Output transformer response test with backwards comparison

Here's a great example of the kind of wacky stuff transformers will do if you aren't paying close attention.  Yeah, yeah, they are just a middle man, converting one to another, blah blah blah.  Look at this:



We see two examples of the same 15K:600 output transformer, measured three ways.  All plots were checked for positive polarity.
The four plots showing a slight resonant boost would be a pair with 10K source, and a pair with 20K source, 10K source being flatter. 
The two plots with the reversed notches show what happens when you run them backwards, driving the 600 ohm secondary as primary, and running the 15K primary into a high-Z load. 

The resonance boost/notch has been addressed in other posts around here, apparently related to differences in each side of a hum bucking winding.  I see it frequently in UTC and ADC transformers, and interestingly it usually goes away if you run the transformer fully balanced on both sides. 

emrr said:

......................and interestingly it usually goes away if you run the transformer fully balanced on both sides.

Click to expand...

Very cool, thanks for the chart!  Please elaborate on your quote.  Were your graphs made in an unbalanced mode?  Would center tap "grounding" make a difference? 
Best,
Bruno2000

A UTC A-10 run typical LA-2A single ended secondary might show a slight notch in upper response, usually more of a single wavy looking dip in the plot.  Run that same A-10 with the secondary CT grounded, to a PP input stage, and the notch goes away. 

My graphs are of a SE output type with no CT brought out, and the notch seen happens in this case when the unit is used 'improperly'. 

Well, the thing is... most of the time, the source impedance driving the transformer windings are much lower than the rated winding impedance. Like a 10k input winding being driven by a modern source of 100ohms or lower.

So, if you are testing these for input signal into a circuit from the modern studio "outside world", you should try to feed it with a low impedance like 60ohms, that most gear uses as standard on the outputs nowadays. I mean both with the 15k winding or the 600. When going into the 15k winding, terminate the lower side with a 600R resistor, and when going into the 600 ohms winding, terminate the higher side with a 15k resistor. Also, loading the secondary with a grid, a base or a gate can change everything, like a zobel would do, but most of the time it only remove peaks. The dips in those tests are really strange, but may also go away with much lower source Z. Give it a try and  let us see how it goes.

If you are testing this for tube output stages, the output impedance of the tube is also generally lower than the rated transformer winding. For a 15k winding, it goes down as 2k for some circuits. But in this regard, the best thing to do is to actually test the transformer in circuit, and measure performance. That's is the only reliable way.

Sure, I've said exactly what you are saying as a warning about using modern iron in vintage tube circuits with little or no feedback, since almost all modern iron is specified for bridging drive.

This is a vintage design from the days of matching, and is specified as such.  I included the 20k source to show what happens when your drive source is weaker than matching.  The real story here is the resonant boost turning into a notch with reversed connection. 

A lot of vintage iron will develop resonant boosts at both extremes with you drive it with a bridging impedance.  The Audio Cyclopedia makes mention of this phenomenon, and I see it occur in iron designed for matching. 

Yes, the notch is really strange. But I though that it would be nice to see how it specifies with bridging impedance drive, as you are on it.