I know I'm late to this party, but I'll throw my own grenade in to see what falls out -- and up front, I'll readily admit to hardly being the last word here on this -- or on anything else for that matter. That said --
It seems to me that the one irrefutable fact in all of this is that Impedance is equal to the square of the turns ratio. I know, I know. Impedance is not simple resistance since inductance and reactance are also part of what makes up the total impedance seen by the tubes, which traditional load lines can't account for because those elements are unique to each transformer and speaker load employed. That's also why tube manuals don't specify effective plate-to-plate load "impedance" under their typical operating conditions offered, but effective plate-to-place load "resistance".
But still, as the impedance rule relates to an unknown transformer being measured and whether it should be measured with a loaded or unloaded winding -- how can a transformer represent two different turns ratios at same time -- one determined from a loaded test, and one from an unloaded test? Obviously, it can't. So irrefutable fact number two is that a transformer's turns ratio -- whatever it is -- is, what it is, what it is, all the time, whether the transformer is loaded or not.
My conclusion then is that the most accurate way to determine the true turns ratio, is to measure it with unloaded windings, as any load will introduce loses to distort that finding.
OK. As at least one data point, measured with unloaded windings, the results I measured with Dynaco transformers agrees very closely with that published by Dynaco for their transformers -- which we thankfully have manufacturer's data for. Now admittedly, the same results don't holdup for Acrosound transformers, which always measure higher than the information states by the manufacturer. So as Jack states, there are obviously a variety of factors involved that seem to lack standardization by manufacturers, at least back in the day, anyway -- if not still today as well.
But to finish my point, if the impedance rule is what it is, and the actual by the count turns ratio within a transformer is what it is, then what of the two different results produced by the two different test methods?
Again, my belief is that the simple reflected impedance presented by a transformer is most accurately represented by the turns ratio as determined using an unloaded turns ratio test, and that the apparent increase in reflected load impedance as determined by a loaded turns ratio test more realistically reflects the same simple reflected impedance, plus losses that bring into play the efficiency of the transformer being measured.
After all, the otherwise increased impedance determined from a loaded winding turns ratio test results in reduced power output -- all else being equal -- which is also exactly what happens when a real world output transformer is inserted between the plates of the output tubes, and the load. Operated as given in the typical operating conditions, the stated power output produced is never achieved in the load in practice, precisely because of output transformer losses.
Over the years then, I've come to rely on the unloaded test as the most accurate way to determine simple reflected impedance by a transformer. But from there, actual power delivered into the load is must be reduced by the very real losses that any transformer has (I find that most are about 81% power efficient ), with any design limitations relative to frequency producing additional losses on top of that. For me, this approach has produced very close agreement with the power output a load line suggests will be developed, and that which is actually delivered.
Based on load lines then, using a transformer whose impedance -- as determined by an unloaded turns ratio test -- matches that of the load line drawn, will allow the power the load line represents to be developed. But all else being equal, redrawing the load line to reflect the impedance that the transformer offers from a loaded turns ratio test will be more indicative of the actual power output the load will receive -- if you chose to want to go about it that way. Personally, I prefer to use a standard power efficiency of 81% and unloaded turns ratio tests, which have always returned very accurate results in my experience.
As I say, I'm hardly the last word on transformer considerations, so I'm more than willing to accept whatever holes that can be poked in all of this -- and I must assume some can. So for those can, please elaborate any if you will . Otherwise, I'm just trying to add my own experience to the mix to offer whatever benefit it may have.
Dave
On a better note, the A500 is still on the bench following the second round of measurements, so I'll measure that one for now.
