Input Transformer Saturating From Bias Current?

Hello all,

I was wondering if someone could shed some light on whether or not one needs to use a capacitor to isolate the secondary of a mic input transformer from the grid of the first stage in a tube preamp.

I've noticed on one of my preamps that I get a little more headroom if I isolate the mic input trafo from the first grid with a cap and use a 1M grid resistor to ground on the tube side of the cap. Without the cap and grid resistor, it appears that the small amount of bias current travelling through the input trafo secondary is enough to saturate the transformer a bit.

I see lots of direct coupled input transformers in tube pre designs. How does one get around the problem of having bias current flowing in the input transformer secondary?

Thanks,

Dean

> Without the cap and grid resistor, it appears that the small amount of bias current travelling through the input trafo secondary is enough to saturate the transformer a bit.

Are you sure of that?

An alternate explanation of the observed "little more headroom if I isolate the mic input trafo from the first grid with a cap and use a 1M grid resistor" is that when you drive the grid positive, the grid current clamps the cap and bias shifts more negative. Instead of outright clipping, just the first peak clips, the rest pass clean. While this can be a very useful effect, you might consider a higher grid bias or a lower-ratio transformer (or even a mike-pad) if you are really running into grid current.

Thanks PRR,

when you drive the grid positive, the grid current clamps the cap and bias shifts more negative. Instead of outright clipping, just the first peak clips, the rest pass clean.

Click to expand...

I'm not sure I understand. Is the coupling capacitor charging and discharging, thus temporarily increasing bias enough to show the increase in headroom?

you might consider a higher grid bias or a lower-ratio transformer (or even a mike-pad) if you are really running into grid current.

Click to expand...

The transformer ratio I'm using is 1:8. I ran the same experiment with it wired 1:2 and did not see nearly as much(or any) difference on the scope between direct coupling and using a capacitor and grd resistor, as far as headroom goes. That is, if I bring the signal right to the brink of clipping the tube, then bypass the coupling cap, the signal remains clean when using the 1:2 ratio. With the 1:8 ratio, the signal clips big time when I bypass the cap.

Does this go along with what you are telling me?

Dean

> the signal remains clean when using the 1:2 ratio. With the 1:8 ratio, the signal clips big time when I bypass the cap.

Clipping a tube with a 1:2 step-up? How big is that signal???

Anyway: in grid current, the tube input is about 2K. A 1:2 tranny presents less than 1K source (assuming you drive with a mike, which I suspect you are not doing), so clipping is slight. A 1:8 tranny presents around 12K, which is huge compared to the 2K of a forward biased grid, so the wave just flat-tops.

> Is the coupling capacitor charging and discharging, thus temporarily increasing bias

Yes.



The SPICE simulation starts with no signal, computes the DC operating conditions, then starts a full-power (7V peak) sine wave.

The tube is biased with 2.6V on the cathode, 10mA nominal current. (Not really typical, but shows the effects fine and simplifies some math.)

Blue curve is a perfect sine source.

Purple curve is the direct-coupled case: that source through a 3K resistance (mike and transformer impedance) to the grid. It is badly bent above +3.6V. It would be bent more with a higher source impedance. (Results are very approximate: my SPICE model's grid is crude.)

Green curve is a tube biased the same way but with 0.01uFd and 1Meg grid R-C network. Because I cleverly started my sinewave on the negative (no grid current) side, the first half cycle is clean. All the voltages track.

But as the sine swings up to +3.6V, something happens. It goes very bent, and actually changes path. Even after the input drops below +3.6V, the grid voltage is offset negatively. The negative half-cycle is clean, but look at the next positive half-cycle: it is much less bent than the no-RC case. The cap has charged up enough voltage to keep the positive peaks just-barely distorted with grid current. The exact amount of clipping is hard to figure, but with a reasonable source impedance and a big grid resistor it can be under 1% (after that first horrid +half-cycle).

The yellow curve is a highly smoothed (to hide signal waves) plate-current plot. The no-signal DC current is 10mA, but after the grid-cap charges it is about 6mA. (You will see this on a DC ammeter or in the cathode voltage.)

So...it appears that the issue won't exist, direct coupled or not, provided that the signal peaks don't exceed the bias voltage; ie, the grid never goes positive and draws current.

What I did not realize is how the tube's input impedance drops so much when the grid swings positive.

A tube's input impedance is super high(several megohms?) outside of grid current, right?

I've seen a lot of that bent purple curve, not really understanding the reason behind the distorted waveform...a combination of grid current and loading.

Thanks for the lesson, PRR.

Dean[/quote]