Hacker News Reader

Comment by palata | Hacker News Reader

That's super interesting! Would you have resources to recommend to learn about that?

For instance: where do you get that "the free space path loss at 20kM is 126dB at 2.4GHz in perfect conditions, or 134dB at 5.8GHz"? And why does 1 Watt translate to +30dBm?

View on HN

https://www.antenna-theory.com/ is a good starter for apprentice-level RF voodoo.

30 dBm = 30 decibels wrt to 1 milliwatt. 10 decibels = 1 bel = 10-fold increase of the base quantity, so 30 decibels = 10 x 10 x 10 = 1000x increase, so 30 dBm = 1000x 1 mW = 1 W.

Oh nice, thanks!

I don't know enough to offer any real sources, but I'll share that I've found ChatGPT very useful for learning stuff like this - really solidly documented and established science/engineering.

I asked* "When making a wireless transmission what is the signal loss at 20km for 2.4Ghz?" and it gave an excellent explanation of free space path loss, stepped through the calculations, and gave the correct answer.

I've had it explain various kinds of first-order filters, especially for electric guitar circuits too. And asking follow up questions works well with how my brain works.

*Link in case you are curious: https://chat.openai.com/share/d0fcf90e-9d2e-4078-9bb4-60717f...

> but I'll share that I've found ChatGPT very useful for learning stuff like this

I get your point, but no way I use ChatGPT. Other than all the ideological issues I have with it, I just can't trust it.

> and gave the correct answer.

Are you saying that because you believe it did, or because you already knew the answer? I don't trust ChatGPT, so I don't want to learn anything it says. And I don't need it for stuff I already know.

Because there are other ways (simple Google search, asking an EE friend, etc) of validating the answer. Often I'm looking to learn a concept so it isn't like I'm using it to solve a bunch of things. Just validate my learning. It is easily as good as having one of my EE friends explain the concept to me.

It's not that different to Wikipdia. Trust, but verify. It's usually correct, but regardless it's a useful starting point for a topic.

Wikipedia shows sources, does not hallucinate, and shows the same content to all the users (meaning that if there is a mistake, someone else can actually report it).

That is very different to me.

Those are accurate details, but if you zoom out they're still contextually very similar - a swathe of text on a subject written by someone (or something) you have no veritable trust in but is usually fairly accurate.

You should independently fact check either source. Wikipedia is slightly easier, but after you have a base subject matter understanding you should have the terminology to validate anyway.

I suspect you just don't like it, which is perfectly fine.

> Those are accurate details, but if you zoom out they're still contextually very similar

I cannot go against that. To me, the points I listed make them fundamentally different.

> I suspect you just don't like it, which is perfectly fine.

Because you don't understand why I see it as a problem that people don't make the difference between Wikipedia and ChatGPT does not necessarily mean that I am just making up arguments because I don't like it...

That's fair. I think we're arguing semantics here. Expanding my "you just don't like it" would be "you have immovable philosophical issues with one, ergo you won't pick it regardless of its accuracy". I didn't mean to imply you were just trying to be contrary - apologies if it came across that way.

I do like the ideals you have there, but I think for most people a tool is just a tool, and if it ms mostly accurate they're happy.

Reply on news.ycombinator.com