acceptable ripple limits - versus unacceptably anal ripple reduction

[eschertron]

I've seen guidelines for ripple limits for output stages, and the comparison of what ripple means to an SE output versus PP. I'm interested in getting some ideas about what's considered acceptable, ideal, and obsessive for ripple figures in the first stage. Here's the scenario:

DC heaters on stage 1 and 2, AC elsewhere. PP output stage. Planning to put a line out in parallel with the speaker (downstream of the OT) and so desire the quietest signal path possible.

I've run Duncan's PSUD and seen ripple at the preamp stage from millivolts down to 3uv or so. How much is enough?

My first guess is that about half the ripple on the first node cap (Ra and ra in parallel) is mixed prior to the second stage, and that this ripple - getting the most amplification; from stage 2 on - is the prime contributor to the ripple hum apparent at the output. What else do I need to consider, and is my first guess about 'ripple injection' fairly close?

[kg]

i think it depends highly on the signal level.

obviously the first stage's noise contribution is going to swamp subsequent stages, in proportion to the gain.

shot noise is the real devil.

[soundguruman]

Separating the Power supply ground from the preamp ground, and attaching them at opposite ends of the chassis, is a good compromise, without excessive filtering...

[Mick Bailey]

Ripple needs to be considered in conjunction with the level of amplification and as a percentage of the DC voltage level. As the voltage to the stage increases, so the effects that contribute to hiss. So ripple expressed as a percentage (rather than an absolute figure) can effectively decrease at the expense of other noise increasing. With a well-balanced PP output stage the inherent ripple within the output stage gets cancelled out and you're rightly considering the early preamp stages.

Bear in mind that when you're taking a line out you're reducing the signal level, so what might be audible through a speaker may not be of concern when you process the line out signal.

To get ripple right down you need to be looking at Hi-Fi PSU schematics. Guitar amps don't come anywhere near the level of filtering in amps intended for music reproduction, rather than as a source of music. You don't say if your amp is to be a hi-gain monster or clean. It's questionable that you need to be obsessive about ripple in a guitar amp. Look at the many of the Fender designs and they can be whisper-quiet with mechanical hum exceeding any ripple hum.

I've found that increasing capacitance alone has diminishing returns, given other factors remaining constant.

The recourse which works exceptionally well is to apply global negative feedback. It has a remarkable effect on hum and hiss, and also extends the frequency response. You get rid of noise at the expense of a stiff, clinical sound though. You can also apply local feedback within the preamp stages.

PSUD doesn't consider ripple from (say) radiated fields inducing ripple in the filter capacitors. It's good, but practical results can vary.

[teemuk]

Acceptable ripple levels also largely depend on PSRR of gain stages. Gain stage designs with high PSRR will tolerate more ripple than gain stage designs with low PSRR.

[R.G.]

I've seen guidelines for ripple limits for output stages, and the comparison of what ripple means to an SE output versus PP. I'm interested in getting some ideas about what's considered acceptable, ideal, and obsessive for ripple figures in the first stage.

Why worry about it? The only problem with reducing ripple to essentially zero is where you have to produce a lot of power. If you want negligible ripple on small signal stage power supplies, just make it so. You can flatly regulate the ripple down to arbitrarily small values with a MOSFET and a few small parts for a single or a few stages.

This is only difficult if you restrict yourself to the parts available in the early 1960s. We have better parts now.

[eschertron]

Okay. Thanks for the feedback, and I have my homework cut out for me. For starters, I'll have to review the Mosfet Follies and http://music-electronics-forum.com/t...mosfet+follies.

I'm trying to keep hum and hiss separate in my mind, but if the flicker noise or the resistor noise is the limiting factor, then zero ripple doesn't really matter after all...

@Mick: I'm thinking medium-to-high gain (no metal monster) for this topology. Also, for the last PSU node it seems to me that if the voltage drop can be tolerated, a bigger resistor (10k or more?) can make the ripple get *really* small.

[Thinking out loud/] How big a resistor in the PSU until the thermal noise surpasses the ripple? [/Thinking out loud]

[eschertron]

Acceptable ripple levels also largely depend on PSRR of gain stages. Gain stage designs with high PSRR will tolerate more ripple than gain stage designs with low PSRR.

Hi, Teemuk. Thanks for responding. Can you give me an example of a gain stage with a high power supply rejection ratio and a counterexample of one with low rejection?

[eschertron]

Why worry about it? The only problem with reducing ripple to essentially zero is where you have to produce a lot of power. If you want negligible ripple on small signal stage power supplies, just make it so. You can flatly regulate the ripple down to arbitrarily small values with a MOSFET and a few small parts for a single or a few stages.

This is only difficult if you restrict yourself to the parts available in the early 1960s. We have better parts now.

I hope I was clear; I meant ripple in the first gain stage, not the first power supply node.

I took a gander at the mosfet follies and more... Among the drawings I've looked at there's a section that show a zener/mosfet circuit that reduces the B+ arbitrarily, but in those drawings (I think) the ripple will simply ride on top of the reduction and not be eliminated. Can you point me to the circuits you had in mind?

[Mick Bailey]

Hi, Teemuk. Thanks for responding. Can you give me an example of a gain stage with a high power supply rejection ratio and a counterexample of one with low rejection?

I've found from practical experience that both halves of a twin-triode (12AX7 or low-noise equivalent) configured in SRPP has much better characteristics for PSU ripple rejection.

Probably one of the worst is any typical triode stage from any amp - 100k plate load/1.5k bypassed cathode. But that probably illustrates that ripple isn't too much of a concern in established amp designs.

An LM317 can be floated to give a well regulated output voltage for powering preamp stages.

[eschertron]

I've found from practical experience that both halves of a twin-triode (12AX7 or low-noise equivalent) configured in SRPP has much better characteristics for PSU ripple rejection.

Configured for common-mode rejection, I presume?
I'm not sure how to do that without an inverted input signal, like a PP stage.

[Mick Bailey]

It's self-split, so can be driven conventionally. There are a few downsides in that it's low distortion until pushed, then the distortion isn't very musical. It also works best driving a fixed-impedance load. In effect it can be viewed as a transformerless low-power output stage, but I use it in clean preamp and SE driver circuits.

[Enzo]

In a typical tube amp, the first B+ node feeds the output plates through the OT. It may have a few volts of ripple. The second B+ node typically feeds the output tube screens, and maybe the PI or reverb or something. AT that point we are typically free of ripple, or close to it. The remaining B+ nodes , usually ever lower voltages, serve the preamp stages. The filter caps there are not mainly there for ripple, they are mainly used for decoupling the stages.

If I find hum in an input stage, I don't expect it will be from ripple.

A poorly executed grounding scheme might result in ripple CURRENTS causing hum, but that is not from a lack of filtration.

[eschertron]

It's self-split, so can be driven conventionally. There are a few downsides in that it's low distortion until pushed, then the distortion isn't very musical. It also works best driving a fixed-impedance load. In effect it can be viewed as a transformerless low-power output stage, but I use it in clean preamp and SE driver circuits.

What little I know of self-split stages (I'm thinking LTP) is that the resulting signals now are 180 out of phase. how do you invert one so that they can be summed?

[eschertron]

In a typical tube amp, the first B+ node feeds the output plates through the OT. It may have a few volts of ripple. The second B+ node typically feeds the output tube screens, and maybe the PI or reverb or something. AT that point we are typically free of ripple, or close to it. The remaining B+ nodes , usually ever lower voltages, serve the preamp stages. The filter caps there are not mainly there for ripple, they are mainly used for decoupling the stages.

If I find hum in an input stage, I don't expect it will be from ripple.

A poorly executed grounding scheme might result in ripple CURRENTS causing hum, but that is not from a lack of filtration.

Point well taken! I'm just beginning to grasp principles beyond the basics. I've read a lot about grounding, but don't yet have the ramifications learned as familiar concepts. Your answer goes a long way to "getting me over" a focus on PS ripple at the first preamp stage. I know there's a point at which ripple reduction just won't have any effect on the hum heard at the output. I'm still trying to put a number on that ripple voltage. Having played with Duncan's PSUD I've modeled some historical power supplies, but still want to come up with a figure that represents a desirable minimum ripple voltage at the input stage. Right now I'm crunching some numbers to compare ripple against tube and resistor noise at the input stage.

...so "for decoupling" you're saying that the primary purpose is to prevent oscillation, like motorboating?