okay, that last entry was just too damn long… i had to break it into parts. let’s just get right into it.
in the previous post, we ended up with two completely basic amp designs, absolutely devoid of tricks or imagination. both would work fine, as we might imagine. because i had set a few arbitrary “rules” for this experiment, it might be time to consider the import of some of them. we have a 450 volt supply for the B+. it’s a relatively simple thing to make all sorts of other less positive voltages with dropping resistors, voltage dividers, decoupling networks (caps) and even more amplifiers (regulators). with clever implementation, even negative voltages can be arranged for, just from the one supply. but all those things add to the complexity of a design. and this is to be a “basic” amp. obviously i have some nefarious motives for limiting myself so… i wanted this “overview” of amplifier arrangements to include a fair amount of hybrid techniques to do things that tubes alone could not. radical simplification has a way of flushing out interesting stuff.
vacuum tubes have more than one “input”. it is easy to forget that and treat them all the same way… as grounded cathode amplifiers, with the control grid driven by the signal. we have already demonstrated all the reasons why this is popular: you get a simple and reasonable result. but what if you aren’t that interested in reasonable?
each “input” has it’s own issues that direct and indicate design. the control grid presents, by far, the easiest set of things to deal with! that is why the standard way is standard. convenience. all the other ways to drive a multigrid tube like a 6L6 are more complicated. a low or particularly variable impedance is generally a big part of the job… both are things that old tubes are not best suited to deal with. yes. there are many things old tubes have a hard time with. but it is possible to help them out… with more modern stuff.
for example, as stephie bench has previously proven, a tube plate is a perfectly acceptable input for a signal… and the output can be taken from the grid. however, the amount of swing needed to get a very small swing into a load can be really high. and, because of the relation between mu and gain in this arrangement (they are inversely related), this mode of operation is mainly limited to low mu tubes… the lower, the better. also, the need for a particularly heavy grid reduces the choices to power tubes, small or large. stephie solved these problems by parallel operation, and with regulator triodes. clearly, something like that is possible with a 6L6GC wired in triode. but not today. she already built that stuff. i want to look for some “new” stuff.
the other “inputs”, meaning the screen grid and cathode, are very interesting. both offer a low and extremely variable impedance that will put demands upon whatever is driving them. in addition, driving the cathode will change rP, depending on operating point and the screen arrangement, and therefore the proper load. still, is there more there, there to be found?
paring down the circuits to their “bones” can really teach us something about what really matters, in terms of amplification. it helps a lot when we then want to optimize… which is a whole other can of worms.
if we were to start by looking at available stable sources of DC in the previous efforts, two immediately present themselves: the voltage at the cathode of the 6L6 isn’t going anywhere… in either the triode or tetrode arrangement. and neither should the screen, in the tetrode version… at least in the previous examples. can they be used as sources of power for other stages or functions? yes they can.
the number of solid state devices that live with a 30 to 40 volt supply is immense. there are, of course, high voltage devices as well, that could replace tubes more or less as a drop in, with some caveats… a lowly LND150 could drop into place of 1/2 a 6SL7 and save heater current right off the bat! but it’s no fun, even if it would work just fine (it would).
we can loosen up a bit with the standard idea, and then look around a bit.
a general purpose N channel JFET can be arranged in grounded source, using the cathode bias voltage as a supply. the 6L6 cathode resistor will have to be adjusted a bit to accommodate the JFET’s current. an unbypassed source resistor will provide some local feedback to linearize the JFET. a big problem immediately presents itself. since there is only 32 volts of supply, there isn’t nearly enough swing available for the triode strapped 6L6. we can’t even make a watt into the load. i know this could be so easily solved by taking the voltage from the 450 volt supply, but… that means dropping resistors and decoupling. bleh! however, if we switch to tetrode operation, things look much much better.
adding negative feedback, as well as increasing the gain in the first stage by “bypassing” some of the source resistance with a red LED puts us immediately back into the same performance as the other “basic” amps previously discussed. also, by connecting the screen to the B+, we have eliminated a resistor and a cap. it’s a bit hard on the tube, but hey, it’s a guitar amp tube! we now have 8 watts at roughly 1% THD. the nfb has allowed the amp to get nearer to the hairy edge before crapping out. still, the swing is so limited by cathode voltage here. the tube wants more swing than the JFET can deliver. of course a tube wouldn’t even play on the same level as the general purpose JFET in this application, but we should be shooting for improvement over the standard amps reference performance. not simply matching it. there are some low voltage tubes we could use here, but most of them wouldn’t come close to the JFET for gain, or headroom with just 30 volts B+.
switching the JFET’s load resistor out for a choke changes everything. now the stage can swing above the cathode and even more swing into the load is available at lower distortion. not bad! 10 watts at 0.2% THD. this could also be optimized in a number of ways. better JFETs, harmonic distortion canceling, etc. one practical note about the choke. this can be an old output trans from a cheap pp amp, such as fisher, or eico or even fender. wire all the taps in series to max the inductance… there is enough slop in those old trannies to have high enough inductance for a milliamp or two. this amp can be built with garbage. or, call or email dave slagle if you are in the states. he will hook you up with a royal choke. here in europe i like christof kraus – silvercore.
those of you who have read “sound practices” may remember an amp i did way back called the dinosaur. in the original version, the first stage was a 2SK30, with its supply taken from a 50 cathode. there was a a 6SN7 in between giving both gain and buffering, driving the 50. it was a great amp, and i loved the mix of 30’s stuff with 70’s japanese JFETs. in any case, it was fun and irreverent. i promise you that this works just fine. and “there’s more meat on them potatoes”, as mike matthews used to say. this idea can be extended to stacked or cascoded stages too.
okay, now i feel loosened up.
screen driven power tubes for audio, have a mixed history. whereas there is a long long history of radio frequency use of screen driven amplifiers, david berning is given credit for doing it first in hifi. i am not sure why he was allowed that patent, but there is little doubt that the first commercial audio amp that used the technique was his. and very few have played with it since. one should ask why. it is more complex than the simpler standard arrangement, but not so much more. at the minimum, all you need is a buffer. a cathode follower, or an emitter/source follower, ahead of the screen and a completely ordinary amp follows.
in fact, we can simply borrow a page from the standard amplifier, and you get the circuit above. berning did it with pp and BJTs. the 6L6 is not the ideal choice for this mode of operation, although it isn’t shabby either. the screen has a much bigger influence on the current going through a tetrode or pentode. that is the whole point, in fact. sweep tubes (horizontal or vertical amplifiers) lend themselves fabulously to screen drive because of the high peak currents available in the cathodes of these tubes (a 6KG6 has an amp to dump, compared with the measley 120mA of a 6L6GC). also, sweep tubes are often designed to work well with low screen voltages… handy for hybrid. still this is in no way an obstacle for us right now. as you can see the amp above compares very well with the others. and yet, this is nothing new. berning (and nearly any screen modulated amp or oscillator or mixer from 50 years ago..) has done this too. and it has too many parts for this current experiment too. bleh, i say! (i have already published a screen drive amp similar to this, but all tube, in “sound practices” quite a while ago)
but isn’t there something else to explore?
if we think about the impedance of the screen as a load in itself, another possibility presents itself which has little to do with berning’s method, although it drives the screen. a PMOSFET can be used off the B+ and biased so as to match the operating point of our choice. the above amp is very interesting and i can’t say i have seen it before. although that means very little. some ham or japanese hacker has tried this i’ll bet. and never applied for a patent since it’s too close to all the stuff done before we were all born. except of course, for the PMOS. this is something that is harder to get away with using vacuum tubes. a sweep tube could do it… but not as easily. the trick here is that the screen draws a fairly nonlinear current through the course of a signal excursion. but what happens at the plate can be quite clean, well before any error correction is applied (very convenient). a scope on the screen under signal can be confusing as it can look like hell there, but things look very nice at the output. the above example is not meant to be definitive, but point in an interesting direction. it isn’t shabby either, compared to the others…
one important note: rP is significantly changed in this mode of operation. something a little different than the more common screen drive approach. in this case, the 5K transformer will need to be strapped for 10K (put the 8 ohm load on the 4 ohm tap, or 16 on the 8…). more about this later, as it leads one to an interesting insight into rP. this will also come up in the next approach… driving the cathode. all in all, this is an interesting circuit.
okay next, we can drive the cathode two several ways, but generally grounded grid operation is going to be series or shunt. adding an NMOSFET turns the 6L6 into a “super pentode” or a “super cascode”. hahahaha! but not a bad thing at all. it is clear this is a good thing. should be good for something. there are many ways to bias the MOS, not only this way.
and last for today is the PMOSFET version of the same thing… this is a little different. can you see why?
i am done for today. more soon.
One thought on “but for the kindness of strangers (angels part two)”