Effect of value of input R (Grid Leak)

[jjasilli]

What is the effect on input signal voltage from guitar into the typical amp input with a grid leak resistor of 200K or 1M?


Per the annexed LTspice simulations, there is no voltage drop from 1Hz - 10KHz!  (Though phase shifts a bit differently).  This did no change change with the tube's Miller capacitance is added to the circuit.  I think Miller capacitance is in parallel with the grid leak R; and Merlin says that for a 12ax7 with 1VAC in the Miller capacitance is 192.2pF.


Hopefully someone will verify these sims.

[pdf64]

To my shame I'm completely clueless regarding spice or any simulation software, but I think you may be trying to represent the input to the amp (guitar and cable) as a pure voltage source, ie zero source impedance? That's very much not the case, see Lemme's fantastic page on the topic http://www.buildyourguitar.com/resources/lemme/

[PRR]

The grid must have a DC path to a bias (often ground). It can be a high value. We normally pick it to have "little" effect on signal, if possible.

As PDF64 says, 1 Meg across a !ZERO! impedance voltage generator has NO effect. Most SPICE voltage sources will output "infinite" power if asked. A guitar has much resistance and reactance in it. All pedals have some impedance at the output. G-amps often have 34k from jack to grid (compare putting your 1Meg before or after this).

[jjasilli]

Thanks, guys!  I now see that the guitar's output reactance cannot be avoided, despite my laziness.  I'll draw the Lemme schematic, his Fig. 7. Fortunately his table gives some of the values for the virtual components in that equivalence schematic.

[shooter]

curious;
do you have to "create" the math for a "created" circuit 
in case that makes no sense 
I use Excel for my math, so I create the formula, "ask" for relevant values and excel fills in the formula and gives answer
thanks
dave

[PRR]

> peak resonance voltage at about 650Hz.

What is the 0.022uFd cap? Smells like the guitar Tone cap, with knob turned full dark.

For full-bright I would expect the peak to be 2kHz-6kHz. But too lazy to sim that tonight.

[PRR]

> do you have to "create" the math for a "created" circuit .... I use Excel for my math, so I create the formula

SPICE is slightly like a spreadsheet; it does the math you tell it to do (not always the math you think you are telling it to do).

Raw SPICE is not a spreadsheet, it eats a stack of punchcards (now a text file) describing the circuits and parts and outputs several types of analysis. A Spice suite adds GUI schematic entry and analysis plotting.

But unlike stock Excel, SPICE "knows" how to math reactances and BJTs JFETs MOSFETs (and "subcircuits" of almost anything, including chips and tubes).

BTW: basic Excel only finds deterministic solutions to linear problems. (Advanced Excel hackers please stay quiet a while.) But in electronics the most "interesting" problems do NOT have a direct linear path to a solution. What SPICE does is repetitive guessing. For two parts in series it assumes the mid-point voltage is "half". Then it computes the currents implied. They generally won't agree. So it shifts the initial assumption and tries again. It gets better guesses, iterates fast, and often comes to a close-enough (8-place) answer "quickly". It can do this for hundreds of parts and thousands of time-slices; since Pentium days, usually in seconds. (But on an 8088 I remember setting a buzzer to call me back when a 7-part run finally finished.)

[pdf64]

Cable capacitance significantly affects the response, it doesn’t seem reasonable to ignore it just because it’s a variable; surely better to choose a mid way typical value, eg 1nF?

[shooter]

it eats a stack of punchcards (now a text file) describing the circuits and parts

 
I built a punch card reader for my LAST merit badge for Eagle Scout 

thanks for the explanation, I'm looking at a program, ROC, that does exactly that, but it uses so many !%{*%}[] symbols that my brain isn't playing 

[shooter]

you're a better man than me Genghis 

so if you lower the 1M (grid leak), freq peak moves right?

[Williamblake]

So i build 1M gridleak into a Fender PA 100 replacing 100k and the owner wondered if i did it right but was satisfied after double checking and marked the modified input (for the guitar player). It had to be turned way louder before.

[pdf64]

Please consider and respond to the query raised by PRR in post #7.
ie the tone control looks to be set to minimum.

[PRR]

Here's a sweep with tone-knob omitted (or very high tone-pot resistance) and cable C swept from 1 foot to 100 feet.

[jjasilli]

Learned how to put multiple traces in a plot pane.  The secret is to give the component (here the grid leak R) a variable value.  Then insert a .step command into the schematic to plot a range of specified values.


Here's the graphs in V & in dB for a grid leak R ranging from 100K to 1M, in steps of 200K.  Now I can see why 220K is minimum recommended value for a 12AX7.

[shooter]

insert a .step command into the schematic to plot a range

you're writing code 
for x = 0 to 100 step 3 

[jjasilli]

Yes, more like simple commands.  In the early days of Microsoft DOS I wrote *.com files to automate commands to open programs, etc.  The dot commands in LTspice are not too bad.  If I can do it, anyone can!

[shooter]

*.com

yup, did the same.  IIRC DEC used .bat, never bonded with Linux.

I just wish there would have been a "standard" for symbols, computers might have become efficient tools instead of cumbersome sloths 

[PRR]

...I can see why 220K is minimum recommended value for a 12AX7.

Actually, "for guitar". An unloaded pickup has a top resonance, and we like it. Loading to 200k damps the resonance. 50k will shave the upper midrange.

Nothing to do with "for a 12AX7" except that's our most popular pickup input stage. (12AU7 or 6AU6 would slightly change the grid capacitance and resonance, but not-much compared to pickup and *cable* capacitance).

Grid resistors 50k and less are found in Hi-Fi where the source is not a highly wound pickup but some previous amplifier (preamp, tuner, CD player) or the lower-Z phono pickup which is intended to have its top-resonance tamed with 47k.

OTOH the 12AX7 has no "maximum" grid resistance (unlike fatter tubes). The bias will change a bit with 3Meg or 22Meg, but it isn't going to run-away and melt. Ampeg used 3Meg to be very-sure the input did not load the guitar the least bit. Early cheap amps used 10Meg to avoid using cathode bias.

But the =47k for phono and >>200k for guitar is about taming or not-spoiling the brightness of the top-end of the wound pickup.

[jjasilli]

Below is what I'm getting for an unloaded Strat PU.  (I'm using single coil for simplicity's sake.)

You have confirmed my guess that hi-fi would use a 50K or less grid leak R.  My reasoning is that it has a much flatter curve, at least out to about 3kHz (loaded), and is thus more true to the incoming signal. NFB could adjust the post-3kHz portion, which is not in the most critical range of human hearing.

[shooter]

just voyeur ing, is the knee up the "roll off"?

[jjasilli]

Sorry, that plot was misleading.  It shows only an upward spike, not a roll off.  The expected roll off which is quite steep, is off that screen, beginning at about 10.3 kHz.  Below is a new plot extended out past 10kHz to 100kHz, in dB.

The spike indicates that the combined reactance of the PU's DC resistance, inductance, and parasitic capacitance literally causes a ringing or resonance at 10.3kHz.  This causes the spike, and the subsequent roll off.

The ability to model circuits and see these things is the benefit of sim programs.  It would be difficult and tedious to get this info any other way.  It would be helpful to someone who likes to build one of a kind amps. 

[shooter]

Thx JJ, I did get confused with a roll-off going UP 
We used a lot of calibrated cables for work, the system was basically a really expensive spectrum analyzer.  with it we could figure out the cables Q, BW, and attenuation.

I like to build one of a kind but that's past my attention span 

[jjasilli]

Don't know what BW is.  Q is the shape of the knee.  A simple RC filter would have a smooth, nicely rounded knee.  A more complex reactive circuit, such as adding inductance, skews the shape of the knee; maybe causing a hump or a spike or a dip, before the roll off.


Attention span I think is a buckdancer's choice.  Either spend it physically tinkering with physical components; or in design & planning.

[shooter]

bw = bandwidth
Q in my world was the resonance, BW was the slopes on either side
hi Q typically = narrow BW
lo Q wide BW
since I'm ADD's poster child, enough math to get started then tinker