[AMRadio] push pull output XFMRs and neg feedback

[Jim Candela]

>
> Message: 3
> Date: Mon, 18 May 2009 10:42:01 -0400
> From: John Coleman <jc at pctechref.com>
> Subject: Re: [AMRadio] push pull output XFMRs and neg feedback
>
>
> Thanks Jim
>        I'll give that a try before switching the XFMRs out.  I guess what 
> your saying is that the slight tolerance differences of components and 
> layouts me be enough to cause this and that the XFMR wires are color coded 
> different doesn't mean they or all wrong.  In his schematic the resistor 
> is 3300 ohms and the capacitor is .001 uf.  The plate resistance of the 
> 6SN7 is supposed to be about 7000 ohms.  With this data, how does one 
> calculate the roll off and how does the un-bypassed cathode resistor and 
> feedback to that cathode effect the response.  I have always done this 
> sort of thing experimentally but I figure that someone with a degree may 
> know the actual methods to calculate all of this.  BTW the cathode 
> resistor is 1000 ohms and the plate resistor is 47K.
>
> Thanks for the tip.
> John, WA5BXO
>


Reply by Jim, WD5JKO:

John, I'm not sure how exactly to calculate that R-C.  One thing for sure is 
that the more your transformers deviates from ideal, the less GFB (global 
feedback) you can tolerate before the output is phase shifted enough that 
the inverse feedback becomes positive feedback, and oscillation occurs at 
anywhere from 10 Khz to 100 Khz.

So lets say your circuit has 24 db GFB gain reduction, and it oscillates at 
15 Khz from transformer phase shift. If you do something to lower the GFB at 
15 Khz to say, 14 db (yet still 24 db at 1 Khz), the oscillation will either 
go away, or move up in frequency. With the R1-C1 is in that circuit at the 
link I gave you in my post, the GFB will continue to decline as the 
frequency goes up. As Patrick Turner said, you must fine tune the GFB for 
every transformer-circuit combination.

I personally prefer less GFB while stressing linearity first, hence needing 
less GFB as a result for a given amount of distortion. One example with 
multi-grid output tubes is to use the 4, 8, 16 ohm secondary's as a source 
of cathode feedback. Do this by grounding the 4 ohm tap along with floating 
the Common, and crisscross the cathodes to the output common, and the other 
to the output 16 ohms. Doing this will lower the gain 3-6 db typically, and 
it will make a low end transformer look like a more expensive transformer. 
This approach seems to in effect reduce the primary to secondary leakage 
inductance.

With the R1-C1 in place, you will surely see that the transformer phase 
makes a big difference. Then start tweaking toward optimum. With a square 
wave at 10 Khz (and resistive 8 ohm load), compensate for best response such 
that the damping is not over, or under damped.

Regards,
Jim
WD5JKO