Well, of course it matters. Anyone for class C?, otherwise known as fuzztone?😀 Or you may require the absolute maximum p-p signal swing before clipping.
But...
What I get from my reading, including this DIYaudio thread are a few confusing points, as follows...
At first glance, it appears that lots of little amps run grid current in their preamp tubes (based on the Fender, Ampeg, Magnatone/Estey, and Gibson schematics in my Jack Darr guitar amp book), but they don't, actually, or they'd sound hideous. Some of them run the tube on 130VDC, with a plate voltage of around 80v. As far as that goes, a coupe circuits run a 12AX7 with no grid bias at all. This isn't supposed to work.
Except it is - according to the design chart in the 12AX7 datasheet, with a 10M grid resistor, you use no cathode resistor, with a 240k (220k) or 510k (470k) plate resistor, and DC supplies from 90v to 300v. Where is THIS bias coming from, and does anyone talk about it in relation to guitar circuits? The same chart recommends a 100k grid resistor rather than the guitar-amp standard 1M. Does the 1M affect the bias actually seen by the tube, or is it still too small to matter? This also implies that a 10M input impedance for a piezo would have to be biased differently, no?
A couple of other designs run the tube from a 485vDC supply. This isn't supposed to work either. I suppose if you drop it all across the plate resistor and control the input, and/or operating point, so that the tube never cuts off, you could, I think, get away with it. Or is it just a "you'll get away with it if you don't abuse anything else at the same time" situation, like power tube voltage, current, and power ratings seem to be?
Power tubes supposedly only sound good when overstressed. But preamp tube high-voltage operation is recommended for better linearity, not more distortion. Does either of these have any bearing on preamp distortion? Why choose a 150v or 350v preamp supply, as long as the driver can drive the power amp? Does it matter?
Nearly everyone in the world uses 100k plate resistors (except for Ampeg and a few Gibsons, which use 470k), pretty much regardless of plate voltage, bias point, or tube function. The tube manuals suggest a variety of plate resistors, as a way of getting the desired amount of gain per stage with differing plate supply voltages, or for other design criteria. Why the love for 100k?
My tentative conclusions...
1. The 12AX7 likes at least 100v on its plate, but isn't too picky about either this or it's nominal maximum of 330v, or much of anything else, for that matter.
2. The 12AX7 develops a fair amount of self-bias, which may or may not show up in normal use. Anything higher than the bog-standard 1M input impedance should consider it, but nobody uses higher impedances so no one mentions it.
3. In most cases, the true relevance of the bias point only shows up when you analyze the whole amp's distortion characteristic.
4. The "best" distortion happens when each stage clips about ten or twelve db before the previous stage, which is difficult to ascertain when the stages are all nonlinear. Beyond that, the nature of desirable nonlinearity is mojo.
5. Ultimately, analyzing the amp's distortion character is almost completely mojo-based, with actual theory and measurements not really able to explain why one amp sounds fairly good, a nearly identical one sounds incredible, and another similar one sounds like a staff meeting of zombies.
My goal, for what it's worth, is to build a tiny guitar amp with a 12AX7 preamp, a 6CG7-based quasi-parametric midrange control, and a 6AQ5 single-ended power amp. I am considering switching the power supply from a 375v doubler to a 188v FWR, for a high-power/low-power mode, and trying to figure out how the preamps will handle this. I could regulate them to 180v, but this will limit the last preamp stage and power amp drive differently in high-power mode.
So, my final tentative conclusion is that the preamps will almost certainly function either way, so I should perhaps just go ahead and build it, and resign myself to hand-tuning the cathode resistor and bypass for each stage, until it sounds good at both voltages.
But...
What I get from my reading, including this DIYaudio thread are a few confusing points, as follows...
At first glance, it appears that lots of little amps run grid current in their preamp tubes (based on the Fender, Ampeg, Magnatone/Estey, and Gibson schematics in my Jack Darr guitar amp book), but they don't, actually, or they'd sound hideous. Some of them run the tube on 130VDC, with a plate voltage of around 80v. As far as that goes, a coupe circuits run a 12AX7 with no grid bias at all. This isn't supposed to work.
Except it is - according to the design chart in the 12AX7 datasheet, with a 10M grid resistor, you use no cathode resistor, with a 240k (220k) or 510k (470k) plate resistor, and DC supplies from 90v to 300v. Where is THIS bias coming from, and does anyone talk about it in relation to guitar circuits? The same chart recommends a 100k grid resistor rather than the guitar-amp standard 1M. Does the 1M affect the bias actually seen by the tube, or is it still too small to matter? This also implies that a 10M input impedance for a piezo would have to be biased differently, no?
A couple of other designs run the tube from a 485vDC supply. This isn't supposed to work either. I suppose if you drop it all across the plate resistor and control the input, and/or operating point, so that the tube never cuts off, you could, I think, get away with it. Or is it just a "you'll get away with it if you don't abuse anything else at the same time" situation, like power tube voltage, current, and power ratings seem to be?
Power tubes supposedly only sound good when overstressed. But preamp tube high-voltage operation is recommended for better linearity, not more distortion. Does either of these have any bearing on preamp distortion? Why choose a 150v or 350v preamp supply, as long as the driver can drive the power amp? Does it matter?
Nearly everyone in the world uses 100k plate resistors (except for Ampeg and a few Gibsons, which use 470k), pretty much regardless of plate voltage, bias point, or tube function. The tube manuals suggest a variety of plate resistors, as a way of getting the desired amount of gain per stage with differing plate supply voltages, or for other design criteria. Why the love for 100k?
My tentative conclusions...
1. The 12AX7 likes at least 100v on its plate, but isn't too picky about either this or it's nominal maximum of 330v, or much of anything else, for that matter.
2. The 12AX7 develops a fair amount of self-bias, which may or may not show up in normal use. Anything higher than the bog-standard 1M input impedance should consider it, but nobody uses higher impedances so no one mentions it.
3. In most cases, the true relevance of the bias point only shows up when you analyze the whole amp's distortion characteristic.
4. The "best" distortion happens when each stage clips about ten or twelve db before the previous stage, which is difficult to ascertain when the stages are all nonlinear. Beyond that, the nature of desirable nonlinearity is mojo.
5. Ultimately, analyzing the amp's distortion character is almost completely mojo-based, with actual theory and measurements not really able to explain why one amp sounds fairly good, a nearly identical one sounds incredible, and another similar one sounds like a staff meeting of zombies.
My goal, for what it's worth, is to build a tiny guitar amp with a 12AX7 preamp, a 6CG7-based quasi-parametric midrange control, and a 6AQ5 single-ended power amp. I am considering switching the power supply from a 375v doubler to a 188v FWR, for a high-power/low-power mode, and trying to figure out how the preamps will handle this. I could regulate them to 180v, but this will limit the last preamp stage and power amp drive differently in high-power mode.
So, my final tentative conclusion is that the preamps will almost certainly function either way, so I should perhaps just go ahead and build it, and resign myself to hand-tuning the cathode resistor and bypass for each stage, until it sounds good at both voltages.
It probably would not make that much difference with single 12AX7, but if you have more gain stages, you normally bias them hot, cold, hot, cold... so the waveform is clipped/shaped along the way, then, the bias does matter.
Without bias (even contact bias) you might as well swap the tube for a diode as one-sided rectification of the audio signal is about all you're gonna get out.
Grid current.KFW said:
Yes. Grid current bias is commonly used in low signal low hum environments, such as tape head preamps. The idea is that the cathode can be grounded so hum from heater-cathode leakage is eliminated. The grid current will create distortion with anything other than a small signal unless the source impedance is low.
The 12AX7 has a rather narrow window between grid current and grid cutoff. This because of the high mu. Low anode voltage make this window smaller. In hi-fi this can be a problem; in a guitar amp this can be an advantage?
I think I need to read up a bit more on tubes. There seems to be two different things called "grid current". There's the current that flows into the grid when a large signal forces it positive, and there is apparently electrons that pile up on the grid when the grid resistor is really high.
That begins to explain a few things, such as the Valco amp schematic I'm looking at (yes, I know, an off brand). It's first stage has a 6.8M grid resistor, a 270k plate resistor, and no cathode resistor. This will work, in the sense that the tube will amplify quite normally - until you overdrive it, when it probably will sound awful compared to a more normal input stage. A great accordion amp.
The "hot/cold/hot" idea may be useful. Without figuring out yet how to paste a schematic together from fragments, I'll describe the architecture as a:
12AX7 gain stage, driving a
tunable Wein bridge into a
6CG7 unity-gain differential amp, driving a
preamp volume control
6CG7 gain stage (lower gain and output impedance than a 12AX7), driving a
tone stack of undetermined design
12AX7 gain stage, into the
master volume, followed by the
6AQ5 output stage
I think there's enough gain in there to justify both a pair of volume controls, and some tweaking for distortion quality. And, as I mentioned earlier, I'd like to run it off both 380vdc and 190vdc without changing bias. That part, I suspect, is a crapshoot as far as getting the whole chain tuned at both voltages, but obviously there's some tuning to be done. And it seems perfectly possible in the sense that there are lots of component values that produce working amp stages at either voltage.
So, my next step is to assemble the power supply module, which implies figuring out approximate current draw at both voltages. With that up, I can start breadboarding.
That begins to explain a few things, such as the Valco amp schematic I'm looking at (yes, I know, an off brand). It's first stage has a 6.8M grid resistor, a 270k plate resistor, and no cathode resistor. This will work, in the sense that the tube will amplify quite normally - until you overdrive it, when it probably will sound awful compared to a more normal input stage. A great accordion amp.
The "hot/cold/hot" idea may be useful. Without figuring out yet how to paste a schematic together from fragments, I'll describe the architecture as a:
12AX7 gain stage, driving a
tunable Wein bridge into a
6CG7 unity-gain differential amp, driving a
preamp volume control
6CG7 gain stage (lower gain and output impedance than a 12AX7), driving a
tone stack of undetermined design
12AX7 gain stage, into the
master volume, followed by the
6AQ5 output stage
I think there's enough gain in there to justify both a pair of volume controls, and some tweaking for distortion quality. And, as I mentioned earlier, I'd like to run it off both 380vdc and 190vdc without changing bias. That part, I suspect, is a crapshoot as far as getting the whole chain tuned at both voltages, but obviously there's some tuning to be done. And it seems perfectly possible in the sense that there are lots of component values that produce working amp stages at either voltage.
So, my next step is to assemble the power supply module, which implies figuring out approximate current draw at both voltages. With that up, I can start breadboarding.
There are two different aspects to grid current, but not quite the two you have identified.KFW said:
Grid current has two contributions: electrons and positive ions. Electrons are the main part of the current when the grid is positive with respect to the cathode; this is often what people mean when they say 'grid current'. Positive ions come from gas in the valve and can lead to thermal runaway; this is also what people mean when they say 'grid current'. Usually the context makes it clear which is being talked about.
The electronic part of the current doesn't stop when the grid goes a little negative, as the hot cathode gives electrons enough energy to overcome the voltage. The current is small in this region, which is why you need a big grid resistor value to see it. Electronic gird current usually stops when the grid is more negative than somewhere around -0.5V to -1.5V - the exact value varies with valve type and from one sample to another and also changes with age. It is this small current which is used for grid leak biasing.
The large resistor (usually 10M) means that the static grid current is small, but it varies with grid voltage and this variation is not linear - hence distortion. RDH4 gives a good explanation of grid current.
Another practical point is that you have latched onto grid leak bias for your input stage as you keep referring to it. There's a reason why grid leak bias isn't used in modern tube guitar amps; it simply won't work if there's an active device (e.g. effects pedal) between the passive guitar pickup and amp. In fact many vintage amps are that originally came as grid leak bias are modded to cathode bias for practicality.
I agree that grid leak bias doesn't seem to be a good idea. If nothing else, cathode bias provides negative feedback at DC and stabilizes the bias point. I'm more concerned with the fact that it seems to affect real amps that aren't, so far as most discussion is concerned, supposed to be using it. It appears that a 1M grid resistor for a 12AX7 is right on the edge of producing enough grid leak bias to affect the tube.
For example, the midrange control I plan for the amp has a 1M tuning pot AND a 1M resistor between the grid and the cathode resistor (it's a unity-gain amp so the cathode is far above ground). It appears that this may cause the frequency pot to induce crackling, and bias shift, into the midrange amp. Enough to matter? I guess that's a breadboard question.
Most of the schematics I've seen that use grid leak bias isolate the input with a capacitor. But not all. A bunch of old Ampeg amps have a DC-coupled input stage, a 5.6M input resistor, a shorting jack that loads the grid with a 47k resistor when the bright, but not the normal, input is used, AND a cathode resistor. Do they INTEND the bright and normal inputs to bias the input stage differently? If so, why do they copy the identical input circuit in a FET-input solid state amp, where it doesn't have that effect?
A related bias question... Looking at a handful of Ampeg amps from the same era, they seem to have the same modular design, but more or less features (2 preamps, tremolo, reverb, 2 or 4 output tubes, etc.). Okay, this makes a lot of sense. But the input stages, of identical architecture and running from basically the same V+, all have different operating points. 270k/5.6k, 220k/2.2k, 100k/2.2k, 100k/4.7k, 270k+150k/5.6k (this one's a bass amp). Were they that good at voicing all these amps slightly differently? If not, what were they up to? Buying whatever resistor value was on sale?
Or an old Gibson amp with two identical input channels, except one has a 1.5k cathode resistor and the other a 1.2k. The schematic says it makes a difference of 2v (198v vs 200v) on the plates of the two halves of the tube. WHY???
For example, the midrange control I plan for the amp has a 1M tuning pot AND a 1M resistor between the grid and the cathode resistor (it's a unity-gain amp so the cathode is far above ground). It appears that this may cause the frequency pot to induce crackling, and bias shift, into the midrange amp. Enough to matter? I guess that's a breadboard question.
Most of the schematics I've seen that use grid leak bias isolate the input with a capacitor. But not all. A bunch of old Ampeg amps have a DC-coupled input stage, a 5.6M input resistor, a shorting jack that loads the grid with a 47k resistor when the bright, but not the normal, input is used, AND a cathode resistor. Do they INTEND the bright and normal inputs to bias the input stage differently? If so, why do they copy the identical input circuit in a FET-input solid state amp, where it doesn't have that effect?
A related bias question... Looking at a handful of Ampeg amps from the same era, they seem to have the same modular design, but more or less features (2 preamps, tremolo, reverb, 2 or 4 output tubes, etc.). Okay, this makes a lot of sense. But the input stages, of identical architecture and running from basically the same V+, all have different operating points. 270k/5.6k, 220k/2.2k, 100k/2.2k, 100k/4.7k, 270k+150k/5.6k (this one's a bass amp). Were they that good at voicing all these amps slightly differently? If not, what were they up to? Buying whatever resistor value was on sale?
Or an old Gibson amp with two identical input channels, except one has a 1.5k cathode resistor and the other a 1.2k. The schematic says it makes a difference of 2v (198v vs 200v) on the plates of the two halves of the tube. WHY???
perhaps when you refer to examples of something, you might give us the model number in question, so we can see the schematic you are looking at. Also, then you could refer to part numbers. None of these tube stages happen in a vacuum - so to speak. Everything happens in the context of a circuit. SO if the first stage has a triode biased to a certain point, it might make sense to have the second stage biased to allow a lot more signal level to provide headroom for the growing signal. So it would be helpful to know you refer to R2 versus R3 or whatever.
AMps back in the day also had different goals than modern amps. Today it is expected we play guitar through them, but you will find many amps of that old era with inputs labelled for guitar and for accordion. They are expecting different signal levels, and the input stage designs might be reflecting that.
Schematics sometimes show voltages, good enough. But there are two sources for those. One is to set down the planned voltage, which can vary 20% and still be OK, while the other is the measured voltage of a sample amp, which can vary widely as well. None of it is gospel, never forget that these are guitar amps, not precision NASA space probes.
And I don't mean this to sound snobby, but I suggest not looking at things with rules in mind. By that I mean, "this is not supposed to work" or "They only sound good overstressed". They fall into what I consider the never/always trap. When someone starts saying something "always" this or "never" that, I expect they will be surprised soon enough to find that it is not the case. When we LOVE the sound of an overdriven fender Deluxe, we can;t forget that overdriving the thing was the last thing on Leo Fender's mind. Read the interesting AMpeg story book. Those were designed to be hifi in sound. he most surely did not want distortion or "overtressed" power tubes.
AMps back in the day also had different goals than modern amps. Today it is expected we play guitar through them, but you will find many amps of that old era with inputs labelled for guitar and for accordion. They are expecting different signal levels, and the input stage designs might be reflecting that.
Schematics sometimes show voltages, good enough. But there are two sources for those. One is to set down the planned voltage, which can vary 20% and still be OK, while the other is the measured voltage of a sample amp, which can vary widely as well. None of it is gospel, never forget that these are guitar amps, not precision NASA space probes.
And I don't mean this to sound snobby, but I suggest not looking at things with rules in mind. By that I mean, "this is not supposed to work" or "They only sound good overstressed". They fall into what I consider the never/always trap. When someone starts saying something "always" this or "never" that, I expect they will be surprised soon enough to find that it is not the case. When we LOVE the sound of an overdriven fender Deluxe, we can;t forget that overdriving the thing was the last thing on Leo Fender's mind. Read the interesting AMpeg story book. Those were designed to be hifi in sound. he most surely did not want distortion or "overtressed" power tubes.
I think we're closer in approach than my language may suggest. I find it easier to understand how something is working outside design limits if I know the limits first. I don't take "shouldn't work" as a moral statement, merely one that the theory I'm currently using is insufficient. If something works that shouldn't, it is almost certainly my sense of "should" that's wrong, not the equipment. So then it behooves me to find out what is really happening. On the other hand, it's good to notice when something isn't working according to conventional theory - that's how I learn the unconventional theory. And nearly all theory of "good" distortion is unconventional theory.
"Amps back in the day also had different goals than modern amps. Today it is expected we play guitar through them, but you will find many amps of that old era with inputs labelled for guitar and for accordion. They are expecting different signal levels, and the input stage designs might be reflecting that."
Now you're leading somewhere. I wonder if what they weren't designing for wasn't specific GAIN, with linearity being just what the tube offered. Tubes are pretty linear with tiny signals anyways. That would explain much of the variability - one amp has a tremolo with a couple decibels of insertion loss, so its preamp is tweaked for a couple decibels more gain. Another has a more complex tone stack, This one has two channels, so there's a little loss in the channel mixing that they want to make up somewhere. And that would actually explain the Gibson - when I look closer, there is ONE difference between the channels - one has a tremolo LDR and the other does not. Hence, the need for a little bit more gain in one otherwise identical channel.
Regarding bias points, though, if I am amplifying a 100mv signal to 5v, it's still not very clear to me why to set the operating point to any particular place, aside from actually having a 5v swing available, and having a gain of 50. With a decently high V+, there's a pretty wide range of values that work for small signals. I'm also thinking it might make more sense to design backwards from the power amp, as the later stages have more constraints and so should be easier to specify.
One last thing, to hear something pretty close to how Leo thought Fender amps should sound, I listen to Speedy West and Jimmy Bryant. But what I'm after for the amp I'm building is the widest variety of useful tones - I'm not even that particular, at this point, what they are, as long as they don't all sound like Adrian Belew. I have effects boxes that do that 😀
"Amps back in the day also had different goals than modern amps. Today it is expected we play guitar through them, but you will find many amps of that old era with inputs labelled for guitar and for accordion. They are expecting different signal levels, and the input stage designs might be reflecting that."
Now you're leading somewhere. I wonder if what they weren't designing for wasn't specific GAIN, with linearity being just what the tube offered. Tubes are pretty linear with tiny signals anyways. That would explain much of the variability - one amp has a tremolo with a couple decibels of insertion loss, so its preamp is tweaked for a couple decibels more gain. Another has a more complex tone stack, This one has two channels, so there's a little loss in the channel mixing that they want to make up somewhere. And that would actually explain the Gibson - when I look closer, there is ONE difference between the channels - one has a tremolo LDR and the other does not. Hence, the need for a little bit more gain in one otherwise identical channel.
Regarding bias points, though, if I am amplifying a 100mv signal to 5v, it's still not very clear to me why to set the operating point to any particular place, aside from actually having a 5v swing available, and having a gain of 50. With a decently high V+, there's a pretty wide range of values that work for small signals. I'm also thinking it might make more sense to design backwards from the power amp, as the later stages have more constraints and so should be easier to specify.
One last thing, to hear something pretty close to how Leo thought Fender amps should sound, I listen to Speedy West and Jimmy Bryant. But what I'm after for the amp I'm building is the widest variety of useful tones - I'm not even that particular, at this point, what they are, as long as they don't all sound like Adrian Belew. I have effects boxes that do that 😀
Don't assume that everything in a commercial circuit was carefully designed after much thought. Sometimes it was. Sometimes it was just copied from another circuit without thinking. If you see something which puzzles you it could be a very clever 'deep' idea, or it could be a mistake. Mistakes can even propagate from one circuit to another.KFW said:
Different bias points will have different distortion, including different compression, and different output impedance. However, as I said above, don't assume great thought went into it: it may, it may not.
Yeah, Marshall amps were designed backwards compared to a Bassman because they were being built on the other side of the pond, right?
I recall some amplifier manufacturer copied a major manufacturer's schematic even the mistake on the schematic. And it worked, sort of.
No, they were designed the same way, the educated way to mean that Marshall was a blatant Fender ripoff.
So much so, that the input jacks side appeared "on the other side" simply because the chassis was bolted to a case *bottom*. instead of to a cabinet *top*.
Just that.🙄
Or to say in in another way: topple a Tweed Bassman flat on its back, look at the "front" panel,and you'll "see" a Marshall head.😀
So much so, that the input jacks side appeared "on the other side" simply because the chassis was bolted to a case *bottom*. instead of to a cabinet *top*.
Just that.🙄
Or to say in in another way: topple a Tweed Bassman flat on its back, look at the "front" panel,and you'll "see" a Marshall head.😀
Very high value grid resistors and no cathode resistor form a kind of bias called grid leak bias.
If you have a perfect tube connected in a circuit with plate voltage and an open circuited grid, random electrons emitted from the cathode will hit the grid and collect and build up to create a negative voltage on the grid. With no path for current flow the negative voltage will increase and eventually cut the tube off. The usual approach is to bleed off (leak) the electrons at a controlled rate through a high value resistor creating a negative voltage on the grid.
There are indeed two sources of grid current and they are of opposite polarity. Both of these are in effect without even considering the current drawn when the grid is driven positive.
There is the previously mentioned collection of electrons on the grid causing the grid to be charged negatively. There are unfortunately no perfect tubes. All tubes contain some "gas". This gas is usually air, other gasses, and various impurities. These can strike the plate and give up an electron, becoming ionized. The newly created ions now have a positive charge and sometimes glow blue or purple. When one of these ions strike the grid, it will steal an electron to return to a neutral state. This causes grid current in the other direction.
Ion type grid current is not usually an issue in small signal tubes, but can overcome the normal grid bias in output tubes that run hot. This is why you will see a maximum grid circuit resistance spec on output tubes. The grid circuit must be capable of supplying enough electrons to neutralize ion current without affecting the bias voltage.
Todays production tubes, and some later production tubes are far worse in this respect than many vintage NOS tubes. This is one reason that grid leak bias is not common any more.
Some of the original guitar amp designs were lifted right from the old tube manuals of the day. Even into the late 60's the Kustom K50 solid state amp was directly copied from the RCA transistor manual. Some guitar amps were copies of other amps, or a mishmash of other amp designs. Others, notably Fender were designed for absolute minimum cost.
Leo never thought that guitar players would embrace distortion, but it turned out that his minimalistic designs distorted in a most pleasing manner.
I posted this to another forum - just copying it here without mods to make it more relevant to the discussion. Hope it is useful.
Quote:
Grid Current is one of the most poorly understood topic and so at the risk of teaching grandma to suck eggs I've written a little essay.
Grid Current and all that – remember that, due to historic limited knowledge, current flow is defined as from positive to negative whereas in actual fact electrons flow from negative to positive.
There are 3 types of grid current
Current flowing into the grid is known as POSITIVE grid current
• When the cathode is heated a cloud of electrons forms around the cathode known as a “space charge”
• Some of these electrons gather at the grid. These electrons then flow out of the grid which is the same as saying that current flows into the grid.
• This positive grid current generates a voltage across the grid leak resistor (Rg1)
• This voltage makes the grid more negative which ADDS to the bias
• If the grid leak resistor is large enough then this positive grid current can generate the entire required bias – This is known as “Grid Leak Bias”
POSITIVE grid current is a low level phenomenon and can easily be overshadowed by NEGATIVE grid current
Current flowing out of the grid is known as NEGATIVE (or REVERSE) grid current
• Negative grid current can be caused by
1. gas ioniszation current
2. leakage current (grid to cathode)
3. grid emission (from grid being heated by the cathode, screen or anode)
• gas ionization current dominates with the other 2 being low level effects
• As electrons accelerate “up” the tube from cathode toward the anode, some of them collide with residual gas atoms. This collision is energetic enough that it strips an outer orbit electron from the gas atom which turns it into a positively charged ion.
• The positively charge ion accelerates back “down” the tube toward the cathode
• Some of these positively charged ions collect at the grid (which is usually the most negative potential of any tube element). Electrons must flow into the grid to neutralize these ions which is the same as saying that current must flow out if the grid
• This current generates a voltage across the grid leak resistor.
• This voltage makes the grid more positive which SUBTRACTS from the bias and results in increase tube current.
• This effect is proportional to tube (anode) current and so is worse in power tubes
• This effect also is worse in old “gassy” tubes.
• This is why there are always 2 specifications for maximum Rg1 values. One value for cathode (auto) bias where the increased current is opposed by an increased bias due to uncreased voltage drop across the cathode resistor, and another smaller value for fixed bias where there is no action to oppose the current increase.
The mechanism of NEGATIVE grid current, reducing the bias, increasing the current, increasing the negative grid current, reducing the bias etc. etc. round and round is called thermal run away and is what causes a lot of power tubes to self destruct.
This is made worse by the fact that Rg1 values in most guitar amps ignore the recommended maximum Rg1 values. This is done so as to not load down the output of the phase splitter too much. This is usually compensated to some degree by biasing the output tubes at 70% of rated maximum dissipation, that is reduce the tube idle current by 30%. That allows use of an Rg1 value of about double the recommended maximum which is based upon running the tube at 100% of its dissipation rating. This helps at idle but does not help much when running the amp with full signal.
So NEGATIVE or REVERSE grid current is something you really need to watch in power tubes.
It can be a problem in small signal tubes as well, particularly high mu triodes which led to the RDH "Rule of Thumb" that for high mu triodes (like 12AX7, 6SL7 etc.) that Rg1 should be no more than 3 times the anode load resistor for cathode bias and no more than twice the anode load resistor for fixed bias.
Those familiar with 12AX7 circuits used in guitar amps will note that Rg1 is often ten times the anode load resistor value. This is because one of the defining characteristics of a 12AX7 which actually makes it ideal in guitar amps is unusually low NEGATIVE grid current.
The above also explains why grid leak bias does not work with older gassy tubes. The negative grid current from the gas ionization opposes the positive grid current needed to establish the grid leak bias.
There is one more type of grid current, GRID RECTIFICATION Current. When the grid is taken positive with respect to the cathode the grid to cathode circuit starts to look like a forward biased diode. Current into the grid increases with more positive voltage and usually this has the effect of clamping the positive going signal at the grid. The current also charges up any interstage coupling (DC Blocking) capacitor and this is the root cause of blocking distortion.
Cheers,
Ian
Quote:
Grid Current is one of the most poorly understood topic and so at the risk of teaching grandma to suck eggs I've written a little essay.
Grid Current and all that – remember that, due to historic limited knowledge, current flow is defined as from positive to negative whereas in actual fact electrons flow from negative to positive.
There are 3 types of grid current
Current flowing into the grid is known as POSITIVE grid current
• When the cathode is heated a cloud of electrons forms around the cathode known as a “space charge”
• Some of these electrons gather at the grid. These electrons then flow out of the grid which is the same as saying that current flows into the grid.
• This positive grid current generates a voltage across the grid leak resistor (Rg1)
• This voltage makes the grid more negative which ADDS to the bias
• If the grid leak resistor is large enough then this positive grid current can generate the entire required bias – This is known as “Grid Leak Bias”
POSITIVE grid current is a low level phenomenon and can easily be overshadowed by NEGATIVE grid current
Current flowing out of the grid is known as NEGATIVE (or REVERSE) grid current
• Negative grid current can be caused by
1. gas ioniszation current
2. leakage current (grid to cathode)
3. grid emission (from grid being heated by the cathode, screen or anode)
• gas ionization current dominates with the other 2 being low level effects
• As electrons accelerate “up” the tube from cathode toward the anode, some of them collide with residual gas atoms. This collision is energetic enough that it strips an outer orbit electron from the gas atom which turns it into a positively charged ion.
• The positively charge ion accelerates back “down” the tube toward the cathode
• Some of these positively charged ions collect at the grid (which is usually the most negative potential of any tube element). Electrons must flow into the grid to neutralize these ions which is the same as saying that current must flow out if the grid
• This current generates a voltage across the grid leak resistor.
• This voltage makes the grid more positive which SUBTRACTS from the bias and results in increase tube current.
• This effect is proportional to tube (anode) current and so is worse in power tubes
• This effect also is worse in old “gassy” tubes.
• This is why there are always 2 specifications for maximum Rg1 values. One value for cathode (auto) bias where the increased current is opposed by an increased bias due to uncreased voltage drop across the cathode resistor, and another smaller value for fixed bias where there is no action to oppose the current increase.
The mechanism of NEGATIVE grid current, reducing the bias, increasing the current, increasing the negative grid current, reducing the bias etc. etc. round and round is called thermal run away and is what causes a lot of power tubes to self destruct.
This is made worse by the fact that Rg1 values in most guitar amps ignore the recommended maximum Rg1 values. This is done so as to not load down the output of the phase splitter too much. This is usually compensated to some degree by biasing the output tubes at 70% of rated maximum dissipation, that is reduce the tube idle current by 30%. That allows use of an Rg1 value of about double the recommended maximum which is based upon running the tube at 100% of its dissipation rating. This helps at idle but does not help much when running the amp with full signal.
So NEGATIVE or REVERSE grid current is something you really need to watch in power tubes.
It can be a problem in small signal tubes as well, particularly high mu triodes which led to the RDH "Rule of Thumb" that for high mu triodes (like 12AX7, 6SL7 etc.) that Rg1 should be no more than 3 times the anode load resistor for cathode bias and no more than twice the anode load resistor for fixed bias.
Those familiar with 12AX7 circuits used in guitar amps will note that Rg1 is often ten times the anode load resistor value. This is because one of the defining characteristics of a 12AX7 which actually makes it ideal in guitar amps is unusually low NEGATIVE grid current.
The above also explains why grid leak bias does not work with older gassy tubes. The negative grid current from the gas ionization opposes the positive grid current needed to establish the grid leak bias.
There is one more type of grid current, GRID RECTIFICATION Current. When the grid is taken positive with respect to the cathode the grid to cathode circuit starts to look like a forward biased diode. Current into the grid increases with more positive voltage and usually this has the effect of clamping the positive going signal at the grid. The current also charges up any interstage coupling (DC Blocking) capacitor and this is the root cause of blocking distortion.
Cheers,
Ian
Good summary, except that I wonder if you have negative and positive grid current swapped over? (Or have I swapped them over myself?)
The thing which can catch people out is that the current can flow in the opposite way to the applied voltage, so the grid becomes a tiny generator. This, of course, is how grid leak bias works. To stop the grid current you need to set the grid negative by somewhere around -1.0V. Some 12AX7 circuits don't do this so suffer from grid current distortion if the source has a high impedance. I say 'suffer', but in a guitar amp this might be an advantage!
The thing which can catch people out is that the current can flow in the opposite way to the applied voltage, so the grid becomes a tiny generator. This, of course, is how grid leak bias works. To stop the grid current you need to set the grid negative by somewhere around -1.0V. Some 12AX7 circuits don't do this so suffer from grid current distortion if the source has a high impedance. I say 'suffer', but in a guitar amp this might be an advantage!
I think I have them correct although different texts seem to have them swapped as well. I think that has been one of the main problems when discussing the topic, there has been a lack of consistency with what is called POSITIVE and NEGATIVE grid current.
Add that to the fact that POSITIVE grid current results in a NEGATIVE voltage being developed at the grid which adds to the bias and it easy to get confused. I think I have followed the definitions as used in RDH. Certainly let me (and others) know if I have it wrong. The last thing I want to do when trying to help out the newer guys is to tell them stuff which is not correct.
To hold a 12AX7 "out of grid current" you need to keep the anode voltage up, typically greater than 180 Volts and keep the anode current less than say 1 to 1.5 mA. This makes the design of the input stage of a guitar amp a little tricky, you want high'ish current for low noise but need to keep the anode voltage up too to limit grid current.
Cheers,
Ian
No, more like 'in' vs. 'out' convention. If I apply a small negative voltage to a device then I would expect a small (positive convention) current to flow out of the device. If the device is a valve grid then in fact I may get a small current flowing into the device. As this is the opposite of what is expected I would call that a negative current. If I put a significant positive voltage on a valve grid I get a significant current flowing into the grid - I would call this positive current as it is what is expected.
The confusion arises because we tend to think of a valve grid as a load, but it is actually a generator for some of the voltage range we use it at.
The confusion arises because we tend to think of a valve grid as a load, but it is actually a generator for some of the voltage range we use it at.
- Status
- Not open for further replies.