Resonance problem between switchmode power supply and electronic load

[skysurf76]

Its been my job over the last couple of months to come up with a test/adjustment procedure for MVP8 switch mode power supplies (datasheet attached).  When I started all we had for loading was a bunch of 250 watt wire wound resistors of various (usually inconvenient) values.  I needed an electronic load for my lab at home so I ordered a KL238 300 watt DC electronic load from ebay and took it to work. 

The model of supply we use has 6 modules with 7 voltages. (-24/-2.2)(12/-5.2)(24)(-12)(5)(5)

The DC load was working great for every module.  I could load them up their rated power and the supply would regulate the voltage to its specification.  When I got to the -5.2 volt output everything changed.  The voltage started at -5.2 and with only a 1 amp load the voltage would drop to below -6 volts, after about 4-5 amps the voltage would then jump into the low -4's.  I could also hear a high pitched whining coming from the module.  After verifying this behavior across several supplies I almost convinced myself that there was a serious design problem with the -5.2 volt module.  Then it hit me and I realized the voltage output as I increased the load was that of a resonant circuit.  I took one of the available 250watt wire wound resistors and put about a 2 amp load on the -5.2 volt rail and its voltage held steady, and with no whining.

I'm assuming the control circuitry in the load has a frequency of operation similar to the switching being done on the -5.2 volt output.  My question is if there is anyway to verify my theory and overcome it.  If this were a resonant LC circuit I would just add capacitance to change the resonant frequency, but seeing as how the control circuits in both devices are causing it I don't think a capacitor would change anything.

I really do find this phenomenon very interesting and would love to hear any knowledge or advice anyone might have on the subject.

[exe]

Two control loops (in psu and in the load) got in conflict   I'd try a resistor in series/parallel with the load, this should move operating point of the load and may help. But this will affect readings and settings.

PS I had the same with my DIY load which I use as a resistor (no chopping / step response / etc). I just slowed down its response with a 10nF cap in feedback loop, but that was an invasive change.

BTW if your load has 4-wire measurement you can add external compensation.