I will just add that I can recognize good technical writing in 3 stages of sophistication:
Stage 1) Someone who has barely mastered the topic and is at the most basic level just telling you the steps to reproduce the result or operate the items or workflow. Once you learn from the material presented (unless extremely complex or required to keep as reference), you rarely go back and study and appreciate this kind of writing repeatedly.
Stage 2) Someone who has done the analysis and exercise enough that the writing evolves into not just instructions or rote description, but narration through the reasons why they did it this way, the pitfalls to watch out for, what not to do. Starts you thinking about what to do next.
Stage 3) Someone whose telling of the technical detail is so fluid and practiced that the basics are unquestioned -- the writing now includes well-informed opinions and jokes about past attempts at the outcome, why they failed and their historical context in technical terms. Gives advice on the difficulties that will be encountered and the limits of the method. The writing is so informative that you may return to read it again and again throughout your career.
One of my favorite texts that reaches Stage 3 is Numerical Recipes in C (which unfortunately I don't have much occasion to refer to nowadays), and the way that programming philosophy, advice, and technical instruction combines in that work is very impressive. Also, many of Griffiths' physics texts, who manages to write with both technical and conversational/narrative mastery.
If only more writers could be like this.
I found the other recent thread on technical writing useful, too: Four kinds of documentation