Parameters with 14 or 15 significant figures are very impressive! How was the grid current measured this precisely? However, as it is just for positive grid it may be of limited usefulness for the 12AX7 as that is almost always used with negative grid.
Not exactly sure what Robert did to get his model, but most likely some curve fitting algorithm. Grid current models for the 12AX7 is of some interest to guitar amp designers, since they are always driven very hard...
I suspect the 68K resistor is a grid stopper to filter out RF.
I usually use 10K in my designs.
This is weird, but by biasing the cathode follower so warm that grid current flows on positive half-cycles, Fender (or his technician) accidentally created a cathode follower that generates a huge amount of fairly smooth-sounding distortion. Positive input cycles are softly clipped due to the inability of the preceding stage to supply enough grid current.
After the Bassman failed spectacularly at its intended purpose - amplifying bass guitar - guitarists began plugging into them, probably because they'd already paid for the lemons. Eventually some blues guitarist turned one of them up loud enough to distort heavily, and liked the resulting sound.
And eventually, someone figured it out, and the accidentally-badly-designed cathode follower went on to be intentionally designed into later guitar amps.
So the reason for intentionally allowing grid current is to deliberately create huge amounts of asymmetrical clipping distortion (rich in even harmonics).
When I say "huge amounts", keep in mind, in some music genres, guitarists use distortion levels up to nearly 100% THD. That's not exaggeration - the RMS value of a square wave is 1, that of a sine is 1/sqrt(2), so if you distort a sine all the way into a square of the same peak amplitude, it delivers twice as much power into the load; that means that equal amounts of power are in the harmonics and the fundamental, so you actually have 100% harmonic distortion!
This is absolutely mind-boggling for those coming from the world of Hi-Fi, where 0.01% THD is considered to be too much by most. We're talking about ten thousand times more distortion in the guitar amp!
Even if death-metal isn't your thing, guitar distortion levels are high enough to shock anyone familiar with Hi-Fi design. In the world of electric guitar, a "clean tone" may contain 25% THD!
(I guesstimated this using an oscilloscope capture from a guitar amp of my own design, and attempting to recreate the captured waveform by adding sine waves graphically.)
-Gnobuddy
As Merlin Blencowe points out in his book, the 68k resistor also does a nice job of generating most of the equivalent input noise in the entire guitar amp. 😀
-Gnobuddy
Thank you, this will be very helpful.FYI, here's real data:
Tim
What was your setup to obtain this data? I am planing some testing too, which I will outline shortly. Anyone with feedback would be appreciated.
Obtaining some data
Frustrated with lack of grid current data on popular SE tubes for guitar amps, I am proposing to conduct the following testing.
A number of years ago I converted an old monster of a 100W Traynor head to a Marshall JMP Lead "2203" circuit. I intend to take it apart in order to allow some probing and use it as my test bed. The intention is not to obtain plate curves, just Grid curves in the overdrive regime.
Included here is a diagram of the input and DC follower stages that will be tested.
Test1: Figure 1, all JMP's=IN. This will obtain pure IG vs. VG.
Test2: Figure 1, JMP1=OUT. This will hopefully indicate how IG is affected by a dynamic reduction in the plate voltage (IG should be higher than in Test1).
Test3: Figure 1, JMP1 and JMP3=OUT. This will indicate how IG is affected by both a dynamic change in VP (plate voltage) and VK (cathode voltage).
Test4: Figure 2. This test of the DC follower will show IG for not only the gain stage (in a manner similar to Test3), but also the steady-state and dynamic IG of the follower.
Any feedback or suggestions is welcome.
Will this data be useful/valid? Am I missing anything? Is this a waste of time, "do it the right way"?
Frustrated with lack of grid current data on popular SE tubes for guitar amps, I am proposing to conduct the following testing.
A number of years ago I converted an old monster of a 100W Traynor head to a Marshall JMP Lead "2203" circuit. I intend to take it apart in order to allow some probing and use it as my test bed. The intention is not to obtain plate curves, just Grid curves in the overdrive regime.
Included here is a diagram of the input and DC follower stages that will be tested.
Test1: Figure 1, all JMP's=IN. This will obtain pure IG vs. VG.
Test2: Figure 1, JMP1=OUT. This will hopefully indicate how IG is affected by a dynamic reduction in the plate voltage (IG should be higher than in Test1).
Test3: Figure 1, JMP1 and JMP3=OUT. This will indicate how IG is affected by both a dynamic change in VP (plate voltage) and VK (cathode voltage).
Test4: Figure 2. This test of the DC follower will show IG for not only the gain stage (in a manner similar to Test3), but also the steady-state and dynamic IG of the follower.
Any feedback or suggestions is welcome.
Will this data be useful/valid? Am I missing anything? Is this a waste of time, "do it the right way"?
Attachments
Last edited:
I was just thinking about this today.
I wonder however if Marshall "screwed up" or did this intentionally (see my Figure 1 above). Assuming Fender was and is the "standard", 68k is not the correct value. If one studies the input jacks closely, the two 68k's end up in parallel when input 1 is used, therefore the input grid stopper is really a 34k resistor. When input 2 is used, the Rin drops to 136k, and the two resistor form a 6dB pad. This input was meant for "line level" equipment, or a heavily degraded guitar tone.
But yes, this resistor determines 99% of the input noise of the amp (and rolls off the high end).
Okay, that makes the thermal noise voltage a whole 3 dB better.
I suppose a little input noise didn't matter much for the sort of low-gain, clean tones Fender intended.
The thing is, the source impedance from the guitar is wildly unpredictable. Leo's guitars use 250k pots these days, so worst case, up to around 63 kilo ohms of source impedance. And that's not counting the reactance of the pickup itself, which can be quite substantial, and will raise the source impedance even more.
I guess maybe the volume pot in the guitar can actually contribute even more thermal noise than that 34k effective grid stopper!
-Gnobuddy
The pickup reactance won't contribute anything to the noise figure, but its DC resistance (anywhere from 2.5k to 14k) certainly will.
Agreed. My point was that it doesn't help the cause in terms of having a predictable low pass filter, since it changes the source impedance (as well as making it reactive).
Not to go too far from the thread topic, I've wondered if grid current in the input valve is a factor in guitar amp distortion. The typical input half-12AX7 seems to usually be biased at between -1 and -1.5 volts, and it seems many guitar pickups are capable of spitting out 1 volt peaks, maybe more.
Those two numbers suggest that the input stage might already be contributing some compression and distortion to the guitar signal due to grid current flow on signal peaks.
-Gnobuddy
Agreed. A guitar signal, especially the hot ones these days, can and will cause the input stage to venture into grid limiting. It doesn't sound like hard or harsh limiting though, more like compression or loss of dynamic peaks.
Too bad there isn't more interest in this aspect of tube modeling.
Too bad there isn't more interest in this aspect of tube modeling.
'Twas just as it says: apply voltages and measure currents. Didn't notice any drift with time, no need for pulsed measurements.
Tim
- Status
- Not open for further replies.