Control loop interaction problems of cascaded SMPS's?

[treez]

Hello,
We wish to power a 160W Buck LED driver (40V, 4A LED string) from the output of a 160W offline forward converter SMPS. (48V output).

Is it true that the feedback loop bandwidths of each of these converters cannot be near each other in frequency as this would result in control instability problems?

 

[D.A.(Tony)Stewart]

When the forward Converter output impedance is as high at the Buck regulator input impedance at similar frequencies or harmonics , there becomes a load regulation instability between the two regulators. LC decoupling may be needed to isolation them from similar frequencies and harmonic mixes


Without impedance plots and conducted noise or Buck load curves vs frequency (Spectrum on current) or something to measure supply/load impedance ratio .... I can only hand wave possibilities., But yes many things are possible with coupling, but not that critical with proper decoupling.

I once had a PWM 5V Buck driving a 12V PFM Boost regulator. It had true Chaos noise. where my Boost inductor sealed in epoxy SMT was hissing like running water in a loud lab. (Chaos theory > true random noise) with full scale ripple. Lowering the source impedance of the BUck output or raising the input impedance of the Boost regulator were options I had including LC filtering in between.

For load regulation or stability problems always analyze the source/load impedance ratio. For a good design < 1% for a bad design > 10%.
This applies to speaker amplifiers , linear regulators and filter cap ESR too. where you want the source ESR to be <1% of the equivalent impulse load.

In your case 3W LED's are around 0.3 to 0.5 Ohms each ESR. , 1W LEDs~1 Ohm, 100mW LEDs 10 Ohms, old 30 mW LEDs ~30-100 Ohms etc. ballpark....

VERY IMPORTANT RULE OF THUMB to REMEMBER ABOUT LEDs Which varies with wavelength and Mfg process to some degree ( eg UV vs IR are different)
Notice the max power * ESR product...
3W * 0.3R ~ 1
1W*1R =1
0.1W*10R=1

The units of this rule of thumb are [W/Ohm of ESR]


More details

P= I(I*ESR+Vth) / ESR where Vth is around 2.8V for power LEds and Vf is 3.0 and up depending on ESR and I

With 40/48V your duty cycle will be around 90% so the harmonics will range up to 1/10% or up to the 10th harmonic roughly, so the Buck regulator ought to be at a higher frequency.

It depends how much storage time you want in your inductors in case of a power dropout to determine the value of L in both cases, but you will want to use PFC in the forward converter.

Running 3V~3.5V LEDs from 40V means you probably have a string of 13 in series or 39W per string with a 160W barely capable of driving 4 parallel strings. (LOwer VF LED voltage means both higher power capacity and better quality from source. ( Best are 2.9V @ 1000mA )

So this means for 13S4P LED load is 13*0.3/4 or around 10 Ohms. so for source impedance it would be nice to have Buck output caps around 1% of this or <=100 mOHm. ( Polymer Caps only)