The core issue
So here I am this one fine day with half a dozen or so switchmode power supply transformers wound on E80-sized ferrite cores and some of the transformers are okay while some of them are not. Measuring the winding inductances, I find that the good ones had inductance values corresponding to core AL coefficients in the neighborhood of six thousand mHy at 1000 turns, but the bad ones had much lower numbers, very much lower!. One bad transformer's core set AL coefficient was only five hundred.
Clearly, the bad transformers had some kind of ferrite core trouble and when I investigated further, that was exactly what I found.
Referring to the above sketch, the good transformers were made with no gaps as shown on the left. However, the bad transformers either had their coil bobbins improperly cut. They were overly long and/or the bobbins were not cut at an accurate right angle to their length. In both cases though, the two ferrite core halves could not come completely together. They were not in proper mechanical/magnetic contact.
This was why we had the inappropriately lowered AL coefficients.
Taking some of the bad transformers apart for visual inspection , we found situations like this one in the following photograph.
The frustrating part was that the folks who were assembling these transformers were habitually careless about doing it. They just didn't seem to appreciate the requirement to keep the core faces completely in contact with each other.
"Hey, the cores are touching each other, right? So what's the problem?"