From: Shea @
Date: 3/9/2002 8:42 PM
Subject: Re: How *does* self-split work?
When a triode has an unbypassed cathode resistor, then a voltage appears at the cathode that is in phase with the voltage inputted at the grid. Signals at the cathode get amplified in phase at the plate, where as signals inputted at the grid get amplified out of phase at the plate.
In your self-splitting output section, the cathodes are connected together. When a signal is applied at the grid of one tube, that in-phase signal appears at the cathode. The cathode is the same for both triodes, so that signal at the cathode is being amplified in-phase in the other triode, the one with the grounded grid. Meanwhile, the signal at the grid of the first triode is amplified out of phase.
Triodes that operate with a cathode input have their grid grounded. I'm not sure why. I have an idea, but there's no point in sharing guesswork since someone here must know for sure.
The part I don't understand is how the triodes end up with balanced outputs, and why the in-phase input in a long-tail pi generally is NOT connected straight to ground. That's something Ken Gilbert and some of our other wizards would know.
Shea
From: kg
Date: 3/9/2002 9:54 PM
Subject: here's how...
Triodes that operate with a cathode input have their grid grounded. I'm not sure why. I have an idea, but there's no point in sharing guesswork since someone here must know for sure.
they don't HAVE to have their grids grounded.. that's just the way they normally are wired. in fact, you could use both inputs (grid and cathode) as a mixer.
The part I don't understand is how the triodes end up with balanced outputs, and why the in-phase input in a long-tail pi generally is NOT connected straight to ground.
they end up balanced because they share the tail current.. as one tube turns on and hogs the current, the other must give up some.
it's the same effect as what happens in a cathode biased amp that shares a single Rk (like a vox) when you pull two output tubes.. the plate current on the other two tubes goes through the roof. you have decreased the current through the pulled tubes to zero. without their plate current, the cathode resistor is "undersized"--there's not enough current through the Rk to produce the required voltage drop that IS the bias voltage, and the remaining tubes burn themselves up.
the same thing happens in the long tailed pair... the cathode resistor is 1/2 the size you'd use for one tube, and thus, if one tube is cut off, then the current through the other is 2x what it is at idle.
the reason why it's not directly grounded is that the balance and phase splitting is most perfect if the current through the tail never changes. after all, then you'd have the scenario such that the decrease in current from one tube EXACTLY EQUALS the increase in current from the other, which means perfect balance.
to accomplish this with passive components requires a large voltage drop across a large resistance. to accomplish THIS without a -ve rail in the amp requires you to float the whole shebang above ground... the grids are no longer at 0vdc, they are some 40v or so higher, allowing you to increase the value of rk and more closely approximate a current source (well, it's a current SINK actually) in the tail. there is little or no AC signal present on the 2nd grid--it's usually grounded via a cap, and fed the requisite DC voltage via a large value resistor from the other grid.
ken
ps. one of the caveats is that you never go beyond cutoff in either section. if you do, the balance falls apart. this can be observed as an AC signal at the junction of the cathodes WRT ground, and is an indication of how well your splitter is splitting.
From: Chris B
Date: 3/10/2002 3:52 AM
Subject: Re: here's how...
they don't HAVE to have their grids grounded.. that's just the way they normally are wired. in fact, you could use both inputs (grid and cathode) as a mixer.
Kg, once again you've taught me something. I've got a Magnatone amp that has a triode stage set up like a mixer; the reverb and tremolo channel is fed to the cathode of the tube while the microphone input goes to the grid of the stage.
Chris B