Ben, reading that post felt a bit weird, as though I was reading one of my own old posts I'd forgotten about! Literally everything you mentioned in your post has crossed my path as well!
I too tend to leave the volume at max on my guitars, accepting the reduced versatility of the instrument that comes with this. (Several of the most talented guitar players seem to constantly vary their guitar volume and tone as they play, allowing them continuous variations in tone colour. Jeff Beck is one example.)
Many years ago I built a JFET preamp into one of my guitars, and powered it down the guitar cable - the JFET's drain load resistance was inside the guitar amp, not inside the guitar, so the guitar cable was carrying the drain current and powering the JFET. (Like phantom powered microphones, which I didn't know about at the time.)
It worked, but that guitar would only work with that one custom amp. Not a practical solution!
I discovered this exact thing when I built my own guitar (including pickups) in my twenties. I didn't know what I was doing, and wound far too few turns of wire. The guitar ended up sounding clean and more like an acoustic guitar than an electric guitar.
But there have been at least two commercial solutions to the "flat, low impedance" guitar pickup problem. Decades ago, Alembic made bass guitars with low-impedance, flat-response pickups, and then put a tunable EQ inside the guitar that re-created that out-of-band resonance peak. Because the peak was now electronically generated, it could easily be shifted in frequency or Q, so the bass guitar was rather versatile.
More recently, there are the Fishman "Fluence" pickups. The "coil" is a few stacked printed circuit boards with only a few printed turns of copper on each one, so the whole coil must be very low impedance. But there is a small active preamp built into each pickup.
No technical details have been released, for obvious reasons, but my guess is that these preamps do more or less what the Alembic onboard preamp did four decades ago - boost the signal to comparable level with high-impedance pickups, and EQ it to sound like a traditional high-impedance pickup.
You have probably seen the aftermarket tone control mod sold for electric guitars: it replaces the standard "tone" pot with a 12-way rotary switch carrying a dozen small-value surface-mount capacitors. Each switch position puts a different capacitance in parallel with the guitar cable, changing the resonance frequency of the pickup coil, and so altering the tone. Some guitarists speak very highly of this approach, but I've never tried it.
Just about every time I struggle with hum and buzz from my single-coil guitars, I wish Leo Fender had thought to put a centre-tap on his pickup windings, ground it, and run balanced output signals to an XLR output jack, like a microphone. But he would have had to use both halves of a 12AX7 to make a balanced input stage, and ol' skinflint Leo would never do that, even if he had known what a differential amp was!
-Gnobuddy
I would not agree in this aspect. Certainly this old-time high-impedance passive guitar circuitry is sensitive to noise currents due to insufficient shielding.
But in a classical setup with a tube amp most of the hum originates from the mains transformer and is picked up magnetically.
Over the years I did many test on that issue, and my recommendation is
-replace all single coils by stacked humbuckers. I go for EMG select, they are perfect for me and cost only 35Euros/each.
-to avoid any cable influence on sound and to enable the guitar vol pot do insert a JFET buffer stage. This typically draws 200uA and with a 9V battery lasts for ages. You may consider replace the battery by a 4V lithium - will last for 2 ages, at least.
-the capacitor switch makes sense. Personally, I do not like a switch as much and prefer a pot. So I designed an artificial rotary capacitor tuned by a pot. Typically I can tune the resonant peak between 1kHz and 5kHz, which is appropriate imho.
All in all I am happy using the guitar vol pot extensively without deteriorating the sound.
Nevertheless - these things are certainly a matter of personal taste
Last edited:
Agreed. Unfortunately, these days there are many other sources of noise to worry about. For example, our plasma TV does not get along with single-coil guitar pickups at all. Same for my active monitors (speakers), which have a switching power supply inside.
Leo Fender's 1950s single-coil pickups were pretty blatant copies of an even older guitar pickup design by De Armond, which he began selling in 1939. I'm attaching an image of the patent application, taken from a Premier Guitar article on the subject. As you can see, the De Armond pickup uses six Alnico pole pieces, a slim elliptical coil, a single-coil winding with 42 gauge wire, virtually every feature of the later Fender pickups. However the De Armond could also be mounted in the soundhole of an acoustic guitar.
Unfortunately for us, every year there is still more electrical and RF interference (for example, WiFi is everywhere now, along with Bluetooth, cellphones, wireless keyboards and mice, etc). A lot of this is relatively recent, but I'm sure new sources of electrical noise and interference have been showing up for many decades now (AM radio, FM radio, police radio, CB radio, etc, etc).
I think that wonderful old 1930s De Armond pickup design is not coping very well with 2016 levels of noise and interference.
Good suggestions! Most of my guitars are already humbucking types. But I always have one 'Strat-type guitar in the collection - they have a unique sound, and it can be very beautiful in the right hands (like Mark Knopfler's).
That sounds like an excellent solution!
On the Aussie Guitar Gearheads forum, a brilliant fellow named Roly Roper posted some schematics for a similar idea. He used variable bootstrapping to change the effective input capacitance of his JFET preamp.
Roly passed away from cancer a few months ago. I've never met the man or even spoken with him, only seen his posts on the AGGH forum. But I felt his passing deeply. It's rare to encounter someone with so much knowledge, so much intelligence, and the desire to use all of it to help other people.
-Gnobuddy
Attachments
thank you for the kind reply. Yes, it is very interesting to see how old some knowledge is sometimes.
Just to mention the artificial rotary capacitor: This is by far not my idea. I was inspired by reverse engineering a VOX Wah-Pedal about 1970. At that time I found a fixed choke and a pot controlling the parallel capacitance - all together formed the well known tunable resonance filter. Very simple, very effective (but very noisy
). Anyway the basics of the circuit I use today.
Concerning EM disturbation everywhere - you are right, these things seem to explode nowadays.
My job relates to EM-measurements and so I do care for design rigid in this modern world. Specially I check my circuitry with a calling cellphone close to the pcbs. There is a lot to be learned!
Just to mention the artificial rotary capacitor: This is by far not my idea. I was inspired by reverse engineering a VOX Wah-Pedal about 1970. At that time I found a fixed choke and a pot controlling the parallel capacitance - all together formed the well known tunable resonance filter. Very simple, very effective (but very noisy
). Anyway the basics of the circuit I use today.Concerning EM disturbation everywhere - you are right, these things seem to explode nowadays.
My job relates to EM-measurements and so I do care for design rigid in this modern world. Specially I check my circuitry with a calling cellphone close to the pcbs. There is a lot to be learned!
Last edited:
I'm sure someone's going to snigger at this but I came across a valuable point...luckily before the lesson was taught. As my amp sits now I had the output wires directly tied to the speaker. OPT>speaker.....Well I wanted to add an output jack to hook into a buddies 2x12 this weekend and I realized the ground terminal of the jack would be tied to chassis....oh crap....the ground of my output wires isn't currently tied to chassis.
While under *most* circumstances it wouldn't be of issue, I could see that in the event the OPT shorts out that I could all of a sudden have B+ on the output wires. Chances are speaker would probably go splodey-poof but it was just another "path" that I hadn't thought of. Sure as ****, pull up about every guitar and HiFi schema I could think of and all the outputs are grounded. So hopefully tonight I'll have a safer amp AND play it through a bogner 2x12!
While under *most* circumstances it wouldn't be of issue, I could see that in the event the OPT shorts out that I could all of a sudden have B+ on the output wires. Chances are speaker would probably go splodey-poof but it was just another "path" that I hadn't thought of. Sure as ****, pull up about every guitar and HiFi schema I could think of and all the outputs are grounded. So hopefully tonight I'll have a safer amp AND play it through a bogner 2x12!
Last edited:
There'll be no sniggering from me, I can tell you. My amp is serving me up large spoonfuls of humility lately, as every "clever" thing I did when I built it, seems to be backfiring, and I find myself taking them out, one by one!
I'm glad your amp is safer now. How did it go with the Bogner cab? (A few years ago, I had the chance to play through a Bogner Alchemist 2x12 in a Best Buy. The amp was bigger than a dormitory 'fridge, and I remember that it was a wonderful sounding amp, even at very low volume.)
-Gnobuddy
I apologize for my part in that, Printer2.
I'll bring it back on topic with a question I'm facing: why are output valve grid stopper resistors so small, typically just a few kilo ohms? Calculation says that much, much higher values can be used (and should reduce blocking distortion).
An example: the 6V6 datasheet lists 9 pF input capacitance. Let's say we had terrible layout and added 11 pF of stray capacitance, pushing the total to 20 pF. Let's say also we want a -3dB bandwidth out to 16 kHz (very generous for a guitar amp). That gives us a maximum grid stopper value of 500 kilo ohms! (In practice, say 180k for the grid stopper, 220k for the grid bias resistor, adding to 500k total.)
Why are people using 1k and 4.7k grid stoppers for these output pentodes, when 100k would be perfectly safe, and reduce blocking distortion?
-Gnobuddy
I'll bring it back on topic with a question I'm facing: why are output valve grid stopper resistors so small, typically just a few kilo ohms? Calculation says that much, much higher values can be used (and should reduce blocking distortion).
An example: the 6V6 datasheet lists 9 pF input capacitance. Let's say we had terrible layout and added 11 pF of stray capacitance, pushing the total to 20 pF. Let's say also we want a -3dB bandwidth out to 16 kHz (very generous for a guitar amp). That gives us a maximum grid stopper value of 500 kilo ohms! (In practice, say 180k for the grid stopper, 220k for the grid bias resistor, adding to 500k total.)
Why are people using 1k and 4.7k grid stoppers for these output pentodes, when 100k would be perfectly safe, and reduce blocking distortion?
-Gnobuddy
There have been amps that go into the 10K to 50K range for a grid stopper on an EL84 to prevent oscillation. My experiments with big grid stoppers ( a long time ago) convinced me that they took something away from the tone, but I can't tell you now what it was. At the time I had purchased a large batch of cheap Chinese KT88's so that's probably what I tested. Maybe it's time for some more experiments.
The maximum grid circuit resistance spec shown in data sheets may not be conservative enough with some of today's new production tubes. Tubes that have a less than optimum vacuum will start to draw grid current as they age. this current will work to overcome the negative bias unless it is bled off (leaked) through the grid circuit resistance. Grid current is the most common cause of output tube runaway (red plate) type failures.
There are several popular "boutique" guitar amps (Soldano) that use large grid stoppers particularly in the preamp where values over 100K are seen.
Amp 1.5 got wired up and tested and I was not too happy with the results. I set the chassis on top of the speaker cabinet and started playing it. There was a vague rattling sound on a few notes that sounded exactly like a speaker with a voice coil failing. It got progressively worse as I played it, also like a failing speaker, but this was 4 watts worth of amp feeding 100+ watts worth of speakers that I have blasted hard before without issue.
After about 20 minutes the sound got nasty, then the amp died. The voltages in the first stage were all wrong, so I put in a new tube and tried it.....it's alive. However, the second tube has become microphonic after only a few minutes of playing. None of this was noticed with the bare PC board laying on the bench.
The maximum grid circuit resistance spec shown in data sheets may not be conservative enough with some of today's new production tubes. Tubes that have a less than optimum vacuum will start to draw grid current as they age. this current will work to overcome the negative bias unless it is bled off (leaked) through the grid circuit resistance. Grid current is the most common cause of output tube runaway (red plate) type failures.
There are several popular "boutique" guitar amps (Soldano) that use large grid stoppers particularly in the preamp where values over 100K are seen.
Amp 1.5 got wired up and tested and I was not too happy with the results. I set the chassis on top of the speaker cabinet and started playing it. There was a vague rattling sound on a few notes that sounded exactly like a speaker with a voice coil failing. It got progressively worse as I played it, also like a failing speaker, but this was 4 watts worth of amp feeding 100+ watts worth of speakers that I have blasted hard before without issue.
After about 20 minutes the sound got nasty, then the amp died. The voltages in the first stage were all wrong, so I put in a new tube and tried it.....it's alive. However, the second tube has become microphonic after only a few minutes of playing. None of this was noticed with the bare PC board laying on the bench.
Sorry to hear about the microphony headaches. I never know how much to trust what I read on the Internet, but if accurate, it seems this is the single biggest problem that has kept pentodes out of almost all guitar preamps.
I once found a white paper on the development of the 6AG5. IIRC, the control grid to cathode distance was around 3.5 thousandths of an inch, or about 90 microns. Considering how tiny that distance is, I wonder how much movement (vibration) of those electrodes it takes to cause microphony? Presumably a few microns of movement would be enough.
I'm using small-signal pentodes in my preamp as well, but not in the input stage. I've also set up the pentode stages for fairly moderate voltage gains (I aimed for 50, and ended up with between 60 and 70). This was done by lowering the B+ supply voltage to the pentodes, and using rather small anode load resistors, in the range of 47 kilo ohms.
I'm hoping those measures will keep microphony problems at bay. So far, so good...(fingers crossed.)
-Gnobuddy
I once found a white paper on the development of the 6AG5. IIRC, the control grid to cathode distance was around 3.5 thousandths of an inch, or about 90 microns. Considering how tiny that distance is, I wonder how much movement (vibration) of those electrodes it takes to cause microphony? Presumably a few microns of movement would be enough.
I'm using small-signal pentodes in my preamp as well, but not in the input stage. I've also set up the pentode stages for fairly moderate voltage gains (I aimed for 50, and ended up with between 60 and 70). This was done by lowering the B+ supply voltage to the pentodes, and using rather small anode load resistors, in the range of 47 kilo ohms.
I'm hoping those measures will keep microphony problems at bay. So far, so good...(fingers crossed.)
-Gnobuddy
Have a 6AU6 on the input of a 12AB5 amp and being a combo the thing is not too happy when I try to use all the gain. Going to have to try a few different tubes in to see if I can find one that is not microphonic and baring that I am going to replace it with a 12AX7 for two channels of plexi fun.
What's the voltage gain of these tubes as used in typical circuits? The "actually seen" capacitance is the grid-to-plate capacitance multiplied by the gain (if it's a gain of 50, for every volt the grid goes up, the plate goes down 50 volts, making the grid-plate capacitance ten times that of when the plate is at a fixed voltage. The buzzword for this is Miller capacitance.
A common-cathode cathode-bypassed 12AX7 supposedly has a voltage gain of 100, which makes its input capacitance that much greater. I imagine this is the reason gate resistors aren't the higher values.
I was talking specifically about pentode (and beam tetrode) valves. Those have very little Miller effect, because the screen and especially the suppressor grids in between the anode and the control grid provide electrostatic screening, and isolate g1 from the anode.
From what I've seen, in many cases, pentode data sheets only quote a lumped "input capacitance", presumably because it is dominated by capacitance from g1 to cathode.
As you say, triodes are different animals entirely, and input capacitance seems to typically be an order of magnitude larger, mostly because of the Miller effect. Around 5-10pF for a pentode, around 50-150 pF for a triode, once you include Miller capacitance.
No wonder early radio designers were desperate for something better than a triode, and very eager to adopt pentodes when they arrived!
Incidentally, I've never seen a 12AX7 achieve much more than a gain of 50 - 60 in a guitar amp. These valves have such high rp and such high anode load resistances that it is difficult to avoid loading down the circuit, lowering voltage gain to well below the theoretical value of mu.
The unloved and unpopular triode in a 6JW8, as I found out, will also produce a (voltage) gain of 50, though mu is only 70. Because ra is much lower, there is less gain lost to loading.
Going back to your point about Miller effect: the datasheet says there is 1.8 pF of capacitance from g1 to anode in the triode section of a 6JW8. That turns into 90 pF input capacitance at a voltage gain of 50. Yikes! (There is an additional 3.2 pF of "input capacitance" specified, too.)
The 6JW8 also contains a pentode. The datasheet specifies 0.01 pF (!) between g1 and anode. It also specifies 5.5 pF "input capacitance", and states that this is from g1 to g2, g3, and the cathode. Even with a voltage gain of a 100, the Miller effect only adds 1 pF to the input capacitance in this case!
Looking at all this, it seems that pentodes really were an enormous improvement over triodes. I'm sure this was well known when they first arrived on the scene, but today, that fact seems to have been mostly lost. Instead, the rose-coloured spectacles of nostalgia have placed 12AX7s and directly heated triodes at the top of the heap (for guitar and Hi-Fi respectively).
-Gnobuddy
From what I've seen, in many cases, pentode data sheets only quote a lumped "input capacitance", presumably because it is dominated by capacitance from g1 to cathode.
As you say, triodes are different animals entirely, and input capacitance seems to typically be an order of magnitude larger, mostly because of the Miller effect. Around 5-10pF for a pentode, around 50-150 pF for a triode, once you include Miller capacitance.
No wonder early radio designers were desperate for something better than a triode, and very eager to adopt pentodes when they arrived!
Incidentally, I've never seen a 12AX7 achieve much more than a gain of 50 - 60 in a guitar amp. These valves have such high rp and such high anode load resistances that it is difficult to avoid loading down the circuit, lowering voltage gain to well below the theoretical value of mu.
The unloved and unpopular triode in a 6JW8, as I found out, will also produce a (voltage) gain of 50, though mu is only 70. Because ra is much lower, there is less gain lost to loading.
Going back to your point about Miller effect: the datasheet says there is 1.8 pF of capacitance from g1 to anode in the triode section of a 6JW8. That turns into 90 pF input capacitance at a voltage gain of 50. Yikes! (There is an additional 3.2 pF of "input capacitance" specified, too.)
The 6JW8 also contains a pentode. The datasheet specifies 0.01 pF (!) between g1 and anode. It also specifies 5.5 pF "input capacitance", and states that this is from g1 to g2, g3, and the cathode. Even with a voltage gain of a 100, the Miller effect only adds 1 pF to the input capacitance in this case!
Looking at all this, it seems that pentodes really were an enormous improvement over triodes. I'm sure this was well known when they first arrived on the scene, but today, that fact seems to have been mostly lost. Instead, the rose-coloured spectacles of nostalgia have placed 12AX7s and directly heated triodes at the top of the heap (for guitar and Hi-Fi respectively).
-Gnobuddy
I got curious about this a couple of years ago, especially because all over the Internet, people omit the output stage when counting "gain stages" in their guitar amps. Why would you do that, when it's a perfectly respectable common-cathode amplification stage?
So I did a few back-of-the-envelope calculations, nothing fancy, just using gm for the output valves, and typical anode (transformer) load impedance.
What I found is that the typical guitar amp output stage has a voltage gain of more or less unity (1), once you include the output transformer's step-down ratio! There may be something like a +/- 50% variation on that, it isn't exact, just a useful ballpark number. There is no scientific basis for it either, just like Moore's Law. It's just an observation.
This doesn't help with Miller capacitance calculations (where you need the gain to the anodes), but I find it does help at an early design stage for a guitar amp, when you're trying to figure out roughly how much drive signal you need at the output stage grids. Basically, you need as many volts as you expect to find across the speaker coil at full output power, give or take 50%!
-Gnobuddy
I seem to remember you were getting around 100 dB SPL from your 2-watt mini 5E3 and a 12" speaker. I'm gonna guess you're squeezing at least 6dB - 10dB more out of your 12AB5s. I can imagine it would be a challenge to keep an input pentode happy!
-Gnobuddy
I seem to remember you were getting around 100 dB SPL from your 2-watt mini 5E3 and a 12" speaker. I'm gonna guess you're squeezing at least 6dB - 10dB more out of your 12AB5s. I can imagine it would be a challenge to keep an input pentode happy!
-Gnobuddy
Had a 12" from a Fender Blues Jr (I think) in it. Reminded me of when I was a kid, haven't been in the cross hairs of a full band in years. I put an old lower efficiency C12N in and it now breaks up at more reasonable levels but still can get the 6AU6 singing on its own. Plan on rewiring a 6K6 amp into something different so what each will end up as is a tossup.
Grid stoppers are that small since they don't need to be larger. The whole point of including stoppers is two fold: to de-Q parasitic LC "tuners" that can make the tube oscillate like a Colpitts. Secondly, to swamp out any negative resistance reflected into the grid circuit from dirty admittance somewhere in the plate circuit via Miller Effect.
You would only make them larger if you were using them to do double duty: not just for stability, but also as a simple first order LPF. You might see that done for guitar amps where you don't need a "DC to daylight" bandwidth.
The spec sheet gives the Cgk value. That would only apply in circuit if you were using the 6V6 as a grounded grid RF amp. As a common cathode amp, the reverse transfer capacitance will reflect into the input by Miller Effect.
Cmiller= Crt(1 + Av)
Ci= Cgk + Cmiller + Cstray
500K would cause a greater loss of highs than you'd expect here.
As previously stated, larger than normal grid stoppers are sometimes used on the output tube grids to reduce blocking distortion, or at least increase the time constant in the blocking condition.
While only a mild issue in most HiFi amps, blocking distortion, also called "farting out" or "farting distortion" can be a big issue in some guitar amps. A guitar amp is often driven to extreme clipping. The drive to the output stage grid can be enough to attempt to drive the grid highly positive causing a diode clamping effect on the positive excursions. Prolong operation in this manner (common practice) will upset the charge on the coupling cap. Upon elimination of the overdrive condition the grid will be biased far more negative than normal, and recovery time can be measured in seconds. During the recovery time the amp will be biased near, or even into cutoff, making sound only on signal peaks. The output is said to be "blocked." Sometimes the time constant is such that the amp will pulse into and out of the blocked condition causing a "fart" sound.
Increasing the value of the stopper resistor will reduce the cap bleedoff issue (resistor much larger than dynamic grid resistance), and increase the time it takes to happen (longer time constant). Often the coupling cap is smaller than normal to reduce recovery time. These values need to be determined empirically since they are highly dependent on the amount of overdrive that the PI can deliver, and the dynamic resistance of the control grid in the positive region.
The other school of thought is a big fat coupling cap, and a PI with a rather high output resistance, say a 12AX7 with 100K or higher load resistors. Usually a happy medium works best.
While only a mild issue in most HiFi amps, blocking distortion, also called "farting out" or "farting distortion" can be a big issue in some guitar amps. A guitar amp is often driven to extreme clipping. The drive to the output stage grid can be enough to attempt to drive the grid highly positive causing a diode clamping effect on the positive excursions. Prolong operation in this manner (common practice) will upset the charge on the coupling cap. Upon elimination of the overdrive condition the grid will be biased far more negative than normal, and recovery time can be measured in seconds. During the recovery time the amp will be biased near, or even into cutoff, making sound only on signal peaks. The output is said to be "blocked." Sometimes the time constant is such that the amp will pulse into and out of the blocked condition causing a "fart" sound.
Increasing the value of the stopper resistor will reduce the cap bleedoff issue (resistor much larger than dynamic grid resistance), and increase the time it takes to happen (longer time constant). Often the coupling cap is smaller than normal to reduce recovery time. These values need to be determined empirically since they are highly dependent on the amount of overdrive that the PI can deliver, and the dynamic resistance of the control grid in the positive region.
The other school of thought is a big fat coupling cap, and a PI with a rather high output resistance, say a 12AX7 with 100K or higher load resistors. Usually a happy medium works best.
In Hi-Fi use, sure. As George has so eloquently explained, things are different when you expect to overdrive the output valves well into grid-current territory.
Put another way, those coupling caps charge via the grid bias resistor (let's call it R1), as well as the (grid stopper R2 + forward biased resistance of g1-cathode). The same caps discharge only through R1.
So, to minimise bias shifts and attendant problems during overdrive, we would ideally like the grid stopper (R2) to be much greater than the grid bias resistor (R1). And quite universally, we find exactly the opposite in every guitar amp!
This problem (and it's theoretical solution) became clear to me when I modelled it in LTSpice a while ago. The traditional measures taken to reduce blocking distortion in overdriven guitar amps (such as smaller coupling caps) merely minimise recovery time for unwanted bias shifts - they actually do nothing to minimise the amount of bias shift. Changing that ratio of R2/R1 is the only way to do that, short of adding additional circuitry (zener diodes, etc).
The issue of Miller capacitance has been discussed a little bit on this thread, just a few posts ago. We're not talking triodes here - for the output pentodes and beam tetrodes I've looked into, the Miller capacitance is typically an order of magnitude smaller than the capacitance between the control grid, cathode, and heater, and therefore I have neglected it (it often isn't even specified in data sheets).
George says he did hear a difference using 100k grid stoppers on KT88 valves, and I would like to find out what it was he heard. In this sort of extremely nonlinear use, maybe using large grid stoppers is reducing some audible traces of intermodulation distortion within the audio band?
Intermodulation distortion is another thing that is despised (rightly so) in Hi-Fi, but many electric guitarists are used to large amounts of it, welcome it, and would be bothered if it went away!
-Gnobuddy
I had a little time on my hands while I was cooking dinner (waiting for things to cook), so here are a couple of LTSpice screenshots to show what we're talking about.
Let's start with the first image. V1 is the signal from the driver/phase splitter stage; it maxes out at 70 volts peak-to-peak in my amp. C1 is the coupling capacitor feeding this signal to the grid of the output valve. R1 is the grid bias resistor for that output valve, and R2 is it's grid stopper.
D1 and the 10V DC source are there to model grid current flow. In my (cathode biased) amp, the cathodes sit at +10 volts quiescent, and the control grid therefore will begin to conduct current if it goes any more positive than +10 volts.
Fairly conventional values have been used for the components, including C1, R1, and R2.
These are all the components you need to re-create the problem (bias shift due to grid current flow through C1). However, it is very hard to see the extent of the problem in LTSpice, because the huge 70 Vpp AC signal covers up the smaller DC shift we're interested in.
So I added R3/C2, and R4/C3, a two-stage low pass filter that removes most of the 100 Hz audio signal, while letting the DC bias voltage at the grid of the output valve come through.
In a perfect world, this output signal (Vbias) would be zero volts initially, and would stay there when V1 begins to drive AC into the circuit.
What happens in actuality is shockingly different. Vbias plunges to (-16 volts) if the input signal remains present for half a second! (A half-second long guitar note is not typical, but not absurdly long, either.)
Remember, this additional (-16V) is on top of the existing (-10V) of bias. So the output valves, which were idling in class AB at -10V bias, are pushed deep into cutoff, with -26V of grid bias, when a long, loud, guitar note comes along.
-Gnobuddy
Let's start with the first image. V1 is the signal from the driver/phase splitter stage; it maxes out at 70 volts peak-to-peak in my amp. C1 is the coupling capacitor feeding this signal to the grid of the output valve. R1 is the grid bias resistor for that output valve, and R2 is it's grid stopper.
D1 and the 10V DC source are there to model grid current flow. In my (cathode biased) amp, the cathodes sit at +10 volts quiescent, and the control grid therefore will begin to conduct current if it goes any more positive than +10 volts.
Fairly conventional values have been used for the components, including C1, R1, and R2.
These are all the components you need to re-create the problem (bias shift due to grid current flow through C1). However, it is very hard to see the extent of the problem in LTSpice, because the huge 70 Vpp AC signal covers up the smaller DC shift we're interested in.
So I added R3/C2, and R4/C3, a two-stage low pass filter that removes most of the 100 Hz audio signal, while letting the DC bias voltage at the grid of the output valve come through.
In a perfect world, this output signal (Vbias) would be zero volts initially, and would stay there when V1 begins to drive AC into the circuit.
What happens in actuality is shockingly different. Vbias plunges to (-16 volts) if the input signal remains present for half a second! (A half-second long guitar note is not typical, but not absurdly long, either.)
Remember, this additional (-16V) is on top of the existing (-10V) of bias. So the output valves, which were idling in class AB at -10V bias, are pushed deep into cutoff, with -26V of grid bias, when a long, loud, guitar note comes along.
-Gnobuddy
