The thing a technical article most fears is becoming a repetition of the product literature. What avoids this is not the writing but the source of the topic being specific enough. Our topics have essentially three entrances, and each entrance has to arrive with a real question attached, otherwise the topic does not hold up.
Entrance one: the same question asked again and again on site
The same sentence being asked on several different projects means this is not an isolated case but a gap in judgement that exists generally. Why is the vacuum level of this machine different from that of another one, why is there a difference in the result after the same oil has been treated: behind questions of this kind there is always a specific person who cannot make a decision at a specific point. What the article has to answer is that decision, not a restatement of the question.
Entrance two: the follow-up question when the test data does not match
There are several ways in which the data does not match: what the instrument measured differs from what is felt on site, two tests differ widely, or the pass judgement runs against the performance in actual use. Each of them has something behind it that has to be traced. Written up after it has been traced, the article is worth far more than a description of the phenomenon. The relationships between water content, breakdown voltage and cleanliness also become clearer through this kind of questioning.
Entrance three: the parameters asked about again and again in purchasing and technical discussion
What is written on a parameter table and what gets asked about are often not the same set. What is asked about repeatedly tends to be the figures that are on the table but whose conditions of validity are not stated: under what conditions a given indicator holds, which part of the problem a configuration addresses, why two figures differ from each other. The answer to questions of this kind exists only in the comparison between the documentation and the actual conditions, and quoting the parameter table does not answer them.
| Source of the topic | What it looks like on site | The question the article has to answer |
|---|---|---|
| The same question asked again and again | The data given by different projects does not match | What is the basis of the judgement and to what range it applies |
| Abnormal test results | The indicator passes but the behaviour on site does not | Which items to look at first and in what order |
| Parameters asked about repeatedly | The entry needed is not on the parameter table | Under what conditions the parameter means anything |
What to write and what to leave out
Before a topic is drafted, go through it once: what action or what judgement can the reader take once they have this information? A topic that cannot answer that is not written. There is another category rejected as well, topics that are not wrong but could be covered under a different wording: each subject is written once, and an angle that has been covered is not repeated. Better to write one article fewer than to write the same subject again under a different title.
As for selection questions, the industry rarely has an answer that holds everywhere, so topics of this kind set out the conditions and the trade-offs: which configuration to choose under which conditions, which premises have to be confirmed first, and no conclusion that a particular machine should be chosen in every case. Data that is not in the equipment documentation is not made up by inference either.
Only the figures that can be checked are quoted
- The values are taken from the machine technical data and the manual, with the corresponding test standard stated
- Water content to GB/T 7600, gas content to GB/T 423, breakdown voltage to GB/T 507, cleanliness to NAS 1638
- Derived values and approximate values are not written; leaving them out is better
- Each standard number is matched one to one with the test item and the two are not mixed up
An article is worth writing only if it saves the reader one wrong judgement; copying the parameters once more helps nobody however many articles are written, and the parameters themselves can be looked up elsewhere.