10
\$\begingroup\$

I'm working on a project that is relatively ambitious for my experience level, and so I've lately been asking many design questions, and gotten helpful answers to them. Now, as this is a project that I'm actually building, and many of my questions are of the form "I have/think I might have problem X, I'm thinking of solution Y, will it work", I tend to eventually learn the correct answer (did it work or not) as I implement the project (often with the help of answers I've gotten from this site).

So once I've actually built and tested that part of the project, I will have an answer that's essentially correct by definition. Often by that time, I've long since accepted the answer which gave the best recommendation or most insight. However, I believe my own answer would also be very interesting to anyone who's interested in the original question, as it resolves the issue or tells if the proposed solution worked at least in the specific case I've dealt with.

So my question is what should I do with such knowledge? I don't want to edit my original question, as that would easily obscure the actual question. I'm thinking I'll just post what I've learned as another answer and leave it at that, but my worry is that people finding the question via google etc will then easily miss that there's an ex post facto answer. On the other hand, I don't want to move the "accepted answer" to mine, as that feels unfair to the people who originally offered their help and who I already once rewarded by accepting their answer, and I of course know more only because I have the advantage of hindsight.

One other option would be to post my answer as usual, and then edit the question to point out in the beginning that I've given an answer that tells what happened. This is what I'm leaning towards right now.

Anyway, what do you think is the right way to do this?

Some examples where I would soon post such answers:

  • About coupling noise from MCU to analog side: I'm now building the device, and I drew the PCB with the RC-filters recommended in the answer. I intend to test first with \$0\Omega\$ -resistors, and then swap those for the RC, so I can actually give numbers in terms of SNR for connecting with and without the RC.
  • The detailed PCB -design: I've built the MCU part already, and I know already that the MCU works, and doesn't couple much noise via the ground plane to the analog output stage. The analog section, including the DACs, is mostly unassembled yet though, but once I get that done and see much it changes things, I'll know how good my layout is, and I could report that in an answer.
  • a sudden worry about possibly destroying my board: turns out my worry was well founded (found out by measuring with a scope, luckily, instead of destroying the board), but for a different reason that I thought. Interestingly, none of the answers predicted the problem that actually occurs. Once I've verified my workaround works, I'll post the details.

Finally, here I already self-answered, and I think think I'm going to accept my own answer, as while the other answer gave helpful hints for the general case, mine gives a decisive answer to this specific question.

\$\endgroup\$
11
\$\begingroup\$

It is acceptable on SE to answer your own question. However, the issues you mention do apply.

You might edit the question by adding a section at the bottom called something like "Solution Eventually Used". Use the HTML h2 tags or similar to make this heading stand out, then you can write your solution below that normally.

People finding your question with its answers from a search will still see the original question, the various answers, and what you finally did. Since it's clear the final solution was added much later, the answers don't look stupid. Some of them may still be good for others in a similar situation.

You had to pick one way of solving the problem, but that doesn't mean some of the other proposed ways might not be good too. Note that your answer might be sufficient by definition since your system workds, but that does not prove your solution was optimal.

As a side effect, editing the question will bump it to the top of the active list. If the question is well written with a good explanation of what you tried and what you ended up doing, it will probably get a few upvotes. The top few answers will probably also get upvotes.

\$\endgroup\$
  • \$\begingroup\$ Good points all. So you recommend just editing the question? I'll do that, except maybe the transient question I linked, where I think my investigation with a scope gives a fuller answer than the others. I'll make sure to write it so that it's clear the others answered before, with less information available to them. \$\endgroup\$ – Timo Apr 24 '16 at 16:54

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .