|[AMRadio] push pull output XFMRs and neg feedback|
jc at pctechref.com
Thu May 14 18:21:33 EDT 2009
This is maybe a little off subject here but I ran across an interesting thing. It may be of interest for someone building higher quality speech amplifiers of modulators for AM.
While constructing two identical amplifier circuits with inverse feed back for a stereo operation, I discovered that the introduction of the feedback caused oscillations in one channel, so I checked my circuits to see if I had wired some part of the push pull drivers or something in the output different from the other circuit. I had not. But reversing the output XFMR plate wires did correct the trouble for the most part. Now here is what is interesting. These are standard old fashion Thordarson push pull OTs for 6L6s. It has three wires on the primary and three on the secondary. Primary is green=P1 Red=B+ Brown=P2. Secondary is Black=ground Brown=8ohm Green=16ohm. I checked the XFMR with the ohmmeter before ever wiring it in but I did not check its through put phase. But the XFMR primary has to be wired different (that is the Green and Brown wires of the primary must be reversed) in order for the feedback to be inverted as it should be.
I said this corrected the trouble for the most part but not altogether. Now here is another phenomenon. I can increase the negative feedback, on the one amp that has always worked OK, to reduce the gain a lot with no oscillations. But on the other amp where I had to reverse the wires, I can only reduce the gain a small amount before the circuit goes into a high frequency (maybe 18000HZ) oscillation that is barely audible. If I put the wires back where I thought they should go, I get about a 500HZ oscillation immediately upon introduction of any feedback.
The circuit is push pull KT88 grounded cathodes, 425 V plate supply, regulated 300volt screens, RC coupled grids with separate bias pots for each grid and set to about 40ma cathode current per tube. Grid leak resistor is 100K from grid to bias circuits. Diver is a differential 6SN7 with 47K plate resistors and with a cathode balance pot.
Phase splitter is also 6SN7 - first plate is direct to grid of second triode. Cathode resistor and plate resistor for splitter is 22K. Appropriate B+ decoupling between stages.
The feedback point is cathode resistor of first stage. Feedback take off is from 8ohm connection of output XFMR. Appropriate decoupling and divider between take off point and feedback point.
This is a very common type of circuit used in many higher quality amplifiers. This one just doesn't have the ultra linear OTs with the screen taps and cathode windings.
I may need to move swap the XFMRs from one channel to the other just to test or prove if it is the output XFMR that is causing all this. But before I do I will open the feedback loop and parallel the two channels inputs. Then do a low frequency test to make sure of my wiring phase all the way through both amp. I guess I will sync the on the input then go dual trace to look at the paths in each amplifier. Then start sweeping up input frequency to see where I get phase shifting that might cause this.
Has anyone else found NIB / NOS XFMRS that don't match but are supposed to. May Thordarson just never intended these to be used with feedback and I just happened to find one that works OK but not the other.
John Coleman, WA5BXO
More information about the AMRadio mailing list
This page last updated 17 Dec 2017.