A in comparison ideal 6Vp-p/5kHz (lowering the test frequency from 10kHz to 5kHz was necessary in order to let the oscillations cling out) square wave signal (tr=500ns) was injected into all four primary windings connected in series and thru a 2.5kOhm resistor (simulating two KT66 type push-pull tubes or UL-connected KT88's as driving source) and measured at primary windings with 8Ohm connected at the secondary (S1+S2). All unused windings were series-grounded. This signal thereby represents the actual transformer driving voltage and as such indicates major HF resonance problems.
Applying the same signal at the HV-windings (S7-8) while moving the load to S1+S2+S3, results in major signal quality improvements (see HV windings stepresponse) even at 10kHz and with no visible stability problems what so ever! In regard of earlier test results on my OPT TI-57613-C, it is easy to deduce these problems to the (unfortunate) used winding technique. This will therefore be revised in later prototypes.
Testing the primary HF response with a stationary sinusoidal signal instead, results however in a smooth roll-off from 0dB@1kHz to -2.5dB@20kHz, which also is the maximum frequency my homebuilt signal generator can deliver.