Hacker News Reader

Comment by dfex | Hacker News Reader

More concerning - can anyone explain why there is such a variation in the results from the DC powered unit vs. the TENG-powered one? The graph at the bottom of the report shows a difference of 30-50ppm between both units when they are sitting side by side on the bench.

View on HN

That's in the normal range of accuracy of modern CO2 sensors, for example SCD40 from Sensirion is described with an accuracy of ±50.0 ppm ±5.0 %m.v.

That’s if the voltage supply was stable and within electrical specs for a sufficient period of time. We can see this is a snippet 2 hours into the discontinous collection.

3.6V is the maximum value that the nrf52832 SoC can handle. I would suspect the VDD is variable.

Actually the real accuracy is much better, especially with optical NDIR sensors.

We do a lot of testing against reference grade sensors and typically see very good performance and reproducibility [1].

[1] https://www.airgradient.com/blog/performance-of-low-cost-co2...

I skimmed the original article and it only mentions the graph and says that it's "comparable to DC powered unit". I'm guessing < 100ppm difference is somewhat acceptable?

You might be right - it's just odd that it's always showing "more" rather than similar amounts.

Also, according to Claude[1] a 50ppm difference is equivalent to around 25 years current atmospheric carbon increase.

* Pre-industrial (1700s): ~280 ppm

* 1958 (when systematic measurements began): ~315 ppm

* 2000: ~370 ppm

* 2015: ~400 ppm (milestone crossed)

* Current: ~420-425 ppm

[1] "What is the normal range for background CO2 concentrations in the air?"

It’s crazy to think that many people alive today experienced a 30% increase in ambient atmospheric carbon dioxide concentration within their lifetimes.

You’re missing some deeply important context there, which is that those measurements are for outdoor atmospheric CO2 only.

Average indoor air quality ranges from 400-1000 ppm CO2, with adverse mental effects starting to appear close to 2000 ppm.

In that context, you can see why a 50 ppm difference is marginal. This is why asking an LLM is not generally a great idea for understanding something - you need to follow it up with more research.

> adverse mental effects starting to appear close to 2000 ppm.

cognition is harmed starting at 1000ppm (https://pmc.ncbi.nlm.nih.gov/articles/PMC3548274/)

I'm having a hard time parsing the data in this paper - is it showing that task focus increases at 1000ppm compared to 600ppm CO2 exposure?

Not odd at all that it’s always showing more — sensor error is often biased. This is within the listed range though.

I have to know, what's the the rise of people like you chiming in with AI sourced tidbits? It's like the people with no knowledge on a subject that use google as a quick catch up tool so they can participate in a conversation, but somehow even worse. Are they the same people, but now lazier or just true believers in the non-sense engines?

Reply on news.ycombinator.com