Hacker News Reader

Comment by HeyLaughingBoy | Hacker News Reader

> often happens that a power rating of some part is exceeded

If this is happening more than rarely, you are really doing something wrong. Or you're prototyping a high power circuit and expect some mortality.

Still, understanding the power dissipation of every component in your circuit should be second nature.

View on HN

A tool like this is a fantastic way to help it to become second nature

All you have to do is read the datasheet and multiply a few numbers to calculate worst-case power. Remembering to do that should be second nature, like remembering to close any parentheses you open. It's a great tool for understanding in greater detail, but it's not the way to avoid breaking stuff.

"You won't make mistakes if you remember not to make mistakes" is nice in theory, but just not how people work in real world. Just like trivial syntax errors are very common while you're doing things. It's better to embrace the common human failures and have a second layer of protection.

I explained in my other comment how it's a great tool for other jobs, but the wrong tool for this job. You won't learn how not to break things by ignoring the proper tool.

You're not going to learn how not to break things this way, but it will help you spot non-critical mistakes while you're learning in different ways.

That's exactly what I'm saying.

if I was to give a kid a breadboard and a bag of components I wouldn't want them to avoid breaking stuff; what I would want is for the child to be able to interpret the event and gain knowledge from the mistake.

This idea can shift the paradigm from "oh, the LED doesn't work", to "Oh, the LED doesn't work, the color around that rectifier area has shifted; why?" , and I think that can help to build intuition.

So teach them to check the power and the paint will help them realize when they forget to do so.

what precisely is wrong with visual feedback?

Nothing. It's great. But it's the wrong tool for the job. Like teaching someone to stop a car by using the speedometer to estimate when to let go of the accelerator pedal, instead of introducing them to the brake pedal.

You are supposed to study datasheets before you start a car?

Please. There's nothing wrong with "hands-on first" approach. And the tortured metaphors completely kill your argument...

It's too late to edit my original comment. I was responding to the part about when you're breaking a lot of stuff all the time.

Checking limits is for preventing breaks. Visualizing heat is for troubleshooting when they happen anyway. You should do both.

> All you have to do is read the datasheet

Have you read many datasheets? Parsing a datasheet can be pretty complex, even if it’s well written (of which many are not). There’s quite a few figures spread across many sections, and a beginner might not even realize they should be looking for the absolute maximums section. Sometimes datasheets aren’t even that explicit about certain failure modes, you could lock up a logic chip by accidentally leaving a pin floating, for example.

I can certainly see some value in a breadboard with instant feedback for when you’ve fucked up. Ideally you’re using a power supply with an adjustable current limit to begin with but this is a cool way to quickly draw your attention to an issue.

It takes 5 minutes to show someone how to read maximum values off a datasheet, and it makes the visualization 100x more useful.

> All you have to do is read the datasheet and multiply a few numbers to calculate worst-case power.

True story: When I was in college the lab mate I had reversed unknowingly reversed power and ground on a chip on a breadboard. I turned off the circuit and reached for the chip and got a blister on my finger.

Maximum power is infinite in that case (well not infinite, but pretty damn high!) I don't recall seeing a data sheet on power consumption when power and ground is reversed...

No offense, but this comes across as "This was the way I was forced to learn it, and dang it everyone else should do the same." But my experience says, I wouldn't have got a blister on my finger if I had that.

Current was flowing through the "body diode" which is intrinsic to the construction of ICs (it's normally reverse biased). The forward voltage was likely around 0.6v, maybe a little higher with a lot of current flowing through it. So roughly that times however many amps the power supply could supply with the resistance of the leads/rails. It doesn't actually take that much power to make a DIP package extremely hot though, lacking any sort of heat sink.

I can't edit the comment now, but I worded it wrong. The quote only applies to exceeding maximum values. The visualization is much more useful if you've calculated worst-case. I use a thermal camera all the time.

Exceeding the power rating is more avoidable but sometimes I just get a footprint wrong or make a schematic error and I love my thermal camera for diagnosing that!

What's good for thermal cameras these days? There is some cheaper stuff iirc but in general the resolutions are just kinda sad (32x32 or 64x64) unless you go to expensive gear. And even then it's like, 256x256, not exactly the 64-100+ megapixels we get in traditional photography. I know you don't really need it but it's nice to have good gear.

iirc some of the phone stuff was like a couple hundred bucks, which is definitely within toy-money range and I'd get some use out of that even with lower resolution. Anything notably great for, say, under $2-3k that'd be worth jumping to over something basic? That's not a number that I'm unaccustomed to with higher-end photography stuff if it's worth the spend.

I've been researching this recently because every time I need to use my FLIR One, it's annoying because I need to find it, charge it, plug it in, and use FLIR's awful software. In the last few years, a Chinese company started selling sensors that seem to be far superior to anything you can get for reasonable prices from FLIR (256x192, and 25Hz since they're not subject to American export controls). I've been looking into buying a Android phone with one of their sensors built in as a replacement, because it seems to have a lot of advantages:

- not needing to futz with a dongle, way better displays than the standalone camera options

- never needing to charge it if you leave it on a wireless charging pad

- lots of internal storage that can automatically sync captured videos/pictures to the cloud

There's tons of options that look great in the $300-$500 range, which is a problem because it makes choosing difficult. This review has me leaning towards getting either a Doogee S98 Pro ($270), or a V20 Pro ($330, newer and has better specifications across the board except for not having wireless charging, so I'd need to get a wireless charging adapter for it): https://www.youtube.com/watch?v=4002_MqVNfY

Its been a little while since I looked at these sensors myself, and if love to know the name of that Chinese company or the name of their line of sensors if that’s easier, so I can search searching for devices using their sensors and even the sensors directly. I’m sure I’ll find them eventually but it’s so hard to search for some Chinese suppliers like this because of the garbage pile of inaccurate product descriptions on sites like AliExpress polluting the search results that basically turns finding a part into a game of cross referencing the effort of sifting through a garbage pile.

Whoops, forgot to mention the name. The company making the sensors is InfiRay. Doogee doesn't even have a product listing for the V20 Pro on their website currently, although you can find one on archive.org that says they're using the Tiny1-C sensor from InfiRay: https://web.archive.org/web/20230601144956/https://doogee.cc...

I searched AliExpress for Android phones with thermal cameras, narrowed it down to three brands (Doogee, Blackview, ulefone), eliminated Blackview because all of their product listings proudly mention that the thermal cameras are by FLIR, and then opened a bunch of tabs to compare specs and search YouTube for review videos that seemed like they weren't paid for. The video I linked above in my comment is from a prominent electronics YouTuber who explicitly states that he bought it on his own and I found another video comparing the FLIR One Pro and S98 Pro that looked promising (https://www.youtube.com/watch?v=QQ7NJPvPsk0), which is why I'm deciding between the two options from Doogee now.

Depends on what you need it for but I have a ten year old FLIR i3 with 64x64 pixels. It’s fine but the feature I would really like is that newer modules have both a normal camera and a thermal camera and they overlay the color image and the thermal image. That makes dealing with the low thermal resolution WAY easier since you can actually see which thing is hot. If you have a bunch of similar shaped chips on a board and all you can see is some rectangle is hot, it takes a moment to figure out which one you’re looking at. (I do this by covering each chip with my finger and seeing when the image is occluded).

So you don’t really need high resolution for most stuff if you have the color image overlay. I haven’t shopped for them in the last ten years before those types were common so I don’t have a specific part to recommend.

The thing is they are a fun toy but sometimes I question whether I should have spent $1500 on mine. I would suggest going on the cheaper end, not chasing specs for no reason, and just get the value out of it without breaking the bank. They’re neat but for my use case I’ve never felt like I really needed to have it. If there’s something that’s $300 go for it.

I don't have any PCB shots but I have some sample ikages from a cheap Noyafa which does 160x120 on my blog. The next model up (also much cheaper than FLIR equivalents) claims double resolution at 256x192

https://bbrks.me/noyafa-nf-583-review/

I have used a FLIR TG267 at work to find if parts are failing and while it's only 120x160 it works pretty well because it can overlay thermal coloration over a standard camera which really lets you use those pixels for temperature not identification.

This lets you know "ah this transistor is dead" instead of "what I am I even looking at".

Circuit simulators should make it easier to see the power dissipation of devices.

Reply on news.ycombinator.com