If there is noise getting through on the output of the previous stage it will be seen on the input of the stage it is connected to, and attenuated to some degree, usually small, by the natural voltage divider created between stages, as you correctly surmised.
But the noise introduced in the previous stage is a problem to be addressed in the previous stage's design, layout, and build.
Notwithstanding, sometimes you create an environment for noise to creep in when connecting two stages together, and it is certainly possible with poor wiring layout, poor component placement, or lack of understanding of grounding networks. The key point is that currents (usually small, but not always) do flow through the ground network, and they can be a cause of noise introduction if you are not cognizant of it. For example, in a cap input power supply, the wire connected between the negative side of the rectifier and negative pin of the first cap carries the full ripple current, so you would not want to connect your first stage grid ground reference in between those points.
Maybe. But one huge star doesn't guarantee a quiet amp or remove the burden on the builder from thinking through how the ground currents flow. What if you find it's noisy? How do you debug it?
A better approach in my opinion is to utilize a star/bus grounding approach. Each stage is locally star grounded, then each stage's star ground connects to a bus (which is really just a wire stringing the local star ground connect points together). Localizing noisier ground connections on one side of the bus and quieter ground connections on the other side, graduating to quieter and quieter ground connect points from stage to stage as you move along the bus.
Something like:
NOISY <----------------------------------------------------------------> QUIET
PT center tap --- output stage --- driver/inverter stage --- voltage amp stage
I'm not saying this is a blind recipe you can follow. You still need to think through how the ground currents flow. It's a general approach that has worked well for me.
A better approach in my opinion is to utilize a star/bus grounding approach. Each stage is locally star grounded, then each stage's star ground connects to a bus (which is really just a wire stringing the local star ground connect points together). Localizing noisier ground connections on one side of the bus and quieter ground connections on the other side, graduating to quieter and quieter ground connect points from stage to stage as you move along the bus.
Something like:
NOISY <----------------------------------------------------------------> QUIET
PT center tap --- output stage --- driver/inverter stage --- voltage amp stage
I'm not saying this is a blind recipe you can follow. You still need to think through how the ground currents flow. It's a general approach that has worked well for me.
Resistance is dV/dI. If the resistor is in between input and output of an inverting amplifier, it will have on it a voltage change equal to Vin*(K+1), where K is an amplification factor of the amp. That means, current change through it caused by Vin would be Vin*(K+1)/R, so the resistance measurable at the input would be R/(K+1).
I've read that a fixed bias applied through the grid leak could be noisy for an input stage. I assumed that was because the supply would be hard to filter. Now I see, I think, that the grid leak and ground would not be referenced to a common point and so couldn't cancel noise. I also found that the potential divider composed of grid leak and output resistance of the previous stage, coming from a battery bias, attenuates the noise but does not change the reference voltage set by the battery. The evidence mounts up, but it still seems so strange to me that this is so. Apparently a reference voltage and a signal (even a noise signal) do not act the same way.
OK, reread it all. That was actually kind of therapeutic- my impression of this thread has been me being obtuse and willful and others being a bit short with me- it (almost) all seems so civil and kind and patient on a re-read.
Perhaps, in fact it is. This seems very commonsensical to me, obvious in fact, and yet it may still not be so.
Perhaps it would reflect reality better to regard the noise as an undulating common reference for everything referenced to it, in the way that the chassis and all the noise flowing in it is connected (at one point) to the ground so everything bounces up and down in unison and the noise does not get amplified.
Perhaps, a directly coupled stage with a cathode referenced to ground but no grid leak is more vulnerable to ground noise that a stage with both cathode and ground cancelling out that noise. Perhaps the same is true of the fixed grid bias stage where the grid is referenced to a different DC wire than the cathode.
Or maybe that's just the second worldview I ponder, where because the whole circuit is referenced to the same more or less undulating potential and so as far as the circuit is concerned, that's zero.
I realize I'm whirling through space very fast, with the whole planet, but since I'm referenced to the planet I seem to be sitting still. That's a weird one but I've come to accept it. It doesn't affect calculations on my scale at all.
Hopefully this clarifies exactly where my confusion lies. I really appreciate your help here- I could go round and round forever on this one, until from my perspective I seemed to be standing still!
This is the only time anyone has addressed this potential divider, but no one has yet addressed it in terms of the grid leak being the series resistance for noise from ground, and the output resistance of the previous stage being the shunt element. In that case, there is a large (grid leak) series element and a small (output resistance of previous stage) shunt element, and the attenuation would be usually large. The reference I found (I admit it, everything I "know" I learned from Valve Amplifiers) to the grid battery bias being attenuated by this same divider leaves me puzzled indeed as to why ground noise would not also be attenuated through this same divider.
Perhaps, in fact it is. This seems very commonsensical to me, obvious in fact, and yet it may still not be so.
Perhaps it would reflect reality better to regard the noise as an undulating common reference for everything referenced to it, in the way that the chassis and all the noise flowing in it is connected (at one point) to the ground so everything bounces up and down in unison and the noise does not get amplified.
Perhaps, a directly coupled stage with a cathode referenced to ground but no grid leak is more vulnerable to ground noise that a stage with both cathode and ground cancelling out that noise. Perhaps the same is true of the fixed grid bias stage where the grid is referenced to a different DC wire than the cathode.
Or maybe that's just the second worldview I ponder, where because the whole circuit is referenced to the same more or less undulating potential and so as far as the circuit is concerned, that's zero.
I realize I'm whirling through space very fast, with the whole planet, but since I'm referenced to the planet I seem to be sitting still. That's a weird one but I've come to accept it. It doesn't affect calculations on my scale at all.
Hopefully this clarifies exactly where my confusion lies. I really appreciate your help here- I could go round and round forever on this one, until from my perspective I seemed to be standing still!
The attenuation would be small. If I have a voltage divider consisting of a 100Ω resistor R1 (representing the output resistance of the stage) in series with a 1MΩ resistor R2 (representing the grid leak of the following stage), the amplitude of the signal will be reduced by a factor of R2/(R1+R2), or 0.01%
Again, kindly clarify for the rest of us, exactly would be considered "noise from ground" if the ground is as you said an “undulating potential” seen by the rest of the circuit. You can also draw an equivalent circuit of the circuit, and mark out your "series" and "shunt" resistance to show us what you are having problem with - we may all learn something new. Since you mentioned Merlin, you may want to review http://valvewizard.co.uk/Common_Gain_Stage.pdf, in particular, the section on AC load.
This has been addressed, I forget who by.tapehead ted said:
You keep going on about "ground noise". We keep telling you that by definition the signal reference point cannot have noise. Why do you keep repeating things?
Always remember, there is no such thing as a voltage. There are only potential differences between points. This is another reason why the 'ground' cannot have 'a potential' - it can only have a potential with reference to another 'ground' (and so on). You need to grasp this, or at least believe it to be true even if you can't grasp it. If not, electronics will always be a mystery.
This seems to lie at the heart of the confusion. To say 1mV of noise at a given point in the circuit, you have to specify what it is measured with respect to.
So what exactly do you have in mind here? Let's say you have a voltmeter, and you are going to measure this 1mv of noise. So you put one of the two probes onto the ground connection. Where will you put the other one, in order to make this measurement and hence give a meaning to the original statement?
Electronics is not a social popularity topic. Put "feelings" aside (mostly).
Teachers sometimes use kindness, and sometimes use a swift kick in the butt. When a pilgrim is heading way off the path, leash-yanking and student anger cuts-through the fog, the lesson is remembered better.
No one author tells all. Suggest you read more books.
Problems with popular books are, authors usually go deep in great details in topics that were hard for them to understand, and omit what was "obvious" for them. Also, they often have own preferences, and even interpret results of measurements wrongly. However, professional books, starting from physics and math, are not such a fun to read, but worth it.
OK, thank you, I'm getting it. That and the "where do you put the other probe of the voltmeter?" question. And "there's no such thing as a voltage. There are only potential differences between points."
Thank you.
There's a lot I've learned from in this thread that I haven't mentioned. I appreciate all the time others have put into it.
I wish I could post the equivalent circuit with this blamed potential divider, but I have no means to do so here at the library computer.
What I am seeing is, there's a potential divider there, series and shunt elements reversed from the usual way of looking at it (as destination and origin are different for something setting out from the base of the grid leak resistor), but it doesn't come into significant play unless grid and cathode are referenced to different potentials. In the case of a fixed-bias situation where the grid is referenced to a negative potential (relative to the cathode) to bias the valve, noise from that negative rail will be attenuated in the same way battery noise is in the example in Valve Amplifiers under "Bias Problems".
This is exactly my experience. I know those books cover to cover, some pages I've scrutinized dozens of times.
I would very much appreciate being referred to other specific books.
I first heard of my standbys right here.
Thanks again.
There is an interesting meta-question hanging over this entire discussion. The question is "Should I trust what I am 'seeing' in my mind's eye?"
Let's consider some historical precedents:
Example one: Anyone can "see" that if you stop pushing something, it stops moving. So the law of motion is obviously "Things move only when they are being pushed".
Example two: Anyone can see that lightning and lodestones have nothing in common. One is the anger of the Gods, the other is magic rocks mined from mysterious mountains.
Example three: Anyone can see that the earth is stationary, flat, and the sun moves across the sky around it. All you have to do is stand in the middle of a large open area, and then look up at the sky.
The trouble is, in all three examples, what is easy and obvious to see turns out to be utterly wrong! It turns out that things keep moving at constant velocity when there is no force acting on them. Lightning and lodestones (magnets) are both electrical phenomena caused by the tiny electric charges on subatomic particles. The earth is roughly spherical, spins on its axis, flies through space, and orbits the sun.
My point? When it comes to figuring out physical phenomena, what we "see" in our mind's eye has almost always turned out to be utterly wrong.
It turns out that the only way to actually "see" such things is to use the hard-won tools of science and mathematics. Reality is to be found in the voltmeter measurements and mathematical equations that predicted it, not in whatever mental picture we happened to hold of that particular phenomena.
Now, if we are lucky enough to grow up from childhood in an informed culture, and to be given the correct information from a young age, it is sometimes possible over a period of many years to build an internal mental picture that actually does jibe with the facts. But even this only goes so far; if you grew up tinkering with vacuum tubes and reading about electron flow, you would still experience a major wrench in your internal world-view when semiconductors and conventional current flow first showed up on a schematic.
And if you didn't first get the immersion in the (experimental and mathematical) facts, then it is virtually impossible to build a trustworthy internal mental picture. When some genius astrophysicist tells us in a NOVA documentary "Black holes have hair!", we can be 100% sure that the resulting mental picture we form will be tragically wrong, and have nothing in common with what the speaker intended, because we do not understand the complicated mathematics that is in fact the only way to understand the phenomenon.
If you want to develop a deeper understanding, stick with those difficult textbooks; work through the math; use contemporary mathematical analysis tools like LTSpice when you can; make your own measurements on your own work-bench. Don't take the internal mental pictures that show up in your head too seriously - they are nice to have, and necessary for creative inspiration, but they are rarely likely to be worth much, until you have a solid understanding of the underlying science and math.
I've tinkered with analog electronics for a very long time, and read lots of textbooks along the way. Even so, my internal mental picture still fails me when confronted with something new: for example, the concept of using an LED for cathode-biasing a small-signal triode. My mental picture was convinced the LED would increase distortion, but actual test-bench measurements by Merlin showed the opposite. Yet again, score one for reality, and zero for my internal mental picture!
-Gnobuddy
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.