Hoffman Amplifiers Tube Amplifier Forum
Amp Stuff => Tube Amp Building - Tweaks - Repairs => Topic started by: Shack on January 27, 2017, 05:05:16 pm
-
This is just a question based on looking at alot of schematics. Different amps use such vastly different values for them, and I dont know why. May be a noob question.
Fender uses a 1.5k grid resistor....Marshall a 5.6k ...and the amp im redoing originally had 47k grid resistors. Just wonder why so different when it uses the same tube compliment of the Fender.
And as far as screen resistors go....this amp originally used a strange ( to me ) setup, where the only screen resistors was a 100 ohm in between the paralleled output tubes instead of a 470 ohm to each screen.
Is there a practical reason for this that relates to tone of an amp ?
-
Grid stopper resistors are handy for avoiding blocking distortion - if the resistance is sufficiently large enough to help prevent the coupling cap from charging up too quickly. However, grid stoppers also interact with the input capacitance of the tube, to produce a low-pass (R/C) filter, more or less in accordance with the formula f=1/(2Pii x R x C), and the rolloff is 20bB/decade (6dB/Octave).
Therefore, the bigger the grid stopper resistance, the more HF gets rolled off. The 'good news' is that pentodes and tetrodes (output tubes) have very small input capacitance, so the grid stoppers can be 'quite large' without affecting the treble response. Hence, a 47k grid stopper on your 5881 will not really affect the audible frequency band (see the Supro Thunderbolt schematic).
Having said that, you want to also consider the grid leak resistance for output tubes (which needs to be lower than what is typically needed for pre-amp tubes). The grid stopper is effectively in series with the grid leak resistor when it comes to working out how much grid leak resistance you end up with - so if you increase the grid stopper, you need to think about decreasing the grid leak resistance. This also has implications for impedance-bridging between the driver stage and the output tube, because if you decrease the grid leak resistance, you will attenuate the signal more. (The grid leak resistance represents the 'load impedance' part of the impedance bridge, and the driving stage's plate resistance and load resistor represent the 'source impedance' part). So it's a bit of a balancing act.
-
Screen resistors, on the other hand, are handy for eating up excess screen current when the plate voltage is peaking. If the Rg2 is un-bypassed (as most screen grid resistors are) then the larger the Rg2 resistance, the more compression you will end up with in the g1 curves. (i.e. more compression = decreased transconductance). Having said that, a Rg2 value of 470R to 1k on your typical 6V6 or 6L6 screen won't have much impact on compression, but 100k will compress it heaps.
-
thank you....wish I knew what you guys know :worthy1:
-
Grid stopper resistors are handy for avoiding blocking distortion ... However, grid stoppers also interact with the input capacitance of the tube, to produce a low-pass (R/C) filter ... Therefore, the bigger the grid stopper resistance, the more HF gets rolled off. ...
Using the grid stopper with the intent of avoiding blocking distortion is mostly a modern thing, especially if the amp is a PA or hi-fi affair.
The primary (but not only) reason is the high-frequency roll-off noted. Long grid wires can act as an antenna, and even straight wire has appreciable inductance at radio frequency (RF). The tube has capacitance from grid-to-plate. You could view that L (wire) as being in series with the C (tube's Miller capacitance), with the potential of being a series LC circuit which could resonate at RF.
Adding series resistance tends to damp the LC-resonate circuit, making it a less-good oscillator. So you could see the grid stopper as "adding a high frequency roll-off" (and reducing gain at RF), or as "reducing the Q of the LC-resonant circuit" (and damping the circuit's peak output at resonance). Both accomplish the same goal different ways (at least they seem different at first).
So if you have long grid wires and more tendency for the output tubes to oscillate, you add grid stoppers. If a small-value R doesn't get the job done, you make it a bigger-R.
High transconductance (Gm) tubes, like the EL34 and EL84, will have bigger changes of plate current with a given grid voltage input than lower-Gm tubes. 6V6 and 6L6 are "low-ish Gm" when compared to EL34 and EL84. That means the latter may be more susceptible to oscillating, and so bigger grid stopper should be seen as "normal".
So this isn't necessarily for "tone" but you can use grid stoppers to attack oscillation, inhibit blocking distortion due to grid current, and even roll off a touch of treble (with very big R).
Screen Resistors:
What Tubeswell said. Plus the below...
The screen is just a coil of wire. If it dissipates enough power, it will melt. When you play your amp loud, plate current goes up, but some of that is diverted as screen current. The plate voltage falls when plate current goes up (because the OT load drops a.c. volts), but the screen is pegged to a power supply node so its voltage stays steady. Power = Voltage * Current, so if Volts stays unchanged and Current gets bigger, the screen dissipates more power and comes closer to the melting point.
Do you know Ohm's Law? Voltage = Current * Resistance; when Current flows through a Resistor, Volts are dropped across the resistor. More R, more Volts get dropped. The designer figures how much screen current is likely to rise when the amp is slammed to full output. They then figure how much screen voltage would have to drop from its idle value to stay within the screen's power rating (Power = Current * Volts; PowerScreen Rating / CurrentScreen @ Max Output Power = VoltsScreen @ Max Output Power).
Adding series Resistance (VoltsScreen @ Idle - VoltsScreen @ Max Output Power) / CurrentScreen @ Max Output Power = ResistanceScreen, which assures screen voltage will drop to a safe value when you're playing loud.
Typically, amp designers used the smallest screen resistor they could get away with while keeping the screen safe from melting. That's because falling screen voltage limits the tube's maximum plate current, and consequently how much output power the tubes can manage. As Tubeswell said, you could intentionally make the screen resistor "too big" to add compression to your amp, though you'll generally hear it only when you really crank the amp.
So this isn't about "tone" but "power response". In the old days amp designers were trying to get every watt they could from the amp, screen R tended to be small (if present at all). But guitarists and their amps beat up on tubes, "every last watt" usually isn't a priority anymore, and some will prefer the added compression.
-
I don't want to hijack this thread but i hate to start a new one just to ask this, but a quick answer without stopping the original dialogue would be appreciated. I have the 5.6k 6V6 grid stoppers on the turret board with about 3-4" of wire going to the tube grid. Would it be a good idea to put the resistors on the grid pins or would it likely be subtle to no difference in any respect?
-
lol, not a hijack at all, because id like to know too :icon_biggrin:
Thank you for the answers , I feel way more confident knowing why I am putting this component in there, than just because the schematic or layout says to
-
So, I think its safe to say that on my 4 x 6L6 amp, using the Fender standard 1.5k grid and 470 ohm screen resistors is the way to go.
-
What my website says about power tube screen stoppers:
Since a screen resistor will restrict the flow of electrons off the screen and cause a buildup of electrons, a screen resistor will accelerate screen voltage drop and gain shift distortion and so quicken the transition from clean to distortion. Increasing the resistance value of the screen resistor will increase the overdrive screen voltage drop and therefore decrease gain. Lower gain during overdrive causes compression and increases sustain. Adding screen resistance also slows the screen voltage recovery time which tends to have an "averaging" effect on the voltage drop and "soften" the overdrive tone.
In other words, increasing screen resistance increases the screen voltage drop and its gain reduction distortion but makes the voltage changes slower and less dynamic.
Reducing screen resistance reduces the screen voltage drop but accelerates screen voltage recovery making the voltage changes more dynamic.
The addition of a screen-to-cathode screen capacitor can reduce the chance of oscillation but more importantly adds a time-variance factor to screen voltage drop and its gain shift distortion. The capacitor is charged by screen current and acts as an electron reservoir. The cap's charge/discharge cycle can alter the character of pentode and beam tetrode distortion with added complexity. The addition of a screen cap and trying different cap values may be worth experimentation when tuning an amp's power amp overdrive tone.
Since screen current affects the tube's operating point along the transfer curve and changes the shape and slope of the tube's transfer curve it has more effect on power tube overdrive tone than grid current. When tuning an amp's overdrive tone close attention should be paid to the screen grid resistor and screen-to-cathode cap values. Taking the time to experiment with different component values can pay great dividends and help differentiate the voice of your amp from the rest of the pack.
Personally I like as much screen voltage fluctuation as safely possible because it changes the shape and slope of the transfer curve which generates tons of harmonic and intermodulation distortion so I use the smallest ohm value resistors that can safely limit overdrive screen current. My suggested screen stopper resistor values are (one screen stopper per tube) :
6V6: 470 ohm (3 watt or higher)
6L6, 5881: 470 ohm 5 watt
EL84 (true pentode) 1k 3 watt
EL34 (true pentode) 1k 5 watt
These values have stood the test of time for tone and reliability.
-
I have the 5.6k 6V6 grid stoppers on the turret board with about 3-4" of wire going to the tube grid. Would it be a good idea to put the resistors on the grid pins or would it likely be subtle to no difference in any respect?
Modify message
Many people say the grid stoppers are most effective when mounted directly on the tube base. I like the way Fender did this in the blackface era. The 1.5K grid stopper was mounted between pins 1 and 5 with the wire from the board connected to pin 1. Put the resistor down close to the insulating base. The 470Ω /1W screen resistors were mounted between pins 6 and 4 with the wire connected to pin 6. Mount these about 1/2" above the socket pins for cooling. Here's a pic...
http://sluckeyamps.com/6v6plexi/P-6V6_05_big.jpg (http://sluckeyamps.com/6v6plexi/P-6V6_05_big.jpg)
Of course you cannot use pin 1 of an EL34 as a tie point for the grid stoppers.
-
... Personally I like as much screen voltage fluctuation as safely possible because it changes the shape and slope of the transfer curve which generates tons of harmonic and intermodulation distortion so I use the smallest ohm value resistors that can safely limit overdrive screen current. ...
I don't follow, because small screen resistors result in small voltage variation, and a more-stable screen voltage.
... Increasing the resistance value of the screen resistor will increase the overdrive screen voltage drop and therefore decrease gain. ...
Pentodes/Beam Power Tubes have higher voltage gain when screen voltage is made smaller. This is a part of why preamp pentodes are often run at a screen voltage very much less than their plate voltage.
You may want to differentiate between "gain" and "distortion" if you were really trying to describe the latter.
What my website says about power tube screen stoppers:
Since a screen resistor will restrict the flow of electrons off the screen and cause a buildup of electrons, a screen resistor will accelerate screen voltage drop...
Forget a pentode's plate (and suppressor/beam-forming plate) for a moment, and see the screen as the "plate" of a triode. For a given grid-to-cathode voltage, raising a triode's plate voltage results in more plate current. The same thing happens in a pentode, except the screen perform the function of the "triode plate" and has the influence of the "triode plate" over plate current because it is physically closer to the cathode.
So it's not that there's a buildup of electrons on the screen repelling plate current, but rather a higher screen voltage is a positive voltage gradient (relative to the cathode) which tends to suck electrons toward it. But being a coarse coil of wire, most electrons accelerated in its direction fly right through it an onward to the plate.
... Adding screen resistance also slows the screen voltage recovery time ... increasing screen resistance ... makes the voltage changes slower and less dynamic.
Reducing screen resistance ... accelerates screen voltage recovery making the voltage changes more dynamic.
Unless the amp has a (smallish) cap at the screen pin and after a largish resistor, there is no "time constant" to speak of, and no "slow recovery". Assuming a typical filter cap for the screen supply node and a low-resistance power supply up to that filter cap, the screen resistor sees a fairly constant voltage at the supply end. On the screen end of that resistor, voltage will change essentially instantly as screen current changes. If the resistor is a low value, the voltage at the screen end changes very little.
Adding a cap after a large resistance from screen-to-cathode may change the frequency response some (shaving bass if the cap is smallish), but does so in the same way as a cathode bypass cap.
-
Unless the amp has a (smallish) cap at the screen pin and after a largish resistor, there is no "time constant" to speak of, and no "slow recovery".
You are 100% correct. I don't know how I missed this. Yes, I have some editing to do.
-
Pentodes/Beam Power Tubes have higher voltage gain when screen voltage is made smaller. This is a part of why preamp pentodes are often run at a screen voltage very much less than their plate voltage.
I don't understand this. If that's true then why wouldn't removing the screen (making it a triode) max out gain?
See "How Screen Voltage Affects Large Signal Gain (https://www.ampbooks.com/mobile/amp-technology/pentode-gain/)"
In a power output stage, the peak plate current works against the load impedance to make our power output. Dropping the screen voltage reduces the peak plate current we get when the grid-to-cathode voltage is made 0v, as in the case where an output stage biased at -30v receives a 30v peak input signal. You can see this reduction of possible plate current in the top graph of page 6 in this 6L6GC data sheet (http://www.mif.pg.gda.pl/homepages/frank/sheets/093/6/6L6GC.pdf). The box in the upper-right corner notes that Ec1 is being held at 0v, but that's relevant because this resulting curve is the saturation plate current for the marked screen voltage.
This is the basis of compression in the output tube due to a large screen resistor.
But where we might commonly use a screen voltage same/similar as the plate voltage in many guitar amps, that wasn't always done in older tube amps. The plate may have 400v and the screen be set at 250v so a smaller idle bias could be used (perhaps provided by a cathode resistor). Now a smaller driving input signal will cause the tube to swing all the plate current it can based on the lower screen voltage (so our "gain" has gone up, and we don't need as much amplification in the preamp).
But power output is plate current swing times plate voltage swing; we could also say Plate Current2 * Load Impedance, and infer the plate voltage swing based on the two factors we have more-direct control over. Since available plate current swing is reduced, power is reduced, but we can raise the load impedance to raise the power output.
Ultimately, design is about juggling all these different factors to find the best compromise which meets our goals.
-
Thanks HotBlue for the response. I found the "How Screen Voltage Affects Large Signal Gain" via Google.
Does this seem reasonable: Since our power tube is being overdriven causing screen current and screen voltage drop, the voltage gain rises but plate current and output power drops. The screen voltage change causes a transfer curve shape shift that generates harmonic and intermodulation distortion.
-
> tube is being overdriven causing screen current and screen voltage drop, the voltage gain rises
High screen current happens when plate is "bottomed", driven as low as it can go. So "voltage gain" is not happening at all anymore. Power output is as high as it can be. Distortion is happening. While a large drop of screen voltage would (and does) reduce power output, the difference is small for ~~1K resistors on the usual bottles. (I once built with 3K and bypassed, got peak power which sagged over a musical time period, and it played well. But the sag was small and I was never sure it happened in normal hard playing, or only on dummy-load.)
-
I have the 5.6k 6V6 grid stoppers on the turret board with about 3-4" of wire going to the tube grid. Would it be a good idea to put the resistors on the grid pins or would it likely be subtle to no difference in any respect?
Modify message
Many people say the grid stoppers are most effective when mounted directly on the tube base. I like the way Fender did this in the blackface era. The 1.5K grid stopper was mounted between pins 1 and 5 with the wire from the board connected to pin 1. Put the resistor down close to the insulating base. The 470Ω /1W screen resistors were mounted between pins 6 and 4 with the wire connected to pin 6. Mount these about 1/2" above the socket pins for cooling.
This is my plan of attack with my build Steve
Edit: Untangled quote, Willabe.
-
... While a large drop of screen voltage would (and does) reduce power output, the difference is small for ~~1K resistors on the usual bottles. (I once built with 3K and bypassed, got peak power which sagged over a musical time period, and it played well. But the sag was small and I was never sure it happened in normal hard playing, or only on dummy-load.)
I think Tubenit has built one or more amps with screen resistors on the order of 3kΩ, and could comment. Agreed that "normal size" resistors don't cause this effect.
... Since our power tube is being overdriven causing screen current and screen voltage drop, the voltage gain rises but plate current and output power drops. The screen voltage change causes a transfer curve shape shift that generates harmonic and intermodulation distortion.
The statement scenario assumes the tube is already being overdriven, then implies screen voltage drop is the cause of "overdrive" (distortion).
If we're already hitting the output tube with a signal bigger than it can handle, there could be distortion due to grid blocking, grid curve convergence below the knee, plate current saturation, grid curve crowding near cutoff, and other effects. If we can't be sure (at least I'll say, "I'm not sure") screen voltage drop is actually contributing to the distortion, why bother claiming it?
Separately, we know from using compressors & limiters that we can have "amplitude distortion" (compression) all day long without the signal sounding "distorted" (harmonic or I.M distortion). I'll acknowledge that one can distort a bass sound using the right (wrong?) attack/release settings of a compressor, but that shouldn't necessarily be construed to mean compression always adds distortion (other than "amplitude distortion" of the resulting signal's amplitude changes don't exactly track the input signal's amplitude changes).
-
How did you people learn all this stuff? I have read gerald weber, kevin o'conner and dan torres books, and built a couple amps, and I dont think ill ever know the info that I get fed here.....incredible :icon_biggrin:
-
How did you people learn all this stuff? ...
I blame PRR. :icon_biggrin:
(Also many years here discussing my "uneducated guesses" until others pointed out how I screwed up the guess, which helped me see how things really work.)
-
FWIW, I don't have anything but anecdotal evidence to add to this. On my recent ab763 build, I made the mistake of using 470k screen resistors as opposed to 470ohms. The result was very low volume output and a mushy, distorted signal. It also affected the bias range.
-
If we're already hitting the output tube with a signal bigger than it can handle, there could be distortion due to grid blocking, grid curve convergence below the knee, plate current saturation, grid curve crowding near cutoff, and other effects. If we can't be sure (at least I'll say, "I'm not sure") screen voltage drop is actually contributing to the distortion, why bother claiming it?
Because in Neumann & Irving's Guitar Amplifier Overdrive they state that screen voltage changes cause not only the tube's operating point to change, but it also causes the tube's transfer curve to actually change shape which generates more nonlinear, harmonic and intermodulation distortion than grid current and grid voltage change (bias drift). So yes, other distortion is going on but screen current generates most of our beloved power tube distortion.
-
Because in Neumann & Irving's Guitar Amplifier Overdrive they state that screen voltage changes cause not only the tube's operating point to change, but it also causes the tube's transfer curve to change change shape which generates more harmonic and intermodulation distortion than grid current and grid voltage change (bias drift). So yes, other distortion is going on but screen current generates most of our beloved power tube distortion.
It's all about relationships, so who got it right? Does anyone have some schematic examples of designers that got the screen vs. grid vs plate vs cathode resistors right? Sunn Pentode which is really the Dynaco HiFi PA section? HiWatt? What is the actual relationship? Is it linear? While I think Marshall and Fender got relationships right, I wonder if it was luck or within a limited range of figured it out? This is starting to sound like Calculus... Which I have no functional knowledge of. Now, to reference the old tube engineering books: I believe the authors understood these concepts but had no knowledge of high gain or even the understanding of "sonics", or how these mathematical relationships resulted in the sort of tones we're looking for... Sad. But room for innovation. On the other hand, Fender and Marshall, Vox and others understood tone and experimented in realms the had little actual, (read math), knowledge in. If only... If only someone like Tesla had designed the guitar amplifier. But then, we wouldn't have anything left to do.
silverfox.
-
> screen voltage changes ... causes the tube's transfer curve to actually change shape
I have not read Neumann & Irving's Guitar Amplifier Overdrive (https://www.amazon.com/Guitar-Amplifier-Overdrive-Ulrich-Neumann/dp/132959665X) so do not know what they say.
They may have looked at 6BA6. See top of page 4, also bottom page 3.
http://www.mif.pg.gda.pl/homepages/frank/sheets/093/6/6BA6.pdf (http://www.mif.pg.gda.pl/homepages/frank/sheets/093/6/6BA6.pdf)
Several other "remote cutoff" RF/IF tubes have similar data.
The dot-line is for Rg2 of 33,000 Ohms. Yes, 6BA6 is a much smaller tube. And the plots for this small-signal tube run much further down from "Max" current than we would ever swing a power tube. Let us dart-board the 6BA6 as a "10mA" tube, and 6L6 as a "100mA" tube. Then 33,000 on the small tube may be like 3,300 on the large tube. This is well above the usual guitar-amp's 470 or 1K.
For 6L6/6V6 in particular the difference is even greater because they have been tuned for lo-low G2 current. For EL34 and similar hungry-G2 tubes it may be somewhat fair comparison. However "remote cutoff" changes several aspects of tube construction; also how far down we plot data.
However also note that a "good power tube" op-point for 6BA6, say 10mA, but with 33K Rg2, leaves G2 voltage around *half* of G2 supply voltage (above the 33K). We _never_ have that much Rg2 drop in power amps.
I remember some guy working with 6L6 variable-gain limiting, but do not recall if he used a large G2 resistor to invoke this curvature.
FWIW (about nothing), my simulator model for 6L6 or 6550 clearly do not model G2 current correctly so I can't ask the computer for clues.
-
> If only someone like Tesla had designed the guitar amplifier.
You do realize that Tesla published very very little hard information in his lifetime? There's the patents for poly-phase motors and systems which did become the major industry (when combined with the innovations and hard work of others). There's patents on "radio remote control" which are slightly ahead of their time and not well documented (useful radio comes from Marconi's internal development). There is his disk-turbine, which has little practical use. There were many-many "WOW!" demonstrations of sparks and glow-tubes, none of it written down by Tesla or qualified observers.
He had the knack of wires and ideas. He did not have the nose-to-the-bench gumption of Fender and White in pushing products to buyers/users.
I am of course overlooking Tesla's trunk of papers which was taken by the government and never revealed. If lost (I know government at work), that's a tragedy. If leaked to jump-start our technology, what came of it? There are few truly-new things in that period. (Maybe the ball-point pen.) Most things we think of as "new in the 1940s" we find have deep roots. The Transistor (which could have been a Tesla secret) is clearly based on observations in the 1920s and 1930s plus crystal diodes from crude galena to the mature microwave diodes at the end of WWII. The transistor had many fathers (not just Bell's B B & S), and Shockley is its grandfather because he worked out _all_ the theory in one mighty book. RADAR goes way back, and practical radar was grunt-work on the complexity (aided by low-cost popular radio tubes). Electric/electronic computers are very old but nobody needed one until WWII gunnery tables. Theory of the Jet Engine was common info in the 1930s but awaited fine data on blowers and much tougher metals.
Yeah, I know. Tesla invented anti-gravity, the notes were taken to Area 51 to build the flying saucers which appeared "secretly" in the early 1950s, and are the forerunners of the flying saucers we all drive today.
-
the notes were taken to Area 51
I went to the area, I still can't walk on water :think1:
-
Because in Neumann & Irving's Guitar Amplifier Overdrive they state ...
I haven't read it either; I'll have to check it out. (EDIT: Order placed for a copy)
Just from the title, one note of caution occurs to me:
If the book is about distortion, there a good chance other 'effects' for a given 'cause' may be ignored because they're "outside the scope" of the book.
I don't know if that's a relevant warning for this text, but I've seen it in other books (mostly meaning that readers assigned some issue/effect a severity totally out of proportion to how often it actually happens; e.g., cathode stripping requiring amps have standby switches for B+).
-
It's all about relationships, so who got it right? Does anyone have some schematic examples of designers that got the screen vs. grid vs plate vs cathode resistors right? ...
That's like asking, "Who makes BBQ right?" Lot of different goals, lot of different flavors, lot of ways to get there.
-
Tesla transistor? Nope, that was stolen from T. Henry Moray.
Jim
-
This is summarized from an (as yet unpublished) writeup I've been assembling.
The "Grid Stop Resistor" or "Grid-Stopper Resistor" (GSR), serves five different purposes:
1) As an input filter to remove unwanted RF signals from the grid by forming a low-pass filter at high frequencies. This prevents high-frequency (> 20 KHz, often > 100 KHz) parasitic oscillations in the tube, but also prevents AM signal from being rectified and low-pass filtered and becoming audible.
2) Limiting flow of current from cathode-to-grid because of the potential differences in VG and VP when the tube is starting up, particularly when the capacitors are already charged. By slightly increasing the resistance the GSR prevents arcing to the grid as a preferential plate.
3) Limiting grid current if the grid is driven positive, thereby preventing grid heating to the point of damage.
4) Damping or isolating any inductive reactance from the grid wiring itself which combines with the tube’s internal capacitance to form a tuned circuit. The GSR effectively lowers the Q of the LC circuit formed from the tube electrodes (Miller Effect), the Grid Leak Resistor, and the Coupling Capacitor.
5) Preventing the coupling capacitor from charging to a DC value.
That's the short version, at any rate. Some of this has been addressed, some not.
-
Almost forgot. Here's the analogous piece for the Screen Stopper Resistor, summarized from a much larger discussion.
The SSR:
1) Limits cathode-to-screen current, protecting each from excessive. Excessive cathode current caused by the omission of an SSR can cause the cathode bias resistor to burn out as well as damage the tube.
2) Prevents cathode-to-screen current when the speaker impedance drops very low and causes B+ sag to the point that the screen's potential is close to plate's potential.
3) Forms an RC circuit with the tube’s internal capacitance (screen-to-cathode and screen-to-plate) to eliminate high-frequency (ultrasonic) parasitic oscillation.
4) Prevents cathode-to-screen arcing, as the amount of current that can pass through the SSR is limited, so the arc can never get started.
Again, some issues have been covered, some not. So I thought it beneficial to post that summary.
-
Thanks for the points.
Implied but not(?) said outright: Williamson taught the use of ~~47K series grid resistance to minimize "grid blocking" on transients in Hi-Fi. Grid blocking is common in hard-beaten guitar amplifiers. Playing to the edge of grid-block is a feature of some player's style, so we do not wish to minimize it too much. Also we tend to smaller grid-caps than Williamson, thus shorter grid-block intervals. However EL84's high sensitivity means it is prone to "crap-out" (going silent) with common drivers.
-
> prevents AM signal from being rectified ...and becoming audible.
This is very real in the input stage. I think this discussion is more about a Power stage. We are unlikely to have the 20V-50V of AM needed to rectify past a power tube's bias; if we do, the input stage is likely totally overwhelmed long before the 'L6/'34 squawks hit-radio.
> remove unwanted RF signals from the grid by forming a low-pass filter at high frequencies. This prevents high-frequency (> 20 KHz, often > 100 KHz) parasitic oscillations in the tube
I had a peek at 6L6 data. It makes a difference metal, glass, or GC. Miller Effect multiplies Cgp. But the worst case seems to be -G in Class A. Cgp is 0.9pFd and gain g-p may be 15. This gives effective 14pFd. Add Cgk of 10pFd, and say 5pFd around the pin. 30pFd? Taking just the common 4.7K grid stopper, this is just past 1MHz. Indeed you can have MHz parasitics. The "transformer" acts like a ~~500pFd lump plus a few microH of lead inductance, a poor MHz tank. We also find "push-push" coupling in layouts that look push-pull for audio. And strays in the grid wiring, as you point out. A few K at the grid spoils the Q of these unintended resonances and damps the oscillation.
The 4.7K does "nothing" about the intended audio response, which is dominated by the ~~40K driver impedance. Even this leads to >100KHz response, far better than we need. (Williamson's 47K stopper, with ~~5K driver, is similar, but critical to his high-NFB approach; it could not be much higher in a Williamson.)
-
... The "Grid Stop Resistor" or "Grid-Stopper Resistor" (GSR), serves five different purposes:
1) As an input filter to remove unwanted RF signals from the grid by forming a low-pass filter at high frequencies. This prevents high-frequency (> 20 KHz, often > 100 KHz) parasitic oscillations in the tube, but also prevents AM signal from being rectified and low-pass filtered and becoming audible.
...
4) Damping or isolating any the resonant circuit formed by inductive reactance from the grid wiring itself which combinesd with the tube’s internal capacitance to form a tuned circuit. The GSR effectively lowers the Q of the LC circuit formed from the tube electrodes (Miller Effect), the Grid Leak Resistor, and the Coupling Capacitor grid wiring. ...
To my mind, #1 & #4 are two ways of looking at the same effect. I've added strikethroughs & italics to edit the statement (C is internal to the tube, plus stray capacitance due to wiring, L is stray inductance mostly due to wiring).
...
2) Limiting flow of current from cathode-to-grid because of the potential differences in VG and VP when the tube is starting up, particularly when the capacitors are already charged. By slightly increasing the resistance the GSR prevents arcing to the grid as a preferential plate.
...
... Here's the analogous piece for the Screen Stopper Resistor ...
The SSR:
...
4) Prevents cathode-to-screen arcing, as the amount of current that can pass through the SSR is limited, so the arc can never get started.
...
I'll take your word these resistor help in this regard; I've never found this to be a problem to solve in audio amps I've owned.
...
2) Prevents cathode-to-screen current when the speaker impedance drops very low and causes B+ sag to the point that the screen's potential is close to plate's potential.
...
Typically the speaker's lowest impedance is equal to the nominal impedance used to design the output stage. That is, the speaker only ever looks like "more impedance" than designed, not less.
Separately, the screen resistor has most impact on the tube when the plate current is very high and the plate voltage is momentarily swinging very low, which results in the screen being a couple-hundred volts higher than plate voltage.
Screen voltage doesn't sag to be "close to the plate's potential" at that moment, otherwise you'd be in triode mode and have very much less power output.
The SSR:
1) Limits cathode-to-screen current, protecting each from excessive. Excessive cathode current caused by the omission of an SSR can cause the cathode bias resistor to burn out as well as damage the tube.
...
I personally wouldn't look to the screen resistor to limit cathode current; G1 voltage is much more effective in that role. It is also a benefit of cathode bias that the cathode resistor would burn open before the tube could destroy itself (or burn other components).
-
For 6L6, I am not over-concerned about grid current.
6L6 has an AB2 rating. It shows 14V positive peaks, which seems to be (from 807 data) 20mA peak grid current, which is in the ballpark of the 0.1W-0.2W grid power specified (on 807) for AB2 use.
Our usual drivers only have 2mA flowing, can't put all of this to the grid, and if the average grid current approaches 1mA the C-R coupling will drive the grid very negative past grid-block.
A sudden power-up could flow 400V through 100K-82K and coupling cap to the grid. 5mA max. Time constant of 100K against say 0.1uFd is 0.010 Seconds. I think 25mA for less than 0.010 Seconds is not a big strain for 6L6 grid. Actually if it came-up this way, the 6L6 would conduct HARD, the B+ would not rise fast and probably not to nominal value, so the actual grid strain would be less.
EL34 has no G1 power spec (except an obscure line in Triode mode). Their "Class B" data is for a Class B1 condition, no positive grid. Conservatively we would avoid any G1 current at all (which is about impossible). Realistically, a EL34's grid is not much smaller than a 6L6's (anybody got some duds to crack?) and will take some current flow.
EL84, 7591, and (IMHO particularly) 8417 have close-spaced grids which would show higher conductance in smaller wire. We could blow these up easier. We do know the EL84 does not blow-up so easy, even though we often run well past 330V and 12W on the plate while hammering the grid.
I may be alone in the opinion that tubes today are CHEAP, cheaper (in real value, gasoline beer and hamburger index) than in the 1970s, and should be considered wear-items, like car tires. I like my car tires to last a long time, but I accept that treads and grids will throw chunks and need replacement every so often.
-
> prevents AM signal from being rectified ...and becoming audible.
This is very real in the input stage. I think this discussion is more about a Power stage. We are unlikely to have the 20V-50V of AM needed to rectify past a power tube's bias; if we do, the input stage is likely totally overwhelmed long before the 'L6/'34 squawks hit-radio.
Fair enough for actual AM talk-radio reception at the initial gain stage. That was just a list of all the purposes I found for the GSR. When I was deciphering circuits it would have been rare for any one writeup to discuss more than one or two of the functions.
> remove unwanted RF signals from the grid by forming a low-pass filter at high frequencies. This prevents high-frequency (> 20 KHz, often > 100 KHz) parasitic oscillations in the tube
I had a peek at 6L6 data. It makes a difference metal, glass, or GC. Miller Effect multiplies Cgp. But the worst case seems to be -G in Class A. Cgp is 0.9pFd and gain g-p may be 15. This gives effective 14pFd. Add Cgk of 10pFd, and say 5pFd around the pin. 30pFd? Taking just the common 4.7K grid stopper, this is just past 1MHz. Indeed you can have MHz parasitics. The "transformer" acts like a ~~500pFd lump plus a few microH of lead inductance, a poor MHz tank. We also find "push-push" coupling in layouts that look push-pull for audio. And strays in the grid wiring, as you point out. A few K at the grid spoils the Q of these unintended resonances and damps the oscillation.
Good point. Interesting how very small changes ruin the Q.
What is interesting is the claims that oscillation robs power. Patrick Turner wrote that amplifier designers routinely neglect this sort of oscillation and proposed adding Zobels to each stage to damp it out. You have any feelings about this?
The 4.7K does "nothing" about the intended audio response, which is dominated by the ~~40K driver impedance. Even this leads to >100KHz response, far better than we need. (Williamson's 47K stopper, with ~~5K driver, is similar, but critical to his high-NFB approach; it could not be much higher in a Williamson.)
Yes, a very good point.
Williamson's paper, which I just looked at, shows a 1 kΩ resistor as the GSR, and the improved amplifier uses the same value. Since Fc = 1 / (2 × Pi × R × C), given R = 1 kΩ and C = 30 pF, that works out to be about 5 MHz. Using 4.7 kΩ drops that to 1 MHz.
Why did Williamson originally pick such a small value?
-
To my mind, #1 & #4 are two ways of looking at the same effect.
Yes, I see that. Good point. They both are addressed by altering the Q.
I was trying to differentiate between RF that gets into the system from outside and oscillation that is generated by the tube itself. They both get damped, but they arise in different ways. Maybe I should consider that as an (a) and (b) as part of the same thing, since the mechanism of damping (lowered Q) is identical.
I've added strikethroughs & italics to edit the statement (C is internal to the tube, plus stray capacitance due to wiring, L is stray inductance mostly due to wiring).
Ah, yes, I see. Thanks for clarifying that.
The SSR:
4) Prevents cathode-to-screen arcing, as the amount of current that can pass through the SSR is limited, so the arc can never get started.
I'll take your word these resistor help in this regard; I've never found this to be a problem to solve in audio amps I've owned.
It's interesting you picked up on that.
I know that the EICO and Dynaco amplifiers did not include SSRs and were not reported to suffer from screen arcing. So there's that data point.
Reading tube papers from the 1950s and 1960s turned up a lot of work was done on reducing cathode-to-plate arcing because of reduced inter-electrode distances. Sylvania touted its new sarong cathode design as lowering that risk. (I can't find much about it in practice, only some ads and a few patents.)
But other reports of damage listed screen arcs as a problem, and it was a casual toss off of something like, and oh, with these higher plate voltages screen arcs are showing up as well as cathode arcs.
I've also read complaints that modern replicas of older designs, the KT88 was one I remembered, suffer from screen arcs.
It just seemed that an inexpensive resistor can save expensive tube (see below) without any consequence, so why not do it?
Typically the speaker's lowest impedance is equal to the nominal impedance used to design the output stage. That is, the speaker only ever looks like "more impedance" than designed, not less.
Separately, the screen resistor has most impact on the tube when the plate current is very high and the plate voltage is momentarily swinging very low, which results in the screen being a couple-hundred volts higher than plate voltage.
Screen voltage doesn't sag to be "close to the plate's potential" at that moment, otherwise you'd be in triode mode and have very much less power output.
Ahhh, thanks for clearing that up.
If I understand this, the speaker is sucking up piles of power, driving B+ into the ground. But the separate screen supply is still fine, since it's not loaded. That's how B+ and Screen cross.
I was thinking of the screen as being derived from the B+, thinking that the screen uses a voltage divider, and that it has some capacitance to filter out fluctuations so I could see how it would have a reserve which would cause it to not as rapidly sag like B+.
The SSR:
1) Limits cathode-to-screen current, protecting each from excessive. Excessive cathode current caused by the omission of an SSR can cause the cathode bias resistor to burn out as well as damage the tube.
I personally wouldn't look to the screen resistor to limit cathode current; G1 voltage is much more effective in that role. It is also a benefit of cathode bias that the cathode resistor would burn open before the tube could destroy itself (or burn other components).
Guilty, but I plead poor wording!
My thought was that if B+ sags and the screen becomes a preferential plate, then limiting the current will prevent damage to the screen, saving the tube. Label that "goal".
That additional current would also cook the cathode resistor, but this is pretty much a "who cares" because the twenty cent resistor is negligible. It is the cooked tube at many orders of magnitude higher that is upsetting. It was more of an aside, not a means to protect a resistor. It was a, hey, look, the cathode resistor doesn't go up in flames! Woo-hoo! Bonus!
Thanks for the commentary. It explained some puzzling bits and pieces.
-
> a list of all the purposes I found for the GSR
Understood and valuable that way. Some of those I never thought of.
> Williamson's paper ..., shows a 1 k
You are quite correct. I am mis-remembering something, but I can not remember what. There IS a paper or plan suggesting more like 47K to G1. Roughly half the value of the grid-leak, though not too critical and must be compromised with treble extension (NFB stability). There is a value which will make the over-slammed average grid bias nearly the same as the idle grid bias. Anywhere close to this is a lot better than real low series resistance. You would think that driver impedance counts, but I'd have to think on (or sim) whether this resistance should be after the coupling cap.
> Interesting how very small changes ruin the Q.
"Small" is relative. At the plate side, you think the OT is a 6Kohm load. At >1MHz (our tubes are good for much more) it is a 1,000pFd lump with an inducty wire to it. So about 166 ohms reactance. 50 Ohms added is "nothing" to a 6K audio load but significant damping on a 166 Ohm tank. Grid-side capacitance is much smaller, higher impedance, so few-K even 1K may be significant damping in what looks like a "100K" interface.
> wrote that amplifier designers routinely neglect this sort of oscillation
I do not think this is quite right. We try to ignore it, and also avoid inviting it, but oscillations DO show up uninvited. When the waveform is bad or the THD number is high, we look for it. Generally a supersonic oscillation will screw-up the bias or the audio gain and give poor performance. I suspect Patrick has much more experience reviewing the work of "boutique" amp builders who may do less bench-testing.
_________________________
> papers from the 1950s and 1960s ...cathode-to-plate arcing ... Sylvania touted its new sarong cathode design
I'd like to see that.
My snap-thought is that _TV_ sweep-tubes are prone to hi-volt arcs. Many of these started as "audio" tubes so may have needed revision for KV spikes.
Googling "sarong cathode" -- "SYLVANIA- 6AU4.. damper tube. Consider, for one feature, the SYLVANIA SARONG CATHODE and how it adds dependability to tube life."
6AU4 is a high-abuse diode. However Sylvania promised to use it throughout the receiving tube line, starting with 6BZ7 6BQ7 6BC8 and 6BS8 (tuner types), for less noise. It seems to replace sprayed oxide cathode with a film applied (wrapped?) on the cathode. In that a film-treatment line may be more uniform than individual spray-jobs, it may be more uniform. I also think that work at the tube factory was dull, and sales-writers would take-off on the smallest details. Maybe the Lamour movie (https://upload.wikimedia.org/wikipedia/commons/b/b2/Dorothy_Lamour_in_Road_to_Bali.jpg) just came through town. A cynic might wonder if making cathode by the mile was just cheaper than spraying it an inch at a time.
Mentioned in Reference Data for .... Van Valkenburg (https://books.google.com/books?id=R67HARlhisYC&pg=SA16-PA3&lpg=SA16-PA3&dq=sarong+cathode&source=bl&ots=TdXEdQEWzM&sig=3Q4L16nEthuIlYMknlh80e-DC1g&hl=en&sa=X&ved=0ahUKEwjLotXx6u_RAhWL3YMKHZnwDFs4ChDoAQg7MAk#v=onepage&q=sarong%20cathode&f=false).
-
... Realistically, a EL34's grid is not much smaller than a 6L6's (anybody got some duds to crack?) and will take some current flow. ...
If only they had a broken 6L6 to compare side-by-side with this EL37 (http://www.r-type.org/articles/art-057.htm).
G1 wire is a bit smaller than G2 wire, but not 50% smaller. G1 is closer to another element than G2, but it's also wound on a finer pitch than G2 so presumably more rigid. Carpentry-by-eye suggests at least a part-watt should be safe (especially since G2 is rated for 6w (http://www.r-type.org/pdfs/el37.pdf)).
EL34 being an evolution of the EL37, I suspect internals are roughly similar.
-
What is interesting is the claims that oscillation robs power. Patrick Turner wrote that amplifier designers routinely neglect this sort of oscillation and proposed adding Zobels to each stage to damp it out. You have any feelings about this? ...
Oscillation is output power, just not usable output power unless you meant to build an oscillator.
You don't need Zobels at every stage in an audio amp; every stage already has a high-value resistor as a load. So that would tend to dampen the plate circuit, and colloquially the recommendation is to make plate wires long to the coupling cap, then keep grid wiring after the cap short (to reduce stray L & C at the next stage's grid).
Radio RF & IF stages don't have high value resistors but use coils. However, they also add peaking caps to tune the output load to a single frequency. Other frequencies (including possible oscillation) are not at the circuit's resonant frequency, and develop very little power/voltage at the output circuit.
"Zobels at every stage" may be taking a radio idea and misapplying it to audio.
I was trying to differentiate between RF that gets into the system from outside and oscillation that is generated by the tube itself. They both get damped, but they arise in different ways. Maybe I should consider that as an (a) and (b) as part of the same thing, since the mechanism of damping (lowered Q) is identical.
I don't perceive it as "oscillation generated within the tube" because the tube is always controlled by the voltages & impedances at its pins. And if you get a tube that can't be controlled by those 2 factors, it is defective somehow & needs to be tossed.
So it's always about what you do externally, even if you're trying to counteract an internal impedance, like interelectrode capacitance. We can't get inside the tube to fight that (we select a different tube in which the manufacturer already fought this battle to come up with a different element geometry).
The SSR:
4) Prevents cathode-to-screen arcing, as the amount of current that can pass through the SSR is limited, so the arc can never get started.
Reading tube papers from the 1950s and 1960s turned up a lot of work was done on reducing cathode-to-plate arcing because of reduced inter-electrode distances. ...
Maybe I've been lucky, but the only arc I ever experienced in an audio tube was a plate-to-heater short in a 6L6GC. It happened from pin 3 to pin 2, and was almost certainly an issue with the base (or its pins). These elements are otherwise as far from each other as it gets within the tube...
-
If I understand this, the speaker is sucking up piles of power, driving B+ into the ground. But the separate screen supply is still fine, since it's not loaded. That's how B+ and Screen cross.
I was thinking of the screen as being derived from the B+, thinking that the screen uses a voltage divider, and that it has some capacitance to filter out fluctuations so I could see how it would have a reserve which would cause it to not as rapidly sag like B+. ...
B+ doesn't sag at the output tube plate the way you envision.
The output transformer is relatively low resistance (little D.C. Volts drop) but a higher impedance for audio (big A.C. Volts drop). The output tube control grid (G1) controls plate current, which is pulled through the OT primary. So while there is a small voltage drop from B+ to tube plate at idle (d.c), there are much bigger voltage drops when audio is applied to G1.
The voltage drop across the OT primary due to audio depends on the size of the plate current and the OT primary impedance; Volts (dropped) = Current (plate) * Resistance (OT primary impedance). B+ might stay the same, but volts dropped across the OT primary changes with the plate current swing. Resulting plate voltage = B+ - volts dropped across OT.
You may be thinking about a power supply with a non-zero internal impedance (which is just about all of them, except a regulated supply well within its limits). If average current drawn from the power supply rises (as it will a tiny amount with a Class A amp, but a very large amount with a Class AB amp), then B+ will sag some in proportion to the supply impedance and increase of average current draw. But this B+ sag is distinct from output tube plate voltage swings (though very severe sag could reduce possible power output).
Separately, the peach-colored line in the graph at the bottom-left of page 1 here (http://www.eminence.com/pdf/Cannabis_Rex.pdf) shows speaker impedance vs frequency. You'll see this 8Ω speaker is only really 8Ω between 200-400Hz, and much more everywhere else (below the bass-resonant frequency doesn't count in this case, because this is a guitar speaker).
I was thinking of the screen as being derived from the B+, thinking that the screen uses a voltage divider, and that it has some capacitance to filter out fluctuations so I could see how it would have a reserve which would cause it to not as rapidly sag like B+.
Only the budget amps use the same filter cap for B+ feed to the OT and to supply the screen. So B+ sag will reduce the source voltage for the screen, in most audio/guitar amps. A cap can't sustain a sagging supply voltage for long on its own.
My thought was that if B+ sags and the screen becomes a preferential plate, then limiting the current will prevent damage to the screen, saving the tube. Label that "goal".
Average Fender amp:
6L6 or 6V6 with plate & screen at 400vdc (or more). The 6L6 or 6V6 might need 50-90vdc on the plate at max plate current (this is at the knee of the 0v gridline on the data sheet). Assuming no B+ sag, the plate swings to 90vdc (or a bit lower) while the screen is at 400vdc (or more), every cycle of the audio output.
The screen already is the preferential plate, which is why screen voltage impact plate current more than plate voltage (and why beam power tubes & pentode characteristic curves do not look like triode characteristics).
But the screen is a coarse mesh of wire; most electrons fly right past the screen and into the plate. In beam power tubes, G2 is wound "in the shadows" of G1, and so the streams of electrons are less likely to hit G2 than in true-pentodes. And this last bit partly explains why beam power tube characteristics look a bit different than pentode characteristics.
-
There IS a paper or plan suggesting more like 47K to G1. Roughly half the value of the grid-leak, though not too critical and must be compromised with treble extension (NFB stability). There is a value which will make the over-slammed average grid bias nearly the same as the idle grid bias. Anywhere close to this is a lot better than real low series resistance. You would think that driver impedance counts, but I'd have to think on (or sim) whether this resistance should be after the coupling cap.
After
When the amplifier is first turned on B+ is live but the heaters are still not hot enough to cause cathode emission. The coupling capacitors have Vp on one side and can charge through the grid, as this path has the lowest potential on the other side, arcing to the plate. Rg is too large to permit such charging. So the GSR limits that phenomenon.
> Interesting how very small changes ruin the Q.
"Small" is relative. At the plate side, you think the OT is a 6Kohm load. At >1MHz (our tubes are good for much more) it is a 1,000pFd lump with an inducty wire to it. So about 166 ohms reactance. 50 Ohms added is "nothing" to a 6K audio load but significant damping on a 166 Ohm tank. Grid-side capacitance is much smaller, higher impedance, so few-K even 1K may be significant damping in what looks like a "100K" interface.
This was very interesting so I looked it up. (Quote from a Navy manual removed as it pushed me over the 5,000 character boundary.) Ok, that makes sense.
You cited 1,000 pF as the normal inter-electrode capacitance. That suggests 100 pF at µ = 10. Isn't that too high?
Common output tubes seem to run about 1 to 2 pF grid-to-plate. So we could maybe treble that for the entirety?
The reactance would be:
XC = 1 / (2 × Pi × f × C)
At 1 MHz:
= 1 / (2 × 3.14 × (1×10^6) × (1 x 10^-12 x C in pF))
= 160 kΩ for 1 pF (assuming 3 pF inter-electrode capacitance @ gain = 1)
= 32 kΩ for 5 pF (assuming 1 pF inter-electrode capacitance @ gain = 5)
= 5 kΩ for 30 pF (assuming 3 pF inter-electrode capacitance @ gain = 10)
At 10 MHz the above numbers drop by 1/10 (only change is f in denominator, so 1 / (delta f)) :
= 16 kΩ for 1 pF (assuming 3 pF inter-electrode capacitance @ gain = 1)
= 3.2 kΩ for 5 pF (assuming 1 pF inter-electrode capacitance @ gain = 5)
= 0.5 kΩ for 30 pF (assuming 3 pF inter-electrode capacitance @ gain = 10)
That's small, so I can see how a few hundred Ω for the GSR begins to materially affect the RC circuit at 1 MHz, and significantly so at 10 MHz. Depending upon capacitance it is anywhere from roughly about 10% to 50%.
-
Was over the 5k limit so split response in two.
> papers from the 1950s and 1960s ...cathode-to-plate arcing ... Sylvania touted its new sarong cathode design
It seems to replace sprayed oxide cathode with a film applied (wrapped?) on the cathode. In that a film-treatment line may be more uniform than individual spray-jobs, it may be more uniform. I also think that work at the tube factory was dull, and sales-writers would take-off on the smallest details.
I spent many hours trying to track this down beyond the obvious details and verify the claims. Very little technical information is available other than the patent which is about what one would expect.
The Sarong cathode impregnates a combustible film with Ba and Sr and then wraps the cathode in said film, which burns away during activation. Much like how the cathode and getter are applied. Now, because this is a film (hence "sarong") the deposition is very even, unlike the dipped cathode which, at the microstructure level, is quite rough.
A smoother surface reduces or avoids the point-source emissions documented by Schade. Overall cathode damage reduces when the islands of charge emitting electrons are contiguous instead of isolated with point-source emissions.
Think of this as a golf course where the high spot gets struck my lightning. Flattening out rocky terrain reduces the likelihood of an arc. That is the superiority of using a film deposition.
My snap-thought is that _TV_ sweep-tubes are prone to hi-volt arcs. Many of these started as "audio" tubes so may have needed revision for KV spikes.
All tubes will arc, and the rougher the cathode the more likely the arc. Increasing the voltage makes an arc even more likely. TV sweep tubes tended to burn out with alarming regularity. That was why damper diodes were created. Interestingly enough, there is no reason why the low-cost damper diode cannot replace a higher-cost rectifier, assuming the voltage drop and current load are appropriate.
Googling "sarong cathode" -- "SYLVANIA- 6AU4.. damper tube. Consider, for one feature, the SYLVANIA SARONG CATHODE and how it adds dependability to tube life."
Rectifiers are particularly prone to arcing and were the leading cause of failure, then as now. Mostly because designers either did not read or did not understand the specifications or, if read and understood, dishonored them as cost-savings measure and pass the cost of failure onto the consumer.
I doubt many consumers then or now could tell the difference between a long-lived design and a short-lived one. Tubes are seen as a consumable, and people don't realize the lifespan is often shorter because corners have been cut on the design.
-
> All tubes will arc
Anything will arc if you put enough voltage across it.
I don't expect arcs in audio amplifiers with appropriate devices. Guitar amps push the assumption because we "do" flog them into reactive kicks which may be 10X the nominal plate voltage.
TV H-sweep amps do exactly the same but non-stop. Not a surprise the early TV designs used audio tubes, stock or with plate-caps. The "360V" limit on the early 6L6 was about the base insulation, not the plate; top-cap 807 had much higher limits.
Yes, arcs are a continuing problem in amp repair, but due to wimpy sockets and especially PCB construction, and lack of respect for proper spacing.
> why damper diodes were created.
The damper diode recovers energy from the H-sweep coil and incidentally uses that for B+Boost (especially in tranformerless sets with low prime B+). Same as free-wheeling diodes in modern fly-back switchers. They also limit the voltage on the switch, but mainly they recover stored energy.
> I doubt many consumers then or now could tell the difference between a long-lived design and a short-lived one.
_I_ can't. In 2001 I got a Honda because they are usually long-lived. It worked good at first. After only 14 years, "something" has gone wrong in the idle, I have spent several months and near $2K diagnosing and replacing, and it's still bad. Before that I had a PoS '79 Ford which ran 1.5X as long and twice the miles, all easy/simple repairs but just got tired of changing everything with an "OR" on the end of its name.
> Tubes are seen as a consumable
Early tubes were light-bulbs, made on lamp machinery, and "obviously" needed replacement. Technical progress could have led to permanent tubes. Economic penny-pinching left us with socketed tubes so a factory could assemble radios and then tube them at the very last moment (even at the dealer) depending on the lowest bids they got and the availability of over-stock and grey-market tubes.
Tubes today are so very cheap that I will live with the risks. If we truly wanted tubes to live forever, we would pay for the best Russian factory to take their time, and we would not run B+ over 250V or Pdiss over 60% of Design Center. A large number of preamp tubes incidentally run far-far below ratings and generally do "last forever", or until hasty welds shake apart. 1940s radios often have bad caps and resistors (nominally non-replaceable parts) gone bad but 4 or 5 tubes working fine (maybe the rectifier burnt when the filter cap went short). If we ran our 6V6es at 250V and 7W, and don't bump them, they may play a lifetime.
-
... All tubes will arc, and the rougher the cathode the more likely the arc. ...
Agreed, but echoing PRR's first statement, I'm not sure we have enough voltage present in most guitar/audio amps to cause the arc.
It's always worth pointing out the amp's of the early 50's (and before) which didn't have G1 or G2 resistors, which are still plugging away just fine. I used to have a '54 Princeton (http://el34world.com/charts/Schematics/files/fender/Fender_PRINCETON_5B2.pdf) and a '55 Tremolux (http://el34world.com/charts/Schematics/files/fender/Fender_TREMOLUX_5E9A.pdf) which fell in this category. Both amps also used a 1-wire heater system and other "no-no's".
Each amp made it 50 years with no special attention paid to arc prevention. I don't know, but I suspect arcing for guitar amps using typical B+ voltages is due mostly to factors outside the tube (wiring, dirty sockets, dust/cobwebs, plus humidity) or by factors inside the tube that amount to defects (trace substances on mica surfaces, insufficient holes in micas to lengthen leakage paths, loose material falling on to base pins inside the tube).
-
If we're already hitting the output tube with a signal bigger than it can handle, there could be distortion due to grid blocking, grid curve convergence below the knee, plate current saturation, grid curve crowding near cutoff, and other effects. If we can't be sure (at least I'll say, "I'm not sure") screen voltage drop is actually contributing to the distortion, why bother claiming it?
Because in Neumann & Irving's Guitar Amplifier Overdrive they state that screen voltage changes cause not only the tube's operating point to change, but it also causes the tube's transfer curve to actually change shape which generates more nonlinear, harmonic and intermodulation distortion than grid current and grid voltage change (bias drift). So yes, other distortion is going on but screen current generates most of our beloved power tube distortion.
Got the book, figured out the discrepancy: Preamp pentode scenarios misapplied to output tubes.
Simply, a preamp pentode has a relatively-large screen resistor followed by a cap to ground/cathode, while the output tube has a relatively small screen resistor followed by the screen itself (no cap). There is no "kinda small" cap at the screen itself to drop charge due to voltage drop at the screen resistor and hold-over a screen voltage shift past the current change which caused the initial voltage drop.
I've got some gripes with the book, which is easy enough to understand when you already understand the subject matter but makes some assertions that can be misinterpreted or misapplied.
-
There may be something you can read.
I've been reading sir but I got lost in the RADAR part :icon_biggrin:
-
"... the effect of altering the screen voltage on a pentode. As it is progressively reduced, the gm (and therefore the voltage gain) progressively falls, while the input sensitivity increases. The basic 'shape' of the curves remains broadly the same, but the grid curves get 'squashed down'. Recognising and understanding this effect is fundamental to understanding pentodes / tetrodes, ... "
From Merlin's page on the effect of screen voltage on gain and transconductance in a pentode
http://www.valvewizard.co.uk/EF86.html (http://www.valvewizard.co.uk/EF86.html)
-
Grid stopper resistors and screen resistors...clarification (http://el34world.com/Forum/index.php?topic=21489.msg228044#msg228044)
Are we approaching clarification? :icon_biggrin:
-
Grid stopper resistors and screen resistors...clarification (http://el34world.com/Forum/index.php?topic=21489.msg228044#msg228044)
Are we approaching clarification? :icon_biggrin:
I live in hope
-
"... the effect of altering the screen voltage on a pentode. As it is progressively reduced, the gm (and therefore the voltage gain) progressively falls, while the input sensitivity increases. The basic 'shape' of the curves remains broadly the same, but the grid curves get 'squashed down'. Recognising and understanding this effect is fundamental to understanding pentodes / tetrodes, ... "
From Merlin's page on the effect of screen voltage on gain and transconductance in a pentode
http://www.valvewizard.co.uk/EF86.html (http://www.valvewizard.co.uk/EF86.html)
Roger, that. But what about:
Got the book, figured out the discrepancy: Preamp pentode scenarios misapplied to output tubes.
-
Seems like there is some confusion between what occurs with dynamic changes in screen current during a signal cycle in an output stage, with setting overall screen idle voltage?
If overall screen voltage is decreased (i.e. throughout the signal cycle), there will (all other things being equal) be an overall reduction in tube current, and therefore an overall reduction in the amount of current transferred to the OT secondary, and therefore an overall reduction in output power. The amount of change in plate current for a given amount of change in grid voltage would also decrease, lowering the input sensitivity and gain. Surely?
I'm happy to be corrected in any misconception I may have, or to discuss the minutiae.
-
... If overall screen voltage is decreased (i.e. throughout the signal cycle), there will (all other things being equal) be an overall reduction in tube current ...
It is insufficient to cite general "If-Then" statements in this way for output tubes. There has to be context provided, because "when does screen current increase" (in order for screen voltage to decrease)? It's not "throughout the signal cycle"...
The attached graph for 6L6 G2 current shows for a given screen voltage, screen current doesn't radically increase until plate voltage is pulled down to the region that would be at/below the knee of the grid curves of the tube.
As a worst-case example, look at the G2=350v curve in the 1st attachment, and compare the screen current at 500v plate vs. 75v plate: it has increased 20mA (from 22.5mA to 42.5mA). If the typical 470Ω screen resistor is used for this tube, the screen voltage drop is only 470Ω * 0.02A = 9.4v!!
Now look at the 2nd attachment:
A Blue line is erected at 75v plate, to the G2=350v curve. The Red line segment approximates 1/5 the distance downward to the G2=300v curve (for a ~10v reduction in screen voltage). The 2 horizontal lines show the plate current for each condition, which drops only ~10-12mA.
This reinforces why manufacturers typically use small screen resistors: because there is little impact to screen voltage, even when there are screen current peaks during peak positive grid-input, corresponding to peak plate current and a plate voltage minimum peak. During the rest of the signal cycle, screen current stays relatively constant.
This also explains why I've said for years that "too big" screen resistors tend to choke back peak plate current and can yield a compressed sound when playing an amp full-tilt.
But my Comment in Reply #44 was in response to Rob's "catch-all statement (http://el34world.com/Forum/index.php?topic=21489.msg228066#msg228066)" for power tubes taken from his website (https://robrobinette.com/Tube_Guitar_Amp_Overdrive.htm#Screen_Currents_Affect_On_Pentode), and he cited the book Guitar Amplifier Overdrive as the source for his statements. It was not until buying & reading the book that I found the book does not cover output tubes at all, and the source material was taken from a section on preamp pentodes.
Preamp pentodes use a different circuit configuration than output tubes, have screen resistors several-times bigger than the plate loads (where output tubes have screen resistors that are a fraction of the plate load), and the tubes themselves are evolved to have different characteristics (power tube mu under 15 is normal compared to a normal preamp pentode mu over 150).
Additionally, the book is about what happens when a grid input slams a tube stage way beyond its normal operating range (it was hard at first to take seriously the authors ramming a 20v peak input into a stage bias at -1v).
It was therefore inappropriate to extend the judgments summarized in that book to output tubes.
-
I have searched the internet & find little hard core info on the screen voltage, other than what Hotblue already posted re small bottle pentodes; EXCEPT for posts by tubeswell on this and other forums. As Hotblue has already pointed-out, it's necessary to distinguish between preamp vs power pentodes.
Tubeswell: as a practical matter what actual circuit do you propose to lower screen voltage? The Tube Cad Journal, among others, advises against voltage dividers. KOC advises rather large 2.2K screen resistors for EL84's for overdrive tone. The tone and "feel" didn't work for me, ruining clean and overdrive tone. This corroborates Hotblue's comments above.
Here's more rambling: http://music-electronics-forum.com/t19103/ (http://music-electronics-forum.com/t19103/)
Of course there's UL but that's a different animal than mere screen voltage reduction, involving feedback & optimal winding taps.
-
Tubeswell: as a practical matter what actual circuit do you propose to lower screen voltage? The Tube Cad Journal, among others, advises against voltage dividers. KOC advises rather large 2.2K screen resistors for EL84's for overdrive tone. The tone and "feel" didn't work for me, ruining clean and overdrive tone. This corroborates Hotblue's comments above.
For power tubes, you can run the screens off a separate winding, which is independently rectified and filtered, like an Ampeg SVT or this old circuit (attached) from a 1960s NZ tube amp. This keeps the screen supply a bit stiffer and improves output efficiency.
-
Many power tube specs call for screen voltage about 2/3 of plate voltage when the plates are run near max; otherwise they can be near equal. Some power tube designs seem to always call for lower screen voltages, like 7027a and KT88. It's not clear how this supports your theory.
BTW: there are aficionado's of lower scree voltage. Sometimes a zener is used to maintain a lower screen voltage.
-
like 7027a and KT88.
I tried fixing G2 on my breadboard with a KT88, and I didn't like the results, although I wasn't running the 88 near meltdown, but I think there are non equal comparisons between big bottle PA and pentodes in the pre
-
Many power tube specs call for screen voltage about 2/3 of plate voltage when the plates are run near max; otherwise they can be near equal. Some power tube designs seem to always call for lower screen voltages, like 7027a and KT88. It's not clear how this supports your theory. ...
Tubeswell is 100% right that a screen voltage lower than plate voltage could be derived from a separate supply.
Where data sheets or practical amps have screens at a significantly-lower voltage than the plate, it is because a higher screen voltage is not necessary to generate the needed peak plate current (which happens when the plate voltage swings to a minimum, because plate current is at maximum and is dropping significant voltage across a load).
To reiterate, my post on March 6 was a reply to Robrob's prior post, which took statements in a book summarizing preamp pentode behavior and extrapolated it to imply some similar behavior in output tubes, based on changing screen voltage. Therein was the error.
(Didn't help that the book was under-whelming, and that the authors tried to claim pentodes behaved differently than triodes, while glossing over the fact they input 0.6v peak to a pentode biased at -1.3v, and compared it to a triode biased at -1v but slammed with a 20v peak input signal. "The pentode doesn't suffer from grid-diode clipping..." Well, duh, when you're not slamming the grid to melt-down like you did with the triode.)
-
> If only someone like Tesla had designed the guitar amplifier.
Yeah, I know. Tesla invented anti-gravity, the notes were taken to Area 51 to build the flying saucers which appeared "secretly" in the early 1950s, and are the forerunners of the flying saucers we all drive today.
:l2: :l2:
-
> If only someone like Tesla had designed the guitar amplifier.
Yeah, I know. Tesla invented anti-gravity, the notes were taken to Area 51 to build the flying saucers which appeared "secretly" in the early 1950s, and are the forerunners of the flying saucers we all drive today.
:l2: :l2:
OK PRR put the cognac down..
-
it's been 5 years, pretty sure he's finished by now :laugh:
although if someone gave me a cognac, it'id still be sittin, un-touched. never developed a taste for something that smelled like butane :icon_biggrin: