Ground .....theory/question

[trem]

In a "typical" guitar amp, it is common to see a triode, with a 1M resistor to ground, from its grid.
The 1M will act as a Voltage Divider/Impedance for the previous stage, and also ties the grid to ground for Cathode Biasing....sets up the grid to be negative compared to the Cathode.
Normally, if you wanted something at ground potential, you would ground it, wouldn't you.?
And that is my question, when does that gird resistor become a problem.?
1M is already a pretty big resistance.....how big could that resistor get, before it would make for a bad ground.?
With almost no grid current, is that why the 1M is not a problem for the grid to still see ground.?
Thank You

[Enzo]

Don't think of it as grounding the grid. It gives the grid a reference to ground. Otherwise the grid floats or uses the pickups as a reference.

Remember there is an AC component here - your signal - as well as th DC voltage. Your grid draws essentially no current. What does Ohm's Law tell you then? Zero current times 1M ohms still yields zero voltage dropped across the resistor, so the grid, at least as far as DC is concerned IS at ground potential. You could use larger resistors, but why? 2.2M? Sure. In grid leak bias, you see even 10M resistors. But remember, everything is in a context, so if you want a 1/2 voltage divider, you gonna install a 10M there and then another upstream?

[trem]

Thank You Enzo. Yes, Ohms Law...That is Kind Of what I was thinking in my head. You put it into words for me. A small, text book of an explanation.
Thanks Again…..I appreciate it.

[eschertron]

The radio tube engineers of the golden age measured and calculated to provide 'design' maximum ratings for grid leak resistors, knowing that equipment designers that followed would want to minimize the loading on previous stages and avoid the voltage divider effect of the resistor to ground. Too big a resistance and the tube becomes unstable, too little and the grid leak interferes with the circuit performance.

Can you change the 1M to something else and not affect the system? Probably. But the pragmatic question - for me - is more about how many 1M resistors you have in your bin and how many other 'sufficiently valued' resistors you want to stock?

[Tooboob]

The signal from the previous stage or the amp's input drives this 1M load. The driving tube has an output impedance (plate resistance) of maybe 50K, and a maybe 100K plate resistor to a supply, so the driving impedance is in the 40K range. This impedance drives the 1M grid leak resistor, forming a voltage divider, at least at frequencies of interest. And my point is ...

If you get significantly smaller than 1M on the grid leak, you'll get a significant signal loss. The cutoff frequency of the high-pass filter formed by the interstage coupling cap and the resistances above will also move higher as you decrease the resistance.

Following a clipping stage, transient response will also change.

[eschertron]

Following a clipping stage, transient response will also change.

Can you link to something that elaborates on that a little further? Sounds interesting. Thanks!

[uneumann]

Can you link to something that elaborates on that a little further? Sounds interesting. Thanks!

Long story, but as a start check Merlin's book for the discussion on blocking
and this thread has some discussion...

http://music-electronics-forum.com/t38664/

[Tooboob]

I think about this stuff a lot. I've come to the crazy conclusion that one can judge the quality of amps/distortion/whatever by whether or not it provides a richer musical prosody. That is, does the process improve the communicative power of the instrument? Now, if we clip a signal on both edges, we lose dynamics, and dynamics are important (along with pitch(es), rhythm and timbre). We do hear that the frequency response curve at the output changes as we vary the input level (the output gets even more or less "square"), but signal dynamics are a high price to pay. Through the magic of electronics, we could sample the envelope of the original signal, and apply it to the output, getting the best of both worlds, with possible fruitful processing of the extracted envelope.

The math for frequency response is fairly well developed. For linear circuits, you can put in a constant amplitude sine wave, and you can make a prediction of the output amplitude. Tubes make the transfer function non-linear for my area of interest, and suddenly the tools get weak. Simulation becomes necessary. So I stand by my claim that if all competent tube amp designers understand these phenomena, most tube amp designers are not competent. My position isn't that tube amp designer are not competent. Rather, I dispute the claim that competent tube amp designers understand these phenomena. I'm glad that Merlin talks about blocking distortion in his book, etc., and Murphy and others have been able to exploit a qualitative understanding to increase even harmonic distortion, but I think there's a hole in the understanding in most cases. I think it's easier to understand the functioning of a tube amp circuit than it is to fully understand how that function sounds. Any decent tube amp design got tweaked during development. Many resistors and capacitors have an effect on the sound. Unfortunately, many of those components work with other components to have their effect, so a quantitative understanding is not feasible. It's the sound of the result that matters anyway, so components are tweaked in a qualitative "better/worse" way. A competent designer knows the "sound" of the various components based on previous experience, and this is more valuable than being able to compute the frequency response to two digits. Of course, this approach assumes that the initial architecture is basically right, and that's a very hard thing to test qualitatively/experimentally.

For example, we see distortion "pedals" that put diodes in the feedback loop of an opamp. So folks experiment with using more than one diode type. If you're looking for preamp tube distortion, you're better off doing this (by my criteria) in two stages, with a cap analogous to a preamp blocking cap in between. More dynamics survive the journey, and they do it in a tubey sort of way. We see and hear this in some distortion pedals.

Now, my approach, based on maximizing prosody, can be used to claim that a cheesy electric organ is inferior to a piano, but the organ still works really great if you're playing "Here Comes the Bride", so I guess we have to say that we should maximize the control of prosodic elements for the desired prosodic idiom. Those kids torturing the fat string on their eight-string while they bleat a few high notes are attempting to use an idiom (and the fact that the stage is surrounded by an all-male crowd says something about their idiom). I just want to improve the prosody of their idiom - at least let them play thirds without so much mud.

[Jazz P Bass]

Interesting how you used the term 'prosody' in a musical sense.

I have always applied to linguistics.

[Tooboob]

Certainly linguistic prosody is more developed, and very interesting. There is information on musical prosody, mostly as it applies to matching lyrics to melody. The interworking of Geddy Lee (melody) and Neil Peart (lyrics) with Rush are great examples. You can hear it in "Red Barchetta" when Lee sings "A brilliant red Barchetta from a better vanished time" and "tires spitting gravel I commit my weekly crime" (same melody), or, in Limelight, "All this machinery making modern music can still be open-hearted". Or think of how pitch and accent line up with the prosody of the lyrics in "The Farmer in the Dell." I believe that musical idioms tend to follow speech idioms. You can divide the world's languages into three prosodies, which are pretty much regional, and the regions have different matching musical prosodies. Music from China sounds very strange to the western ear, and a female news anchor on a Chinese news channel, who I can presume has a lovely voice to Chinese speakers, can sound like a cat being stepped on to me.

In both the linguistic and musical cases, pitch, rhythm, timbre and volume dynamics all play a role. Step away from the idiomatic use of these tools in music, and the result sounds very "experimental".

[The Dude]

Can you link to something that elaborates on that a little further? Sounds interesting. Thanks!

No link, but when a stage is clipped, transients are the first to go. Think of it as a compressor/limiter. When a stage is driven to the point that it clips and gain is maxed out, there is no more gain to be had. Anything above a certain input level is clipped off at the output. Since transients (or in most cases, the attack of the note) are at a higher level than what follows the transient, the transient will be "cut off" or lowered to the level of maximum output.

[Tooboob]

The thread link from uneumann is the previous discussion. My point can be summarized this way - If you take a waveform that varies in volume (like the ADSR of a plucked string), and run it through a triode, clipping one extreme, the lows from the comparatively slow change in the envelope which modify the signal frequencies are filtered out by the following blocking cap (unless it is huge, and it typically isn't). This causes a low-frequency shift in the DC value toward the unclipped edge. It restores half the dynamics to the clipped edge, and takes them away from the unclipped edge. So each preamp stage takes away half the dynamics of the envelope if it clips on one edge, even if they clip the same edge. The value of the coupling cap, grid leak resistor, and output impedance of the previous stage vary this effect in a less than intuitive manner by letting more of the envelope through for faster changes in the envelope.

[Tooboob]

Yes, so if the PI and preamp are running clean, and you overdrive a push-pull output hard (or you just overdrive a push-pull output hard), dynamics get leveled. But I'm talking about clipping one edge, then going through a high-pass filter represented by the interstage coupling cap and its associated input and output impedances. A near DC shift occurs in the signal, putting half the lower-frequency dynamics back on the clipped edge, and taking them away from the un-clipped edge. So if you have back-to back inverting stages with a coupling cap in between and at the end, and you overdrive them hard, the dynamics aren't eliminated. 25% of the dynamics with a slow enough rise/fall time to get blocked by the filtering action of the blocking capacitors gets through. An overdriven push-pull output stage can block them entirely.

[Tooboob]

Ooh! Look at that. The initial pick attack gets through a stage and coupling cap that clips one edge unmolested, and survives and get attenuated in a manner similar to a grid conduction error, with the possibility to totally eliminate it with another haircut on the opposite edge. We can think of the coupling cap as interacting with the rise/fall time of the envelope. I need to do some scope captures for you folk.

[eschertron]

I need to do some scope captures for you folk.

Yes, please I am very much a visual learner.

I have Merlin's book and have read (& re-read) the section on blocking distortion. I understand the time constant that affects DC, and remember just enough college calculus to visualize how the wave can integrate positive or negative charge over time. My understanding is not intuitive, yet.