Any reason not to use smaller grid stopper at input?

[jpfamps]

Here is a simulation using a pickup model based on a real unbranded humbucker I measured. The guitar has a 500k pot turned up full, and a 10k grid stopper is used with 100pF Miller capacitance. Now, let's suppose someone increased the grid capacitance to a full 1nF, which is far more than anyone would choose in reality.

Here's the response with zero length of cable- there is clearly a big difference between the 100pF and 1nF case.



But now look at the case with a mere 4m of guitar cable added. In creasing the grid capacitance by a full factor of ten only moves the resonant peak by a factor of about 1.8. So I think I am fairly justified in saying that with a 10k stopper you can add a lot of grid capacitance but the response is still dominated by the cable.

Thanks for doing some calculations.

This is what is suspected was the case as the grid stopper will isolate the input capacitance of the valve stage from the guitar.

[pdf64]

How does a 1nF grid - ground cap impact the frequency response at lower settings of the guitar vol control (eg halfway for a regular 10% audio taper, so R1 450k, R2 50k)?
How about for a 4n7F cap originally proposed?

Also, just to pick up on a different point,
'And what about the impedance of the driver? The Legend 125 may show a sensitivity peak at 2.5kHz, but the impedance has doubled so you're only getting a quarter of the power into it (assuming a valve amp)...'
Shouldn't that be assuming a voltage amp (or something) rather than a valve amp? Valve power amps tending to have a fair degree of output impedance which acts to track power output to the load impedance; more so in the case of no global negative feedback such as an AC30 (or 5F6A derived amp with the presence turned up).
Thanks all for a really interesting thread!
Pete

[J M Fahey]

Dear Merlin, your very simulation proves what I'm saying, thanks.

1) You are showing that the original pickup peak at 3500 Hz, which is close to the upper range of guitar frequencies (I mean those that can actually pass through the speaker response) and so will be perceived as "bright/chimey/jangly/sparkly" (pick one or more) now goes down to 2 KHz, in the middle of the speaker peak, in an area where our ears are *very* sensitive, and will be perceived as "tinny/nasty/abrasive" , you name it.

2) I beg you to *please* re-simulate with the .0047 cap suggested earlier, and about which you said

the input capacitance can be as big as you like without greatly loading the pickup.

You are showing that "just" 1000 pF varies completely the sound and quality of the pickup, how can you dismiss a value 5 times higher?.
That .0047pF cap will lower the peak by more than one octave.
I bet *that* is noticeable.

3) With all due respect to Mr Aiken, don't know how he got that Marshall cabinet response.
Placing a speaker in a cabinet influences *low* frequency response but not mid-high response.
Most it can do besides that, if poorly made, is to add some low-mid hump , which is perceived as "boxy" sound.

High frequency peaks, not only are not padded, but exacerbated because of beaming and phase cancellation/reinforcement.
Multiple equal speakers in a box cause comb filtering which worsens its response.

I beg you to please walk across a full tilt driven 4x12" speaker like you mention, say 5 to 10 meters in front of it (so you listen to what the audience does) and tell me what you hear.
You will hear a phasing effect, where when you move your head even a foot the sound changes, and the end result will be peaky, buzzy, perforating.

I repeat, multiple similar speakers in a cabinet worsen the frequency response peaks and narrow throughs (because of phase shifting), *never* "smoothen" it.

And please let me trust more frequency response curves measured by respected companies under controlled Lab conditions with the best equipment available than "homemade" unspecified measurements.

4) You found lowering mid-low "hiss" most important, I said that high frequency hiss was more annoying .
OK, besides the increased sensitivity of human hearing at mid-high frequencies, let's check the frequency response, imbalances and peaks, if any, for *many*classic guitar speakers, the most popular ones, so you don't acuse me of focusing on a single one which by sheer chance proves my point.

Let me explain how to "read" a speaker frequency response.
After that, you'll have a reasonably close idea of how a speaker will sound, even before hearing it.
Please click graphs to see them actual size, to see details.
Although looking at them "small", "all together" as on this page, helps see the *great* similarity between them.

a) Celestion Greenback.

You'll see that the frequency response is very different from that shown in your example.
More to the point, the response between ,say, 1600Hz and 4500Hz is *way* higher than that below .
You see 3 basic peaks at 2300Hz , 3500 Hz and 4000 Hz.
The first is in the "nasty" ear sensitivity area, and reaches whopping 108dB!!!
Talk about brain perforating!!
This is made even more noticeable by the huge dip around 1400/1500 Hz which as you will see is typical of *all* guitar speakers.

Before you say "but that's a single speaker" let me add that putting it in a closed box will raise that bump you see around 150 Hz by , say, 3 dB (or 4 if totally unpadded) which is not bad at all, provides "thump", but the change will be below, say, 250 or 300 Hz at most; not higher.



b) Celestion G12T75 "the speaker most widely used by Marshall"

Response below 1KHz similar to the Greenback, but noticeably different above.
The around 1500 Hz dip is deeper, by around 3 dB; the 4kHz response is lower by around 3 dB, there's an extra peak at 5 KHz.
This explains its sound being perceived as "scooped" , somewhat buzzy, "better for Thrash Metal than anything else", compared to te Greenback that's perceived as more "airy" and "natural".



c) OK, let's cross the Atlantic. After all, there's where all this music started !!!

The real thing: Jensen C12N

*Very* good lows, generally not noticed because it's used on small open back cabinets .
I have built 4x12" closed cabinets with them (heresy !!) and the result was incredible.
Although, there *must* be some explanation for Twin's good lows response, even with those tiny open backed cabinets after all!

Back to the curve, it shows its strong high frequency response, with the (yea, I know, I always repeat the same, maybe there's some fact behind it) classic (by now) peak at 2 KHz reaching an incredible over 110dB (almost out of the graph itself).
It also shows an important 108 dB peak at 3500 Hz, by the way close to a typical peak on Humbuckers.
Anybody played a Les Paul through a Twin?
Ever wondered at the bright, full yet punchy and articulated sound?
Well, now you know
And *still* the around 1500Hz strong dip, which we are starting to see on every 12" guitar speaker.
Is there any guitar speaker without that dip? (or much smaller)
Yes, EVM12L , JBLs, etc. Although they are not really "guitar speakers" but PA ones.



d) Let's not forget the very popular Eminences.
The earlier ones were Jensen type (no wonder, they aimed at Jensen Customers, mainly Fender), now they make many "British" type too (and many others, including EV wannabes).
Frequency curves very similar to earlier ones, specially Jensens.

Legend 125 . A *wide* peak from 1800 to 3400 Hz , accounting for its classic bright American sound, quite shy mids and lows. (as expected).



Legend 121 . Similar to its brother, but with larger (heavier) voice coil which accounts for a narrower 2300Hz peak, and a *huge* 1400 Hz dip. Very audible, like you set that band fully down in a graphic equalizer. Not as chimey as the 121, but perceived as more agressive.



Well, I guess that the similarities are clear enough by now.
Please notice that most are quite similar below 1 KHz, and differ a lot from , say, 1500 to 4000 or 5000 Hz.
*There's* where most speaker "flavor" differences are perceived so it can't be sustained that

You can argue that that is where the 'sparkle' is, but it's not where most of the sound is. 4kHz is about three harmonics above even the highest note on the guitar! Most of the energy from the guitar is well below 1kHz.

That's exactly the point!! *Harmonics* are what let us distinguish one sound from another and account for perceived (and hotly debated) differences between pickups, speakers, wood type , etc.
Speaking only of fundamentals means we can't distinguish a guitar fom a piano, for example.

And as a final point, what's the relation of this lengthy discussion with the original question?
That messing with brightness, harmonics, frequency response beyond what's strictly necessary to kill RF interference is *bad*.
Adding more capacitance than the bare minimum can and will mess with the sound.

[Alan0354]

A gain of 40 per stage is abnormally low- typical 12AX7 will give you a gain of closer to 60.
The 4th root of 1 million is 31.6, so four 12AX7s with about 0.5 attenuation between stages will give you a gain of 1 million, which is entirely possible, especially if you include the gain of the phase inverter on top of that.

I only chose 1 million as a convenient number- do the same calculation with a gain of 250000 if you prefer; you've still got a 1V noise signal at the power valve, which is still a lot! You can argue about the minutiae of the amp as much as you like, it doesn't change my point: the noise floor is what we care about, not the SNR, and reducing the grid stopper can help with that. Plenty of people have actually observed the improvement in real life, so I'm not sure why you're determined to oppose me on this one..?

I am not an expert on all the different amp schematic. I just took a quick look at a few schematics of the famous cascaded gain amp like Peavey 5150, Mesa Dual Caliber, Dual Rect etc. Even you take gain of 60 per stage, they don't have your kind of gain. You'll find they have divided by 10, intentionally lower the gain of a stage by cathode resistor of 39K and not degeneration. That stage has only gain of 3!!!! Then you left out the tone circuit that easily divided by 10. You look at those high gain amp, they put so much shunt capacitors along the way you'll be lucky to get over 2KHz BW. Don't believe me, try crank up the amp and lower the volume on the guitar and see how ugly the sound is

The most important thing is first, the guitar hum always dominant at over half volume. Second, Gain is lower than 125000 on the few amps I looked at. Three, nobody in the right mind crank everything to 10!!!! That's is just not real life. You can't control the amp like that.

Yes, you crank the amp up, either turn the guitar down all the way or not even plug the guitar in. Yes, you'll hear difference in the noise with different resistor values. But what does that mean?

I just finished my amp, had the place cleaned up to keep my wife happy. Next time when I open everything up, I'll use the signal generator, create an antenna loop at the end of the coax and see whether I can induce the RF noise. Hopefully my 110MHz sweep generator still works!!! I'll also change the 20K input resistor to a pot and and compare the noise.

[imaradiostar]

This is largely pointless. I don't know very many guitarists that plug a guitar directly into an amp. Almost everyone has at least one buffered effect between the guitar and the amp input. Those that don't usually have at least one non-true-bypass effect in series with the input, throwing off the above calculations.

It's already been proven that it makes a difference. If that difference is significant for your uses then great. If not then don't sweat it. Some commercially made amps use a smaller grid stopper. It doesn't seem like it's worth arguing.

Let's be straightforward about something else- who uses that much gain? Almost no one. It's too much.

Anyway, just observing and wanted to chime in. It seems silly to argue if people have already noticed the difference. It would take about 50 seconds with a soldering iron to make a grid stopper smaller and maybe add a cap and prove it for yourself. Let's just try it.

JT

[J M Fahey]

Dear imaradiostar. I agree.
Simulation is a powerful and useful tool, but the final test is grabbing he soldering iron and tack soldering a few caps, while somebody is busy playing.
I did just that, and my comments on the effect of different caps come straight from there.
By the way, tubes "cheat" a little in that there is always the hidden Miller capacitance.
For my experiments I used the "guitar du jour", meaning my Strat and LP "shop guitars" and anything my customers brought.
Effect was most noticeable on FET and Op Amp guitar preamps, because they have *very* small input capacitance on their own, opposite to the typical 12AX7, so whatever was there was what I added for the test.
I also use (and sell my customers) "ugly" guitar cables made out of flexible core (not solid) RG59U type coaxial cable.
*Heavy* braided shielding, low capacitance (20pF per foot) means a 12ft cable adds only 250Pf .

Dear alan0354. I think than rather what you can transmit using your generator, the problem is *where* you live.
As I mentioned earlier, I live in an interference filled place, maybe you live in a relatively clean one.
I often get called to Stadium sized Rock Festivals, because they have some RF intrusion they can't stop.
I won the bid to supply an interference free mixer for a successful show whose Tascam mixers plagued by garbled "mysterious voices" heard at random ... later it was found that it was a Police base station, on the same block but facing other street.
Also to shield a recording studio, full of mysterious bursts , chirping, etc. which followed a regular pattern, 1 minute on , 5 minutes off.
They were less than 60 meters from a Metro line and its associated high voltage high current switching.
And so on and on.
Please study this classic Shure M68 microphone mixer:
*ALL* gain stages Transistor Q1 to Q6) whether at the input or not, have their base-emitter junction bypassed by an important value capacitor.
Why? Because it's a Pro mixer designed for live radio broadcasting, and often used just a few feet from a transmitting antenna.
Not only the input, but even internal stages can get RF through the wiring and the next diode junction becomes a demodulator .... and from then on it's audio mixed with your clean signal.

[Merlinb]

2) I beg you to *please* re-simulate with the .0047 cap suggested earlier, and about which you said

the input capacitance can be as big as you like without greatly loading the pickup.

You are showing that "just" 1000 pF varies completely the sound and quality of the pickup, how can you dismiss a value 5 times higher?.

You are misquoting me. No one would use a 4.7nF cap with a 10k resistor- that's just absurd. I'm not dismissing it, I'm saying it's irrelevant because no one would use it. Heck, why don't we simulate 100uF while we're at it??

I suppose I should have qualified by saying
"with a 10k stopper the input capacitance can be as big as you like to kill RF (everything above audio band) without greatly loading the pickup."
but I thought that was implied.

I'm afraid I've lost track of what your point is. It sounds like you think the grid stopper has little or no effect on noise and that we should all stick with the sacred 68k/34k...

[loudthud]

I haven't been following this thread but just read it and have a couple of comments.

1) In the 80's I knew a bass player that could solder but did'nt have a clue about electronics. He claimed his bass amp sounded better when input grid stoppers were bypassed.

2) No one has mentioned using an inductor in place of the grid stopper. The inductance would need to be large and they can pickup hum but would have superior RF suppression.

3) There seems to be a decrease in the last few years of people complaining about interference from cell phones. My GMA phone can be heard if it is near a telephone and receives a text message, incomming call or network poll.

4) I built a variable capacitance / buffer stomp box some years ago. As you twist the knob it sounds like a wah-wah. See simplified schematic below. I never experimented with placing a pot in series with the cap to dampen the resonance. It must be the first thing in the effects chain, it works better with a short cable between the guitar and the box. Great between guitar and wireless transmitter.

[J M Fahey]

Dear Merlin.
Really what I'm trying to stretch is that *too much* capacitance kills harmonics that later simply can't be recovered, nothing beyond that.
What shocked me from the OP question was not the 10K but the .0047uF to ground.
My experiments with *a lot* of Musicians showed me that the "nobody complains about" value was around 500pF.
Of course if they are made out of 300pF cable capacitance and a 220Pf ceramic added at the input or if we have to factor in some Miller capacitance, so be it.
And I also made loudthud's experiment, only even simpler.
Try it yourself.
It's a simple 5 way switch (I really use a 6 position one, one being off or no contact) with 5 caps: 100/220/470/1000/2200Pf or if you wish, skip the 100pF and add a 4700 pF .
With short cables as to lower the number of variables plug in as many different guitars as possible and in different amps.
Very interesting.
Simulating is cool but listening is important too.
Both yield results.
Good luck

EDIT: also try a similar switch with 10/22/47/100/220K grid stoppers and another 47/100/220/470/1000 K grid resistors.
I have a full of holes L shaped sheet of aluminum where I mount whatever preamp pots, switches, jacks, etc need to be tested, rough looking but which if needed can even be carried onstage.
Many things that look/sound cool in the Lab do not work onstage, and viceversa.
Oh well .

[loudthud]

Somebody used to sell a tone control for guitars that was really a miniature 22 position switch with surface mount caps mounted on a little circuit board attached to the back. I haven't seen the ad in a while. I think they called it a "lossless" tone control.

[J M Fahey]

You gave me an idea, I'll add the 6 way switch (I have *tons* of them) to my shop guitars and bass.

[raiken]

3) With all due respect to Mr Aiken, don't know how he got that Marshall cabinet response.

If you read the article that graph was taken from (Marshall 4x12 frequency response), it clearly states the following:

"Following is a graph showing the frequency response of a late 70's slant Marshall 4x12 cabinet (checkerboard grille cloth, plastic handles) loaded with the original Celestion G12M25 "cream-back" speakers, miked with a Shure SM57 on the lower left hand speaker in the center of the cone, at a 45 degree angle towards the side of the cone. Measurements were taken with a Hewlett-Packard 3580A spectrum analyzer. "

That plot was not intended to be a lab-grade frequency response plot of a speaker, rather, it is a plot showing the miced cabinet response when using a standard SM57, which is useful in creating speaker emulators that sound similar to what you hear through the board. Since that was done way back in 1999, I don't remember any other details of the test setup, and I have long since retired my old HP 3580A in favor of more modern, computer-based data-acquisition setups.

I still have that Marshall cab, though, and it is by far the best-sounding cab I have ever heard. Nothing compares to it, and modern MIC "greenbacks" sound woefully inadequate by comparison, even after my three-day continuous break-in using all sorts of sweeps and pink-noise. It's funny, because I actually prefer the Chinese Vintage 30's to the Chinnese greenbacks in my 4x12s, and I've always hated V30s - I used to jokingly put a statement in my manuals that said the warranty was void if you used Vintage 30's with the amp.

In keeping with the topic at hand, here are a few real-world observations:

I use different values of grid stoppers, depending on the amp. To my ears, anything lower than 10K sounds odd and a bit harsh on non-master Marshall designs. I typically use 33K on most amps. On high-gain amps, I prefer 10K, in keeping with the philosophy of minimizing resistance values, but I also tend to do a lot of frequency-shaping on each stage of the amp, which, when done properly, can result in a very quiet, yet not dull and flat tone (like the aforementioned Mesa syndrome).

I also like to use a ferrite bead as additional RF insurance, located right on the pin of the tube. I have no empirical evidence to back this up, but it makes me feel good.

At one point while experimenting with grid stopper noise, I observed that metal film resistors were far noisier than carbon film of the same value, the opposite of what one would expect. The noise appeared to be related to some rectifier switching noise pickup or other noise that was riding on the filament string. I didn't have time to further investigate it, I just used a carbon film and planned to get back to that problem "one of these days", which of course never happened. It was an interesting anomaly, and I hope to have time to look into again, one of these days.

RF can be very insidious and sneaky, I learned this the hard way. The very first production amplifier I ever shipped went to a guy in Oregon, I believe, which is about as far from South Carolina as you can get, and therefore the most expensive shipping cost. The amp worked fine here, but it picked up radio stations at his house. I had to pay to have it shipped back to me and pay again to ship it back to him after I fixed it. The amp had 68k grid stoppers, so that wasn't an issue, but the problem turned out to be isolated input jacks with a not overly-long wire going to the ground point. The solution was to add a 0.01uF cap from the ground lug of the input jack directly to the chassis right at the jack. Even a few inches of wire can have enough inductance to render the ground useless at RF frequencies, so you must provide a very short path directly to the chassis at the input to keep RF out of your amp. That was an expensive education in RFI mitigation.

RA

[Chuck H]

It's funny, because I actually prefer the Chinese Vintage 30's to the Chinnese greenbacks in my 4x12s, and I've always hated V30s - I used to jokingly put a statement in my manuals that said the warranty was void if you used Vintage 30's with the amp.

I actually like the upper mid karang of new V30's. But it's not for everyone. But broken in they sound very different. The top end rolls off and the upper mids balance with the bottom end. Much better than other new model ceramic Celestions. But they take A LONG TIME to break in. Longer than other speakers IMHE. Long enough that I hope the one in my personal amp never blows and forces me to break in another. For out of the box classic Celestion/Marshall tone I like the G12H Anni's. Some like the Heritage model better.

Sorry, but I had to come to the defence of my favorite speaker. But this isn't a speaker discussion. So I digress.

[J M Fahey]

If you read the article that graph was taken from (Marshall 4x12 frequency response), it clearly states the following:

Dear Randall, thanks a lot for answering and providing a lot of useful data.
I did not read the article you mention because I only saw that graph, shown absolutely out of context.
I will read it with pleasure.

"Following is a graph showing the frequency response of a late 70's slant Marshall 4x12 cabinet (checkerboard grille cloth, plastic handles) loaded with the original Celestion G12M25 "cream-back" speakers, miked with a Shure SM57 on the lower left hand speaker in the center of the cone, at a 45 degree angle towards the side of the cone. Measurements were taken with a Hewlett-Packard 3580A spectrum analyzer. "

Fine, that explains it at least in (great) part.

a) you did not place the microphone at the *measurement* standard 1 meter distance (or the old US standard 4 feet) but real close to the speaker, in the typical "live reamplification* (or recording) setup.
In that case, the measurement becomes "near field" and *much* flatter. Not surprisingly that's what your graph shows.
Why? Because you minimize 2 effects which cause lots of cancellation peaks and notches:
* the fact that sound from the center and sound from the edge do not leave the cone at the same time (the cone is not rigid at all). and
* there's a lot of diffraction from the boundary between the edge of the speaker frame and the actual cabinet surface. Some Hi Fi speakers taper the cabinet surface backwards to minimize it or put a wide felt ring around the speaker, glued to the cabinet. Unheard of in guitar cabinets.

b) you placed the microphone at 45º angle.
This alone introduces a very noticeable attenuation in high frequencies, which among others would attenuate the strong 2000 to 3000 Hz peak found in most Guitar speakers.

To hear the difference, here are 2 MP3s where the first sounds "bright/jangly/chimey" amd the second sounds dull by comparison.
Both recorded with the same SM57, same amp, distance, everything, but one pointing straight (like in the factory curves) , the other at 45º, like you did.
Straight (0º on axis) http://media.soundonsound.com/sos/au...re0Degrees.mp3
Angled (45º off axis) http://media.soundonsound.com/sos/au...e45Degrees.mp3
Please switch back and forth to better hear differences, human ear has very short memory.
Both taken from a very interesting and useful page: Guitar Amp Recording

Both effects, even more if added (used at the same time) would explain the relatively mild response you show compared to the fierce aggressive one shown in all Factory Curves I quoted.

That plot was not intended to be a lab-grade frequency response plot of a speaker, rather, it is a plot showing the miced cabinet response when using a standard SM57, which is useful in creating speaker emulators that sound similar to what you hear through the board.

Cool, and a very justifiable goal.
You were trying to mike "a good sound" ... and I bet you did.

What I was referring to was a different goal.
Specifically (since we were talking about hiss as perceived by the guitar player and the public) I was interested in what's heard by a "flat mic" (human ears) , "on axis" (people in front of the speaker and looking at it) and *most* important, at a certain distance from the cabinet grille.

I think I mentioned that at , say, 8 to 15 meters away, those peaks become even bigger and worse by phasing and interaction between the *four* speakers involved.
Just walk across the speaker axis and you will hear it.
In fact, I invite you to do so, with that cabinet you mention (or any other).

Since that was done way back in 1999, I don't remember any other details of the test setup, and I have long since retired my old HP 3580A in favor of more modern, computer-based data-acquisition setups.

Thanks a lot, you have provided more than enough info, besides I'm not nitpicking here but trying to learn from people which is likewise experimenter minded

I still have that Marshall cab, though, and it is by far the best-sounding cab I have ever heard. Nothing compares to it, and modern MIC "greenbacks" sound woefully inadequate by comparison, even after my three-day continuous break-in using all sorts of sweeps and pink-noise. It's funny, because I actually prefer the Chinese Vintage 30's to the Chinnese greenbacks in my 4x12s, and I've always hated V30s - I used to jokingly put a statement in my manuals that said the warranty was void if you used Vintage 30's with the amp.

Cool.

In keeping with the topic at hand, here are a few real-world observations:

I use different values of grid stoppers, depending on the amp. To my ears, anything lower than 10K sounds odd and a bit harsh on non-master Marshall designs. I typically use 33K on most amps. On high-gain amps, I prefer 10K, in keeping with the philosophy of minimizing resistance values, but I also tend to do a lot of frequency-shaping on each stage of the amp, which, when done properly, can result in a very quiet, yet not dull and flat tone (like the aforementioned Mesa syndrome).

I also like to use a ferrite bead as additional RF insurance, located right on the pin of the tube. I have no empirical evidence to back this up, but it makes me feel good.

Well, everybody uses them now, there must be a good reason.

At one point while experimenting with grid stopper noise, I observed that metal film resistors were far noisier than carbon film of the same value, the opposite of what one would expect. The noise appeared to be related to some rectifier switching noise pickup or other noise that was riding on the filament string. I didn't have time to further investigate it, I just used a carbon film and planned to get back to that problem "one of these days", which of course never happened. It was an interesting anomaly, and I hope to have time to look into again, one of these days.

Interesting. Thanks for sharing. Will experiment.

RF can be very insidious and sneaky, I learned this the hard way. The very first production amplifier I ever shipped went to a guy in Oregon, I believe, which is about as far from South Carolina as you can get, and therefore the most expensive shipping cost. The amp worked fine here, but it picked up radio stations at his house. I had to pay to have it shipped back to me and pay again to ship it back to him after I fixed it. The amp had 68k grid stoppers, so that wasn't an issue, but the problem turned out to be isolated input jacks with a not overly-long wire going to the ground point. The solution was to add a 0.01uF cap from the ground lug of the input jack directly to the chassis right at the jack. Even a few inches of wire can have enough inductance to render the ground useless at RF frequencies, so you must provide a very short path directly to the chassis at the input to keep RF out of your amp. That was an expensive education in RFI mitigation.

You bet.

I already mentioned my fights against Police Equipment, illegal FM stations (we are chock full of them), various SMPS/Monitor noise and my own nemesis: Marine Radars (I doubt many of you have their shop less than 100 meters from anchored ships under repair or maintenance).
This is the actual corner of where my shop is:



PS: thanks again for posting such a full and comprehensive answer.

[GainFreak]

I also like to use a ferrite bead as additional RF insurance, located right on the pin of the tube. I have no empirical evidence to back this up, but it makes me feel good.

I didn't quite get that. Ar you using the bead after the resistor or how?