DC offset... why?

[composition4]

Hi, I'm building a guitar amp... opamp input (rail to rail 18V) which feeds to a pot then to a 12AX7 stage.

SPICE simulating some ideas... for some reason I'm getting a huge DC offset jut before the grid stopper for the 12AX7, but I can't figure out why! It actualy eems to disappear when I delete the 12AX7 and just have the 68K resistor connect to ground reference..

The picture shows my sim, the graph on the left is the input voltage, the graph on the right is the voltage after the pot and before the grid stopper



Thank you in advance!
Jonathan

[amptramp]

There are a couple of things at work here:

1. Contact potential

With the circuit running with no input, there may be a slight negative voltage at the grids.  In fact, if you had 10 megohms to ground, you would get enough bias to run without a cathode resistor on any signal small enough to not overload a 12AX7.  But your grid resistance values appear to be well below that.  Contact potential is caused by electrons coming off the cathode and hitting the grid.

2. Grid rectification

You show a six volt p-p signal going into a stage with a gain of 5.89.  This output is more than your power rail, so I don't see how you can get a sine wave at the op amp output, but if you have a signal going into the grid which is sufficient to bias the grid positive with respect to the cathode, the grid will act like a plate and rectify the signal.  Electrons cannot go into the grid but you can pull a steady electron flow out if the grid goes positive since the grid intercepts electrons going to the plate.  The electrons come out of the grid and keep the voltage from going up by depositing a negative charge on the right side of C5.  When the signal goes negative, there is nowhere other than through RV1 for this charge to bleed off, but since this is a resistor, electrons going in from the top bias the grid negative.  The voltage decays with a time constant of C5 x (R11 + RV1).  The effect diminishes when RV1 is turned down.

[composition4]

Thanks for the reply.

First up, sorry I meant to say +/-18v supply, so the 5.89*6V should work out to within the rails.

So I understood about half of your second point, I'm new to vacuum tubes and just finding my way around them.  You're right, when the pot is turned down the offset dimishes.  I also noticed if I changed the pot to a 5K, the offset also diminishes, same principle at work I guess.

Do you have any suggestions as to what I should change to get rid of the DC offset so that the opamp/pot stage will lead into the triode stage without problems/offset?

Thanks!
Jonathan

[merlinb]

Strange... valve models don't normally include any grid current effects, so I would doubt that's what you're seeing, although maybe you've got a more advanced model from somewhere?
Sure you haven't accidentally set up RV1(3) to measure the grid-cathode voltage, rather than the grid-ground voltage?

[PRR]

Fifteen Volts into a 12AX7 grid??

The 12AX7 cathode is at 2V. If the grid is negative of cathode, "no" current flows. If grid is higher than cathode, BIG curent flows.

Move your probe over to the 12AX7 grid pin. You will see it is not going more than a few tenths Volt higher than the cathode voltage, and is probably severely clipped, "flat-topped".

You have a cap feeding this network. A cap, a diode, and a big AC voltage *will* make DC voltage. In this case, the grid goes negative when signal is very large.

12AX7 has amp-factor of 100 and is typically eating 300V. So inputs over 3V peak are pretty-sure to be grossly distorted. (If distortion is asymmetric, DC will happen also.) In real life, 12AX7 may have signal gain of 50, peak plate swing of 60V, so inputs over 1.2V peak are mangled.

And at the grid as well as the plate.

> things at work here:
> 1. Contact potential

Agree with Merlin that SPICE tube models rarely show grid leakage realistically, or with low values of grid resistor. With 1Meg, my experience is you will get a few-tenths volt; with ~~100K much less.

> 2. Grid rectification

+1

> change to get rid of the DC offset

Forget that offset. What is coming *out* of the 12AX7?

With 17V pk in, and gain of 50, we expect to see 850 Volts peak output.

This is like expecting 850 pounds of potatoes to fit in a 60 pound bag. Past "mashed".

Put your probe on the OUTput. Sneak the input up from zero until the output just clips. Turn-down a bunch, that's as far as you would go for High Fidelity. Turn-up from clipping 2:1 or 4:1, that's what guitarists call "overdrive". What happens past 10:1 above the edge of clipping is usually meaningless.