Grid Input Transformers

[rdf]

Can a small signal transformer (CineMag, Jensen) not rated for DC current be used to feed a fixed bias grid input? It means applying a voltage to the secondary but as long as the grid isn't driven positive and the bias stays within the winding's voltage ratings I can't see why it shouldn't work. What's the risk? Thx in advance.

[SY]

Should be no problem if the signal level isn't higher than the transformer rating- the grid draws no DC.

[rdf]

Is not done often because of perceived cost? It seems such a nice solution - balanced in, high CMRR, high RF rejection, sub-sonic filtering, no large caps of dubious quality or expense on the cathode (and no rising Zout), even the potential of a little free gain in some cases eliminating a stage - that I just assumed I was overlooking something obvious. Thanks for clearing that up.

[SY]

Why isn't it done more?

Well, there's two issues that I can see, other than cost. First is signal level- fixed bias is used for output tubes where many volts must swing. That means the transformer must be built to handle that amount of signal, which translates to compromises elsewhere. Second is Miller Effect in the output tube, which can also make life difficult for the transformer.

The grid is a high impedance point so stray capacitances must also be taken into account, an account which grows with transformer size.

[rdf]

Sorry, should have been more specific. The Cinemag group buy got me thinking about using one at the first input stage for the generally reduced distortion (and noise?) of fixed bias and the elimination of a time constant in the cathode. Count me among the ranks of the cap-phobic when values get large enough to work with small value cathode resistors too. No intention to use global feedback so not too worried about losing the injection point, and of course power out isn't an issue in this position.

The more I play and learn the more convinced I become of the critical role played by the first stage. The order of harmonics generated here are amplified geometrically by non-linearities in the following stages - 3rd order distortion of the second stage generates 6th order from the 2nd order distortion of the first stage...if you know what I mean (which I'm sure you do.) Wrapping feedback around it squares this once again. Any effort spent on the first tube pays off handsomely. I like the balance of trade-offs an input transformer represents.

Thx once again.

[Steve Eddy]

Originally posted by rdf
The more I play and learn the more convinced I become of the critical role played by the first stage. The order of harmonics generated here are amplified geometrically by non-linearities in the following stages - 3rd order distortion of the second stage generates 6th order from the 2nd order distortion of the first stage...if you know what I mean (which I'm sure you do.) Wrapping feedback around it squares this once again. Any effort spent on the first tube pays off handsomely. I like the balance of trade-offs an input transformer represents.

I know what you mean. I eventually went a step further and made the transformer and the input stage one and the same using step-up transformers to get all the advantages of transformers plus the voltage gain that an active input stage usually provides.

se

[ThorstenL]

Konnichiwa,

Originally posted by rdf
no large caps of dubious quality or expense on the cathode (and no rising Zout)

But the introduction of a coupling capacitor in the input loop, on top of the transformer.

I am always amused if I find people "believeing" that the addition of transformers removes capacitors, it rarely ever does, it merely shifts the position, not that I'm against transformers though, mind you, on the contrary.

Sayonara

[rdf]

Originally posted by Kuei Yang Wang
But the introduction of a coupling capacitor in the input loop, on top of the transformer.

Do you mean the final smoothing cap of the grid supply is effectively in a coupling position? I'm aware of that and am thinking of a couple ways to ameliorate its effect, or effectively eliminate it. I believe it's doable but need to sim and test first.