Attenuation Between Gain Stages - Questions

[exclamationmark]

To start my foray into tube amplification, I've decided to start 'making' (more like experimenting with...) a low voltage rack pre-amp based on some dirt cheap 12U7 tubes (basically a low voltage 12au7, with a gain of 20~ and a plate max of 30v).

Sort of like an upscale distortion pedal so I can experiment with proper pre-amp topologies and the sort without killing myself, and hopefully get something I can plug into a power amp without blowing it up

Anyway... I have a high gain design in mind and was just wondering about the attenuation between gain stages, because I imagine with a 12ax7 under clipping you'd have 200v p-p flying off the plate. Knowing that it only takes roughly 4v p-p depending on where you bias to clip a 12ax7, I know some attenuation is going to occur, but my question is, how and how much for a 'high gain' amplifier? (think Mesa or ENGL's)

My second conundrum is, are more gain stages a 'replacement' for a harder clipping, higher gain triode. Realistically, the gain is only going to be about 7-10 for what I plan to run the 12U7's at, and given I have no idea about cascading gain stages, would this still sufficiently clip following gain stages?

Hypothetically, if you cascaded a hundred gain stages together without them ever coming close to clipping through attenuation, the THD would be huge, and it would certainly sound distorted, but what would happen to the dynamic level? Would it simply end up with a lack of sustain?... At least in my head this is how it would appear, which leads back to my second question.

My third and final question pertains to voltage dividers. I know that this is usually how stages are attenuated. Usually I see the grid stopper is connected after the grid leak resistor, but what would happen if it were connected before it (ie both soldered directly to the grid). I know it would form a voltage divider, but my question is, would it work? And in the case that it actually did work, how would this affect miller capacitance/coupling capacitor calculations?

Phew. That's all the questions for now. Hopefully I shouldn't have to ask that many more when Merlin's preamp book finally arrives

[defaced]

I know some attenuation is going to occur, but my question is, how and how much for a 'high gain' amplifier? (think Mesa or ENGL's)

Somewhere around 50% is common. 470k voltage dividers a common but far from the rule. Your best bet is to find schematics of popular high gain amps and study them. The Soldano SLO 100 is kinda the grand daddy of the 4 stage high gain preamp design. It's a good starting block because alot of amps borrowed form it's design (two channel Dual Recto, Peavey 5150, etc.)

are more gain stages a 'replacement' for a harder clipping, higher gain triode.

More gain stages is for tonal complexity. Yes, you get more clipping as you add stages, but the clipping is more pleasant and musical. You could probably get just as much distortion out of a stage or two, but it would be very harsh. As for 100 stages, you would probably lose dynamics entirely, and probably also have an amp that is so unstable it'd explode as soon as you hit it with B+. Back to the SLO example, 4 stages and a CF driven tone stack is a very well established topology, and a great one to learn from because it's so widely used. There are also alot of tonal possibilities once you start doing things like local negative feedback, parallel stages, and adding tone shaping options.

My third and final question pertains to voltage dividers. I know that this is usually how stages are attenuated. Usually I see the grid stopper is connected after the grid leak resistor, but what would happen if it were connected before it (ie both soldered directly to the grid). I know it would form a voltage divider, but my question is, would it work? And in the case that it actually did work, how would this affect miller capacitance/coupling capacitor calculations?

Try this on for size. I had the same question once.
http://music-electronics-forum.com/t16346/

[Wilder Amplification]

The volume pot is essentially an adjustable voltage divider on the grid of a driven stage so yes a fixed voltage divider between two stages would work.

Maximum available swing is determined by B+ voltage and the load resistor value. The peak-peak output swing of a single ended stage cannot exceed B+. Of course due to the plate characteristic curves of the valve the plate voltage cannot drop quite to zero (given a B+ voltage of roughly 270V, the "Normal" channel stage on a plexi would drop to a max of about 85V on the negative swing according to the load line) so your available peak-peak swing will actually be somewhat lower than B+. This is your "headroom". B+ and the load resistor value determines it. Higher load resistor values will increase the available swing as the plate voltage will be able to pull down lower with a higher value load resistor.

Stage gain determines the "input sensitivity" of the stage (i.e. how much input signal is required to swing the stage to max available swing). It does not determine "headroom". However, as a byproduct of the characteristic curves, increasing gain via increasing the load resistor value will yield a lower plate drop on the negative swing, which will increase available output swing.

[exclamationmark]

Maximum available swing is determined by B+ voltage and the load resistor value. The peak-peak output swing of a single ended stage cannot exceed B+. Of course due to the plate characteristic curves of the valve the plate voltage cannot drop quite to zero (given a B+ voltage of roughly 270V, the "Normal" channel stage on a plexi would drop to a max of about 85V on the negative swing according to the load line) so your available peak-peak swing will actually be somewhat lower than B+. This is your "headroom". B+ and the load resistor value determines it. Higher load resistor values will increase the available swing as the plate voltage will be able to pull down lower with a higher value load resistor.

Yes, but isn't this still something like a 185v p-p signal coming off the plate into the grid of the next stage when it is clipped? I realize it's only typically about 90% of this value because of the voltage divider formed by the grid leak and output impedance of the previous stage but it still seems ridiculously high to me :S...

I took a gander at some pre-amp schematics of some high gain amps, and I sort of understand where they are coming from but the attenuation between stages doesn't seem to be that great. The similarities between the 5150 and the SLO-100 sort of gave me an idea though. They have both a ridiculously cold biased 3rd stage which I would assume would give a higher input sensitivity (ie, more clipping) which I can implement in my own design.

The more I think about this stuff the more I get confused, haha. I think I might just start experimenting to get some experience on changing what values does what in my circuit. Parts should be arriving soon

[defaced]

They have both a ridiculously cold biased 3rd stage which I would assume would give a higher input sensitivity (ie, more clipping) which I can implement in my own design.

Yea, I forgot about the cold clipping stage. It's kinda important. Like in almost every design important. Se let me rephrase what I said earlier, 4 stages, 1 is cold biased (usually the 3rd), CF driven tone stack. For a reference, as you know the SLO uses 39k there, and maybe for a reference low end the Fryette Deliverance uses a partially bypassed 10k. 2.2k||1u in series with 6.8k. The cold biased stage is ALOT of the character of the circuit. If you really want to play with high gain circuits, check out the SLO Clone forum.

The more I think about this stuff the more I get confused, haha. I think I might just start experimenting to get some experience on changing what values does what in my circuit. Parts should be arriving soon

Welcome to la-la land. You're right, building will teach you alot more than talking/reading. Plagiarize-build-play-read-tweak-repeat. I'd say you're probably far enough along on the reading front to get some good info out of building. As a heads up, watch your layout and ground scheme. These circuits aren't as forgiving of mistakes in those two areas.

[Wilder Amplification]

Yes, but isn't this still something like a 185v p-p signal coming off the plate into the grid of the next stage when it is clipped? I realize it's only typically about 90% of this value because of the voltage divider formed by the grid leak and output impedance of the previous stage but it still seems ridiculously high to me :S...

In the case of that particular stage config it would actually clip a bit lower than that...around 125Vpk-pk. Once you drive it harder than that it will drive the grid positive and "clamp" the input signal, which will result in a clipped negative swing at the output.

Assuming you drove it with a 4Vpk-pk signal and the bias were -1V relative to cathode, this would indeed be the case.

However, since a guitar pickup puts out nowhere near that amount the first stage will more than likely never see clipping unless you boost the front end. But it will clip the crap out of the stage that it's driving if you crank the volume control. Most non master volume amps never see this as they'd be WAY loud when approaching the onset of clipping the next stage. But amps with master volumes are usually made for the purpose of clipping the preamp at manageable volumes anyway.

If you're looking at old designs these were never designed with distortion in mind so of course they weren't concerned with large amounts of inter stage attenuation as they never thought anyone would ever crank them that much. But on a circuit that is designed primarily for high gain it's imperative that you use not only more inter stage attenuation, but you'll probably want to attenuate the low frequencies more than the high frequencies in the early stages so you don't get bass distortion, which causes the "farty low end" sound when driving it super hard.