Why do so few datasheets offer such simple and clear feedback loop equations.

[treez]

Hello,

The MAX16833 datasheet, on pages 18 & 19 gives a really easy, clear and simple explanation of how to compensate the feedback loop of LED drivers done with the MAX16833 chip.

The LT3756 LED driver IC, and other similar ones by linear.com have the same type of internal components, so why have linear.com never layed out such a clear and simple version of these equations?


MAX16883 DATASHEET
http://datasheets.maximintegrated.com/en/ds/MAX16833-MAX16833D.pdf

LT3756 DATASHEET
http://cds.linear.com/docs/en/datasheet/375612fb.pdf

 

[schmitt trigger]

From the LT3756 datasheet:

The compensation resistor and capacitor at VC are selected to optimize control loop response and stability. For typical LED applications, a 2.2nF compensation capacitor at VC is adequate, and a series resistor should always be used to increase the slew rate on the VC pin to maintain tighter regulation of LED current during fast transients on the input supply to the converter.

How much simpler can this be? If you ask me, it is actually a little sparse on details, and its actually instructing the user to validate the design.

 

[treez]

The max16833 datasheet's equations leave the customer in no doubt. Why cant the linear one do the same?...those equations can carry over directly to the linear.com one...the only difference is that the linear chip (LT3756) doesn't have the 6.15 multiplying factor in it.

I trust that you also agree with me that for a 1:1 coupled sepic converter, the buckboost equations (as given on page 18 and 19 of max16833 datasheet) also apply to the sepic?

 

[schmitt trigger]

Why can't Linear do the same?................Because they want to leave the burden of proving that the circuit behaves correctly, in the designer's hands.

Also, the engineer writing the data sheet may just have, for lack of a better term, got lazy. Or did not have a thorough understanding of the circuit. Or left the company before finishing the datasheet. It escaped the proof reading. Many human errors come to mind.

About the SEPIC compensation. My understanding is that they have higher order poles and zeros as compared to a buck-boost. There is in fact a white paper by Ridley that claims that the SEPIC's loop is essentially un-tameable.

 

[D.A.(Tony)Stewart]

From what I read in the design, the feedback compensation mechanisms are quite different and both are non-linear so I would not trust the linear equations to be exact in the switchover to discontinuous mode. I would trust the simulations more with Linear's model's provided for their own chip's. One can test the sensitivity of tolerances for each part and compare with actual selected components and layout requiements.

In the end one must consider much more than Rule of THumb's and Mfg. guidelines but rather overall sensitivity, tolerances, ambient rise, chip rise , input step and output tracking response characteristics with loop margin. THese rely on the designer's expertise for environmental stress, input parameters and output behaviours.... including tolerance, regulation margin, stability margin testing and being able to DEFINE THESE REQUIREMENTS is MOST IMPORTANT. ... not just the existence of a theoretical resonant frequency of part values. THese may be fundamentals but not the entire design task. One must consider effects of ESR aging, Permeability drops with saturation, LED failures, design failures, component variations with suppliers, env. stress etc. etc.

I think your intentions are good, but your assumptions are incomplete..

 

[treez]

Thanks all....regarding the above comment that sepic is "untameable" in its feedback loop, I think I read that too, but Ridley was referring to the "uncoupled inductor" sepic, surely you agree that the coupled sepic has the same dynamics as the flyback?..which is the same as the buckboost if 1:1 turns ratio.