EL34 in class AB2 triode mode push-pull?
I understand that considerably more power can be had by allowing triodes in push-pull to go into positive g1 voltage and draw grid current (class AB2). For instance, Norman Koren did this with triode-strapped 6550s in "The Emperor's New Amplifier." However, I've never seen class AB2 advocated for triode-strapped EL34s. Is there some specific reason for this, or is it just something that no-one's ever bothered to do?
No reply yet to this question but I'm still interested in people's opinions. I have the necessary components to insert a direct-coupled cathode follower between the driver and OP stages of my amp but I don't want to do all that work just, to find out that the EL34 (triode-strapped) isn't capable of operatmg like this.
You need to first look at the triode plate curves and see whats going on...
I believe you should have no problem doing this with EL34...
Just keep in mind that to properly extend up into AB2 region, you need to readjust the load-line that you were using in AB1...
Make sure to figure your peak grid current and properly calculate the drive impedance needed for good bandwidth...
I have had good success using 6V6 as a cathode follower as a driver tube.... 1/gm ;)
Why on earth would you want to use AB2 with EL34 ? You would be far better off using EL34 in class A1 or even AB1 where a good 15 watts is available into a 5.5k a-a load . AB2 will get you another 3dB but distortion will rise significantly when grid current starts being drawn , also your driver would probably need to be another EL34 . Not worth it really , if you need more power use the EL34 strapped as UL or pentode
There is method in my madness and it's based on Norman Koren's "The Emorer's New Amplifier" article. He built a PP amp using 4x6650 in triode mode, class AB2. He claims it is sonically excellent and more linear than UL. His enthusiasm has rubbed off on me to a certain extent, because I would like to get the extra power without incurring the pentode's higher order harmonic distortion and high OP impedance.
Modelled in LTspice, the AB2 arrangement seems to be capable of delivering up to 30w before cutoff, which is twice the power of AB1. I don't intend driving it to high such levels on a continuous basis but I would like to have the headroom. It wouldn't cost me anything except effort, since I already have the necessary parts to insert a cf driver where needed. I reckon a 6SN7 would have enough grunt for this.
I have already tried a pure pentode arrangement but I couldn't get satisfactory bass (too boomy), which I suspect is because I couldn't pile on enough NFB to increase the damping factor to a realistic level. UL is out, unless I invest in a new pair of OPTs, which I'd like to do when I can afford it (maybe Lundahl).
I'll attempt to attach the LTspice schematic that I used to model the planned amp, which is only for one channel. Please note that a number of current sinks are included to represent the load impose by the other channel.
If you're going anywhere near drawing grid current , you'll have to lose those grid stoppers
Starting with the "obvious": When the grid goes positive, its impedance drops from infinity to about 1K ohms very abruptly. If the driver's gain falls off at 1K load, distortion is large. As a first approximation, driver source impedance should be zero. Actually, since Gm is rising and Rg falls somewhat slowly, lowest THD may be with a source of 50Ω-100Ω.
Another way to look at it: negative-grid tube is an FET. But positive-grid, it is a BJT: plate current is approximately proportional to grid current, the "Beta" starts at not much more than 10 and soon falls close to unity. It is entirely possible to push a tube with more grid current than plate current. That actually works in RF where tuned circuits transform impedances and store energy. In untuned amplifiers, big grid current is wasteful and there is usually a better way.
The implication of a low "Beta" is that for maximum power and maximum efficiency you should run the output stage at HIGH voltage, and the driver from a lower separate supply.
BTW: most SPICE tube models have very poor modeling of grid current. Some omit it entirely. Some put in a silicon diode and a 1K resistor: this gives you a flat-topped hint if you thought you were not running grid current, but is not a good approximation of the real performance.
It is very hard to make a 100Ω source with tubes. (And you sure need to lose the 1K grid resistors.
I've seen it done with cathode follower 6SN7 but with both sides coupled with a center-tapped choke, which doubles the drive current and halves the output impedance (since only one of the grids is low-Z at a time). And that was a high-power P.A. application where a little mud on peaks was a fair trade for a lower price (or higher profit).
A very traditional way to go AB2, back when we HAD to slap grid current to get to a Watt, was to use a third "output tube" as the driver, working through a 3:1CT step-down transformer.
Trying to drive 1K loads to 10V-30V, i.e. 10mA-30mA peak drive current, from a 300V supply, through tubes of ~6K plate impedance, is doing things the hard way.
If an electrode passes current it has a dissipation. If you exceed rated dissipation, the electrode gets in trouble: often it warps, possibly shorting against another electrode. By omission, the G1 dissipation on all the 5W-40W audio power tubes is "zero" (exception: 6L6 has an AB2 suggested condition, but no explicit G1 rating validating a warranty claim if it melts). Obviously the true capacity is more than zero. Geometry suggests something like 1/1,000 of the plate rating. For a pentode, the G2 rating divided by the G2/G1 Mu might be the gross-overheat point, but the close G1-K spacing says that G1 might short to the cathode before that (G2 has more space to warp without hitting anything).
> Norman Koren's ... claims it is sonically excellent
He may be right. It is hard to know what another person hears. Though I do respect Koren's writings, I take all such descriptions with a grain of salt.
> I would like to have the headroom.
I've made some money with audio. Usually if X is not enough power, X+3dB is still not enough power.
+3dB means double the amplifier, double the weight, nearly double the cost. Running AB2 mostly does not change that. You don't need double the output tubes, but you nearly double the power and output core weights, plus you need that super-driver.
Ah, but what is the worst that can happen? You build it, it isn't sweet or isn't reliable, you re-make it as a conventional amp. No big deal. And the best-case is: it works sweet and well, you come back here and rub our noses in it.
Just don't put much faith in the models. You have to build it and see. While I doubt you have any worry about grid-melt with unclipped speech/music, be cautious in full-power testing. Have strategic burn-out resistors to protect the transformers against tube shorts.
Help - Please explain
Its stated by a few folks above (whose opinions I've come to respect) that runing a triode into grid current significantly increases distortion and that accords with what I've read in Morgan Jones and similar texts.
Then we look at some of the SET designs (2A3, 300B, 845 etc) around the place and the common theme seems to be that you need a decent driver stage to push a bit of grid current into the DHT - so why don't they sound rubish - or do they.
Are we to conclude that any decent SET is strictly Class A1 and doesn't transit into A2 at higher levels?
I'm with Ray on this. If a Class A2 SET sound OK why not Class AB2 PP?
Can anyone enlighten me? I feel my education is incomplete. If I had to take a WAG based upon the above I'd have to conclude that Class A2 SETs probably DON'T sound OK - any comments????.
I had a similar problem to Ray - but I tackled it by using parallel pairs of triode strapped EL34s to get O/P Power up to 40W without recourse to grid current.
edit: Direct-coupling is just about mandatory. The other possibility is using a step-down transformer with a high-voltage driver.
100 ohms isn't that hard, 10mmhos will do it. 6AU6 even, at enough Ip. 6CL6 or 12BY7 (or any number of similar frame grid tubes *COUGH*7KY6*cough cough*) will provide more oomph.
The curves don't lie, class 2 isn't a bad thing. Look at 6SN7 for instance, the GE curves show some +35V (it would seem I drew on my copy, extrapolating curves out to +60V :P ), to which it draws 150mA at a mere 45 plate volts! (And grid current is only one third of the cathode current, about 75mA, not bad for such low Vp and high Vg!) Well anyway, look at the spacing of the curves in this graph: very slight upward curve (indicating decrease of Rp, usually coincident with increasing Gm), greater spacing at higher currents indicating higher Gm (peaking around 200mA; it appears to fall off ever so slightly at the end of the Vg = 35V curve, suggesting full emission (saturation) at maybe Ik = 0.5 to 1A, and at a guess, Vg = 125V, Vp = 500V, for 45W grid dissipation and 350W plate dissipation :devilr:).
Since you've got that knee there, you load it exactly like a tetrode, sliding the low-Vp end right of the loadline right into that pocket. RL = 3Rp goes out the window; we have no use of such archaic rules of thumb. You're welcome to draw a line and see just how linear it is! (While you're at it, check 6V6 (triode or pentode mode) at +160Vp, zero bias class A2...)
So no, the problem isn't the output tube; it must be driving it solidly. Tubes aren't good because they have low Gm, you could use a FET I suppose, but weshallnotspeakanymoreofthis! ;)
You can use a transformer, but that sucks. You can use CF triodes, but those suck. You can use SS, but that sucks!
FYI, one way around the drive Zo problem might be local shunt feedback to a driving amp stage. Plus everything PRR mentioned...
Have fun engineering the circuit conditions, if you wish to do it from scratch. :)
|All times are GMT. The time now is 12:26 AM.|
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright ©1999-2014 diyAudio