I have seen many reasons over the years why the microcontroller on-board circuitry for reset and brownout circuits does not work well. I could list out some of technical reasons but, as I said before, Prahlad has illustrated some of them already.

I can add however the following comments:

- The "digital" type circuitry used to make processor chips is often not particularly suited to making circuits that require precise design margins and tight thresold voltage compliance.

- The process control needed to build a microcontroller must be more loosey-goosey to achieve a reasonable yield due to the large variation of circuit parameters across a complex chip. A small chip of specific function can achieve reasonable yield with much tighter control of process parameters.

- Smaller companies with a focus on special needs market places are producing needed technologies. One example is reset and voltage supervisor chips for which they devise designs and processes for building such chips that may include things that are just not feasible for larger more dense chips. One example is the use of lasers to "trim" or "configure" a chip.

- The engineers that design chips are just as human as those that design PCBA's. Mistakes are made and scope of experience may be limited. I have over 30+ years uncovered numerous chip bugs that can be attributed to "human error". Just because a circuit is encased inside a moulded lump of black stuff does not mean that it is any more robust than any other circuit design.

With all that said I'll comment that it is possible today that there may be some manufacturers that are making microcontrollers that have very good on-board reset generators and on-board brownout detectors. But just because a company may demonstrate proficiency in making both microcontrollers and separate supervisor chips does not mean that the expertise from the left side of the brain gets utilized on the right side of the same brain.

So . . . accepting that it may be possible that some reliable chips have emerged I'll have to say that on MY projects I do not want to have to be the one to discover if this is true or not. My own accumulated experience over the years says that one place that I must pay close attention to is in the area of reliable reset generation and avoidance of software running amuck on a chip that is operating outside its specification limits. That has to include the concerns of chips that have on board programming and debugging circuits since marginal behavior can make very undesirable things happen. Thus every design I make or have a hand in will include a reset generator with reliable and precise voltage level detection and management. This design strategy works toward making reliable products that work well and do not need to be pulled back and re-worked and re-worked. For me as a product developer and the clients that I work for it is more profitable to be in the fast lane of the expressway rather than on the side of the road with a flat tire.

I am sure that there are some engineers that come to this embedded game with a less conservative viewpoint or apply a design strategy to "trust" chips more. This is fine and I wish well to all those that have success discovering if one chip or another is reliable in its RESET and BROWNOUT capabilities. But I must ask those same engineers if they are going to experiment with leaving board designs on two layers without ground planes and with no bypass capacitors.


Michael Karas