Can anyone point me in the right direction as to where I could find a good in depth description of the function of driver stages in comparison to output stages. I understand that driver stages have a resistive load and output stages have a reactive load, but what are the other design goals in a good driver stage and a good output stage. I've read that the main purpose of the driver stage is voltage amplification and the output stage is more of a current buffer? I'd really like to find more in depth information on this. I have Morgan Jones's book, some I'm looking for information supplemental to that.
The main purpose of the driver is to provide whatever drive the output requires. In many cases this is mainly voltage into a middling level impedance, but capacitance can be an issue. If the output runs into grid current (e.g. as in class AB2) then the driver needs to supply current, and hence significant power. The driver in this case might be a cathode follower. Some amps do not have a separate driver stage, but just go straight from the phase splitter.
The idea that the output is just a current buffer is more applicable to solid-state.
The idea that the output is just a current buffer is more applicable to solid-state.
The task of the driver can be quite trivial (e.g. when the OP tube is a pentode, with its high sensitivity and negligible input capacitance) or very significant (e.g. when the OP tube is a low-mu triode, with its need for a high drive voltage and its substantial input capacitance).
In extreme cases, such as when the OP tube is a cathode follower or takes grid current, the ability of the driver to fulfil the exacting demand for high signal voltage and/or current can act as the limiting factor of the whole amplifier. Inadequate driver capability is a very likely cause of distortion that is too often blamed on the OP stage.
You already have the M.J. book, and they don't come much better than that.
Not always. Cgk + Cstray + Cmiller can present a significant reactive load. You forget that at your own peril, and here's where just ESSSSSSSSS-loads of hollow state texts go way wrong: claiming that Class A*1 VTs require voltage only, but not current. Ain't so: the control grid doesn't see the true input voltage until those capacitances charge up. That takes current (always) and the faster the input voltage changes, the higher the current required. If you don't have it, you have slew rate limiting, and that sounds nasty.
A lot of complaints about disappointing performance from 300Bs almost always stems from inadequate grid drive.
I always try to figure the best guestimate for input capacitance, see what the reactance is at 30KHz, and how much current it will be pulling. Then make the static plate current of the driving stage at least five times greater than that capacitive current ("Rule of Five" -- borrowed from solid state design, works pretty good here too.) You find that a lot of the low current types (12AX7, 6SL7, 6AV6, 6SF5) just don't have what it takes to drive the grids of power finals.
Low distortion (unless doing guitar amps).
Not necessarily. Sometimes your driver stage is going to have to deal with significant current demand. That's frequently the case when designing for low-u power triode finals. These have such low voltage gains that they require some significant input voltages. As these don't have a screen sitting between the plate and control grid, the reverse transfer capacitance is rather large. Magnify that by Miller Effect, and even if you don't go out of Class A*1, you're looking at some stiff currents.
In those cases, your best bet is a source follower driver: it can put out some currents, and makes for a much easier load to whatever's connecting to it.
That makes sense, about the capacitance, and I suppose there will always be some capacitance since we don't live in a perfect world. I'll look into the "rule of five" you mentioned. It seems once again that this is a question that has many answers based on any number of variables, which is what makes this all so fun right ? 
Input capacitance is a STRONG function of output stage topology, not because of any "defects" in real-world components, but because of logical things like Miller Effect. MJ really does cover this pretty thoroughly- I'd go back and reread it in some detail, especially his analyses of the Williamson and Mullard 5-20.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.