, but the data sheet for that may still be worth a look to provide some insight into the sort of behavior seen. Curves are provided for drive levels up to 20 Volts for both triode and pentode modes. There are also curves showing grid current which is low, rising at the lower plate voltages. It is worth asking what is planned for the output transformer. If it is a leftover from pentode operation is may be worth moving an 8 Ohm load to the 16 Ohm output tap if there is one. Less compression at peak current may result in slightly less third harmonic and more second harmonic distortion since there would be less symmetry with the waveform compression seen at low currents from falling transconductance.
It can be a bad idea to try to balance off distortion from grid current and distortion from changing transconductance, even though this technique was encouraged 60 years ago. The reason is that the two curves don't match, so you get a reduction in lower orders and an increase in high orders. It may measure better but sound worse.
Yes of course I agree. In the last message I was talking about the effect of the added (assumed stiff) drive restoring a normal triode positive current peak (producing less symmetry). In an earlier post I was relating an observation of others. Although I'm certainly of the school of thought of removing distortion of the drive signal, the main point I should have made clearly is that entering the realm of grid current is better than most think even when there is some distortion. The loading increases are monotonic, which is more that we can say for screen-grid induced kinkiness distortions. And when mild voltage loading is allowed, it can be adjusted to provide soft clipping. That's more tolerable, and occurring at a higher power level, than the hard clipping we'd see with high impedance drive. So it is free power. And for those that enter mild overload occasionally, the gentle nature of it makes the perceived power even greater. That's because the average (undistorted) level/power can be turned up closer to overload since the little overload that does occasionally occur is less offensive. It's easy to get in trouble with things perceived as sweeping statements, because the design hats we must wear, like the active devices we use, have multiple personalities. Sometimes it is easier to step aside and address some things separately since there is so much and some of it is modal; small signal linear transconductance, large-signal shifting transconductance, mild overload, deep overload and recovery, transfer function with positive grid voltage, changing input resistance or reactance, core saturation, changing large signal cathode impedance modulating break frequencies of bypass capacitors, (inter)modulated junction or gate capacitance affecting high frequency phase and gain, kinks and bends at low plate voltages... it's a long list. It needs to be that way, as an overly SPICEy world with smaller and avoided boundaries is a bit too much of a fantasy. People talk as if they know how much feedback they have, like it is a constant. But a 4 to 1 change in transconductance over a large signal swing is a 12 dB gain change to a small signal riding that roller-coaster. The models don't go far enough. There is screen kinkiness. There is plate current beyond zero bias. There can be control grid current (and it is influenced by plate or screen voltage). And there's that potential twisted personality when impedances alter our reality if we let them. Giving names to some device personality quirks, I think it is fitting to call input impedance shift induced distortions Bipolar Disorder. Indeed, beyond the easily controlled slew-rate issues affecting solid state amps at high frequencies, it is mostly the non-monotonically changing wild input impedance variations of output/driver stage switching between mismatched devices (and sometimes transmitting back properties of varied external loads) that generates such offensive "transistor" distortions (audible down into the mid-range) in an earlier high-impedance high gain stage. Those impedance effects are often far worse than those from simple device gain variation change with current seen seen with stiff drive. What we have to deal with in seeing grid current is far milder than those, screen or other problems we routinely face, and that's true whether we're using class 2 as an extension of clean power or as a soft crumple zone sliding into overload. Since this thread was supposed to be about drive with a 12DW7 (the bigger half that's like a 12AU7) for drive, I won't fully detail how I implement op-amp drive. Its' modern design, but off the beaten path. Suffice to say, one can address drive stiffness, transconductance shift, robbing of plate current by the screen, screen current introducing error into cathode-current derived feedback, and still leave independent control of output impedance and damping (down to no output circuit feedback if desired). Except for knowing to provide sufficient swing, the design calculations don't even require knowing that the grid drive needed for the desired bias current is a negative voltage. No tube matching needed, and no adjustments even when changing tube types or pairing two different types. (My hat is off to Steve Jobs for stirring the desire to start with high end-goals in mind and make something different that just works) Modern creative thought is productive. Simulations are useful, but if used too closely in design they can lure us into the box of variation-intolerant designs targeting devices with tightly matched non-ideal behavior. We can avoid designing a virtual burning house of cards. Counting on distortions to complement or add up to cancel by resorting to closely matched tubes and high bias currents which get the curves to fit, can be avoided. We can do better.
Last edited:
Thanks alot for the replies!
If I wanted to try transformer coupled-
The schematic here is for a very small power amp. Could be hooked up to the grid of the el84 using the little output transformer like an interstage transformer? Would the secondary swing enough volts to be any use?
http://home.comcast.net/~stphkeri/milliAmp.pdf
I think the better choice for this would be a dissimilar triode t/T the first side dc coupled cathode follower into the second side.
At least I found out that it could be done without harm to the el84 and that a 12au7 could source enough current which is mainly what I was wondering. Thanks very much for all the great info!
If I wanted to try transformer coupled-
The schematic here is for a very small power amp. Could be hooked up to the grid of the el84 using the little output transformer like an interstage transformer? Would the secondary swing enough volts to be any use?
http://home.comcast.net/~stphkeri/milliAmp.pdf
I think the better choice for this would be a dissimilar triode t/T the first side dc coupled cathode follower into the second side.
At least I found out that it could be done without harm to the el84 and that a 12au7 could source enough current which is mainly what I was wondering. Thanks very much for all the great info!
Its not exactly clear but I meant a dissimilar nine pin triode instead of the el84 for output since I already have nine pin sockets
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.