We really just need a group like ConsumerReports, Labdoor, WireCutter or something that actually digs into the quality and claims of devices. Very few people are going to buy 30 competing options as well as a known good control and compare and contrast them in a scientific context, but that's what's needed.
I remember looking around for a Radon sensor for way too long. An electronic device, a litmus-like strip test, some analog crystal ball -- hell a mail-in lab kit bag of carbon -- but they all seemed inaccurate. The few people that "bought two" always showed heavy variance while the majority that bought and got a number *loved it*. Even buying two that agree often just means the products are consistently and predictably wrong, but most products don't even check this box ffs.
Reviewers that try and apples-to-apples all products by comparing cost, size, advertised features and ergonomics are doing no better than you or I opening up a package and appraising it. For awhile this Google dev was putting up extremely detailed amazon reviews for USBC cables based on their purported features v. reality and it was wonderful, but tucked away and undiscoverable.
The world needs multiple parties providing unbiased in-depth reviews, but there's very little will to pay for even one group to do something like this. Ironically, paying for information in the information age seems almost comical to today's consumers and while there's demand for good information, no one's willing to open their wallets. Labdoor moved from subscription-only memberships to "certified products" and affiliate links because everyone wants to know what's in their multivitamin, but no one wants to pay for an indepth review. We aren't cultivating a supply-side that works for us, but against us.
The result of this is shilling and fake reviews running wild. I mean hell, consumers will buy a product knowing full well they might get a shoddy knock off and yet it's $5 less and they can get it tomorrow so they'll roll the dice. Sadly, we don't live in a world where people are intimately concerned with quality and accuracy in their purchases.
For tools and other items that you might get from Home Depot, Project Farm is pretty much the standard now in my book. https://www.youtube.com/c/projectfarm
His video on Water Purifiers only measures one dimension of water purification: TDS measurement. This leaves out a whole world of other aspects to consider when testing water and filtration quality. While TDS can give an indication of cleanliness of the water you can only glean from what you can measure, to this end the TDS measurement will not catch all contaminants in the water because it relies on electrical conductivity.
Also he does not disclose that his TDS measuring device is actually bundled with one of the devices he is testing. Should have used a third party tool to verify that it is giving an accurate reading instead of a unit bundled with one of the water filters as a marketing ploy. It gives some doubt as to how seriously he takes his other tests. These are the things you miss if you don't actually know any better(I assume the majority of his audience have not looked into what makes good water filtration before watching his video).
Nevertheless, the fact that he is really the only one doing these kinds of tests is enough to give him some sort of super reviewer credibility.
From his pod casts it looks like Linus (Linus Tech Tips) is building his lab up to be a massive consumer review machine, can't wait for that to happen. But I doubt they'd be able to tackle the massive amount of gadgets like USB cables, USB hubs, USB docks, chargers, air purity weather stations, etc. Plus their reviews often aren't in-depth enough for people to make their own mind up on stuff.
Linus has mentioned that part of the intent of LTT Labs/Lab 32 is to not just release video reviews, but to produce comprehensive datasheets on products with some sort of open-access release model. I think the video they did testing HDMI cables[0] is sort of a prototype for that.
Project Farm is poorly controlled, sample-size-one, poorly-designed-but-very-specific "testing." It's entertainment, not actual useful product testing, and a good example of this would be his "let's see how many bolts this impact gun can remove in X amount of time" - each time a bolt is screwed on and screwed off, it polishes the threads, changing the amount of resistance.
It's of slightly better quality than the usual youtube comparisons.
Channels like his are wildly popular because corporations prefer inexpert entertainers aka "influencers" and "creators" over product/market experts, so they shower the popular ones with free shit to "review."
This is one reason you see e-bike companies heavily courting the tech press; the tech press know fuck-all about what to look for in a bicycle, so for example, they don't notice that the bike has Chinesium bearings that won't survive being ridden in the rain, terrible no-name tires with little flat protection and high rolling resistance, and brakes made of chewing gum that perform much worse than proper offerings from major brands.
This is pretty wildly inaccurate specifically in Project Farm's case. Most "consumable" parts of tests, like bolts, are swapped out from test to test. From what I've seen he's pretty transparent when that's not the case.
He also explicitly buys all products with his own money (some by way of viewers contributing, of course) and products are sourced from what folks in the comments are asking for.
Scientifically, one should test multiple “identical” components for each test and average the results (or do something like a box and whisker plot) to incorporate manufacturing variability.
But his results are authoritative as the bar is low.
> and a good example of this would be his "let's see how many bolts this impact gun can remove in X amount of time" - each time a bolt is screwed on and screwed off, it polishes the threads, changing the amount of resistance.
He's smart enough to know that...
> Channels like his are wildly popular because corporations prefer inexpert entertainers aka "influencers" and "creators" over product/market experts, so they shower the popular ones with free shit to "review."
> while the majority that bought and got a number loved it
This seems to be Amazon's main unspoken strategy behind promoting positive reviews... many people want to be helpful and provide a review of a product that they purchased. That's great, but the problem is that most people are thoroughly unqualified to review things: "I took it out of the box and turned it on, five stars!"
(And I'm not even going to mention Vine Voice or whatever it is called, these are reviews you should ALWAYS ignore because they are given free products in exchange for ostensibly unbiased but implicitly biased reviews.)
Agreed, and that's if they review the product at all.
Lots of reviewers talk about the delivery, how Amazon Prime bungled the return or even a completely different product; all completely unrelated to the actual features of the product you'll be buying. Pair that with how Amazon allows companies to put completely different products on the same offering with "variants" that are again abused by companies to transfer reviews from a well received product onto a poorly received one. Most users just look at the five start v. one star ratio in the end anyway so an electrician going over a wire gets the same final weight as someone that confusingly put a question in the review box.
I've put up good quality critical reviews too and had them silently removed. Amazon may be an okay place for getting an item, but too many people use it as a discovery and curation vehicle and that's where the manipulation starts and ends. I get better product recommendations looking almost anywhere else than the Amazon search box.
There are lots of "favorites", but one of my favorites is when someone rates it highly and says that they ordered it, haven't yet received it, but are very excited to try it. So in other words, they have no idea, but they are so hopeful and excited they already rated it!
I would add that something like this often happens outside online shops for devices and more generally house appliances.
My theory is that many people are "proud" to have bought something, either for its novelty or because they managed to get the whatever for a lowish price/discount, or - on the opposite - because they already spent an awful amount of money so it must be exceptional, or because they read somewhere that the thing is exceptionally good.
Short of the thing completely not working, they are (in good faith) convinced that they got the best and talk endlessly on how good it is, and how nothing else is as good.
Once upon a time it was typical of many self-appointed Hi-Fi experts but in more recent years I have seen the same behaviour in many other cases, from vacuum cleaners to Tv's.
Both of these are caused by the emails sent out, where unsuspecting people try to do their best to answer the question or write a review - sometimes thinking they have to.
Yeah. Amazon didn't start the answers section like this. Once Amazon started emailing random purchasers to answer questions about the item they purchased, the quality of the answers went way down.
They'll do more than a discount. They have entire groups dedicated to giving out entirely free products in exchange for five star reviews. They provide entire catalogs of products available; you put in an order request(s) then get a confirmation to place the order. Then you buy the item on Amazon and leave your 5 star review. After the seller verifies your review you are sent full reimbursement typically through PayPal.
I got invited to join one of these groups from a friend and got tons of free products on Amazon this way. I remember getting a free home security system that was like $500 or $600. I never even took it out of the box, just re-sold it on Facebook Marketplace.
Eventually I left the group because I felt bad leaving these good reviews when some of the products were absolutely garbage. It's an unfortunate situation for the sellers though too; people don't buy products on Amazon that don't have reviews. So if a seller can't bootstrap a few good reviews early on then they won't ever start making real sales.
Wait till you see AliExpress reviews. Often it's, I received the package, yay it arrived (photo of package). I haven't opened the box or turned it on yet. 5 stars.
To be fair, just getting your order from Ali is a momentous achievement :P
Reviews are all like 'Such an improvement compared to my 10 year old 30 inch plasma 720p TV' or 'noce, it has Netflix'!
Not a single review mentioned that Android TV takes like 3 seconds to respond to any buttonpress on the remote because it's running a 8 year old ARMv7 CPU, basically like the first raspberri pi.
Since the 1960s we have "Stiftung Warentest" in Germany which does unbiased indepth product reviews and is a very popular magazine. They review more than 200 products every year and the tests are carried out by independent external test labs worldwide.
I bought one of their prior recommendations, the "Kaiterra Laser Egg+ Chemical", and as far as I could tell it did the "TVOC-to-CO2" lookup table conversion mentioned in this article, and so it would give CO2 readings that were inconsistent with same-branded CO2 sensor ("Kaiterra Laser Egg+ CO2"). WireCutter seem to have since updated the recommendation to something with a different type of sensor.
One of the reasons you may be having a problem with radon sensors is that most radon sensors will report results with a resolution in tenths of pCi/L (in the US) while the statistical variance in the range below 4.0 pCi/L (where most houses should be) is +/-25% or more for many common sensors. So you can have two readings which are e.g. 2.4 pCi/L and 1.9 pCi/L, and from the standpoint of a radon measurement specialist, these readings agree!
You might say, well, +/-25% sucks. You wouldn't accept a thermometer or pressure sensor which read +/-25%. And you are right, it does suck, but then try to design a radon sensor with a sensitivity of more than a few counts/hour per pCi/L but which is also affordable... you could spend $10k+ and get an Alphaguard which has a huge sensitivity and thus very little statistical uncertainty, or spend $25 on a charcoal kit which is +/-50% if you are lucky, or any of the options in between. If you buy a detector which does not explicitly list the sensitivity in the specs, you can be sure it absolutely sucks.
Because many detectors are measuring in single counts per hour, you have to measure over many hours (usually 48) just to get enough statistics to make a decent estimate of the radon level. Decent in this case being +/-10 to 25%. Then you add other sources of error like temperature, humidity, air speed, etc. If the sensitivity is absolutely abysmal, you may need to measure for weeks or months to get enough counts to know what your radon level is to +/-10%.
Often the best bang for your buck in radon measurements is to hire someone who has a really nice radon measurement system for a short term measurement. However, if your radon level changes over time then you get a good snapshot only of whats going on right then. For most homes this isn't a problem, but occasionally you run into a situation like one guy I remember whose well system had radon saturated water, so his radon was fine except when the shower was on.
It is an interesting thought to contemplate how one would disclose lightly radioactive water to a potential future purchaser of the home. I got curious and looked it up, apparently you can purchase an aeration system for $3k-6k to take the radon out of the water.
Unfortunately, there is no business model to support this. People don't want to pay to support unbiased product review sites. And as they say - "if you're not paying for the product, you are the product"
Project Farm has unsponsored empirical comparative reviews of multiple products for the same use. He has affiliate links for every product tested so he makes commission no matter which one you buy. Compare this to a one-at-a-time reviewer where the reviewer is motivated to sell each product as much as possible.
The only perverse incentives are (1) he makes more if you buy the more expensive one and (2) he doesn't make money from free/DIY solutions. 1 would require knowing the market really well and faking tests on video. W.r.t. 2 he's demonstrated good faith by including DIY and vintage options.
Was that single topic enough of a reason to throw away all the good content he makes about other things? It's not like his opinion on the value of a vaccine requirement will make him off base in his knowledge of machining and tools.
My dad has garbage political opinions, and is generally what I would call a low information voter who is swayed more by advertising than vote record, but he's still a very talented general contractor who can build anything and knows his stuff.
I pay for Consumer Reports. I used to love wirecutter - and then bought some of their recommendations that were garbage.
Consumer Reports sometimes overindexes to things I don't care about ("complicated technology"), but they usually put in the work and break it down well.
That's a new word to me, "overindex". Had to Google it, and that was kind of fun. It seems to be from economics, but nowadays a bit of a Microsoft term, maybe?
Here's Raymond Chen in 2017 explaining it [1]:
To over-index on an attribute means to give that attribute too much prominence in a discussion or analysis.
Also fun to note that Wictionary [2] does not mention this meaning at all, instead saying it means to index out of bounds in an array which given the technical usage sphere of the word is kind of amusing.
i think they grew too quickly and just hired some people that were "content" makers not real reviewers, who weren't spending the time required to actually understand, nor actually have any scientific method.
I wouldn't be surprised if some of their reviewers were getting paid on the side for their recommendations (it seemed like it might've been freelanced out reading some of the reviews).
I've seen some signs that they might be getting better... let's hope.
But Consumer Reports has always been great for me.
> Reviewers that try and apples to apples all products by comparing cost, size, advertised features and ergonomics are doing no better than you or I opening up a package and appraising it.
Unless you're buying every product on the market and load testing them in your garage, some folks do a lot better than that (e.g. project farm, or torque test channel).
But it clearly requires time and passion, and serious rigs for some of the tests (not everybody has a Skidmore lying around).
There are some good reviewers out there yeah. Sadly, they often rely on the creator's passion moreso than getting paid by their audience to provide a service. Youtube ad revenue and Amazon affiliate links are keeping these people's lights on more than the "thank you's" of their most ardent "supporters".
Google tightening the belt on Youtube monetization is probably, strangely, the best thing that could have happened to mostly-direct payments to content creators. Patreon and the like have made viewers at least somewhat more willing to pay out to access content. On the flip side, it seems to have also led to an explosion of more biased reviewers pushing their own discount codes and affiliate links to make up the difference.
There's a cultural misalignment somewhere. Journalism still hasn't cracked the nut nor has scientific peer review. Information curation and verification isn't cheap regardless of how much we'd like it to be so.
> The world needs multiple parties providing unbiased in-depth reviews, but there's very little will to pay for even one group to do something like this. Ironically, paying for information in the information age seems almost comical to today's consumers and while there's demand for good information, no one's willing to open their wallets. Labdoor moved from subscription-only memberships to "certified products" and affiliate links because everyone wants to know what's in their multivitamin, but no one wants to pay for an indepth review. We aren't cultivating a supply-side that works for us, but against us.
I feel like there might be more willingness to pay for this sort of stuff if people knew that they are being straight up lied to. Many people have a sort of vague "yeah probably the company is doing something scummy" attitude, but, at least in my experience, people still mostly trust that their devices do what they explicitly promise to do. Buying a TVOC sensor that doesn't actually measure TVOC at all somehow feels too outrageous to be true.
It's because we used to regulate against the ability to sell something which didn't do the main task it purported to do. Was that not the entire point of the BBB?
Here's one marketer's belief about its value in 2022:
"at this point, I believe the only people who still think the BBB means something are the older generation. And business owners seem to pursue BBB accreditation mainly so they can slap a logo seal on their site." - https://www.blogmarketingacademy.com/better-business-bureau/
Not sure what the BBB has to do with regulation? They are just a private organisation.
(And you are right, that they don't do a good job of what they are ostensibly trying to do.)
> It's because we used to regulate against the ability to sell something which didn't do the main task it purported to do.
If someone lies about what the product can do, can't you just sue them for fraud or so? (Do a class action lawsuit, if the individual damages are too small.)
Even if such a service existed, it wouldn't accomplish much. OEMs are free to change the design and internals of their products at any time and many have had no choice but to do so because of semiconductor shortages.
Oh, of course. You can compare external product numbers, but often these aren't updated when things change. That's not even getting into buying Levi 501's from Kmart will get you an inferior product compared to buying them from levi.com despite being marketed as the same product. Companies too will often abuse their brand and former good will in order to ship sub par products today that look almost like their products of yesteryear -- doubly so after an acquisition. I have a friend that is into Rubiks cubes and when asked which cube I should buy he just said.
> Any of them are great, except if they come from Rubiks
I still think being defeatist about this isn't helpful. More information, even if it's not complete and timeless is good. I think if this kind of information proliferated too, companies would have a disincentive to update what is loved, works and isn't broken.
Also, I realize the chip shortage is a real issue and has had genuine effects... still companies will find any external boogie man for why their products should get weaker, their turn around times longer and their products' sticker prices higher.
You can't forget the "but verify" part of "Trust but verify"
this is literally nonsense -- every electronic device in the USA sold to consumers must conform to basic standards, and stay that way. At face value, the reply says "you cannot really rate devices"
The most upsetting thing to me several years ago was that I couldn't find a way to verify a product's UL rating. The sticker on the product can say whatever it wants, but I couldn't find anywhere on UL's website to verify that the sticker was genuine.
You are just describing a company, a billion dollar company indeed, because I would buy from a company I could trust on these matters. Happy to pay the extra 10% or 20%, as long as you don't screw me on details, quality, etc.
My cheap co2 sensor shows dramatically different readings for my office with the window closed vs open. It is great at reminding me to crack my window open to get more air flow.
Its temperature sensor is within a couple degrees of my nest's remote sensor.
My Amazon vo2 sensor can tell if I turned my kitchen fan on to a sufficient level to clear out whatever I am cooking.
Lots of sensors don't need to be precise, they just need to indicate if some action needs to be taken.
e.g. If there is a fire, I just need the alarm to go off, I don't care how hot the fire is.
I have an $80 monitor and it doesn’t seem trustworthy. It has very similar low values 90% of the time, barely ever correlated with forecast AQI I get online. And even when it was wildfire season and I walked out in the smoke (150 AQI) with it the values only bounced around a bit but still were <50 AQI 80% of the time and still <100 AQI the remaining time.
I rather like the environment devices for Raspberry Pi. Surely they aren't "professional level," but for things like CO2 and other particles and gases, you can generate scrolling graphs. That lets you always know whether things are going up or down at any given moment.
Most of the store-bought devices just beep if too high etc, but having a visible graph certainly tells you if it is measuring anything at all, and if you need to crack open a window or blow out the novel candle you just bought from the dollar store.
Agreed, as long as deltas are consistent I find the devices useful. I feel like fitness tracker step counts fall into the same category--I don't care much about the absolute number is, but rather how that changes over time and correlates with other behaviors.
Relative VOC measurements at least are able to detect a no. 2 in nappies. Great to distinguish if you should go in and change them or would be a distraction from falling to sleep when checking. I bulit a Pi ZeroW that measures on the side of the bed and publishes a curve of the last hour of the values. You can see a pretty salient drop when "it" happens.
I have an Awair sensor, and I can confirm that the VOC sensor is able to detect...well, pretty much anything from food to hair products to outgassing from some new thing that I opened up and accidentally left too close to the sensor.
What sensor are you using? I bought a few trash ones from Microcenter. The readings were all over the place and I couldn’t get reliable numbers without smoothing numbers and throwing readings out. One of the spec sheets even said you would get crazy readings and would need to throw them out.
I'm a little surprised that no one has mentioned Aranet yet. From what I can tell, they appear to work quite well. However, it is quite pricey.
I did look into the AirGradient when geerlingguy put out a video about it but they were out of stock at the time. It looks like there might be a new version but that may be something worth looking at.
I tried the Vitalight CO2s but the readings were all over the place despite recalibration with 2 devices side by side. Ended up returning these. Do not recommend.
True. You need to know that and see if you can live with it for your use case. Some information is available at least, like the temperature coefficient of the pressure sensor [0]. Bosch offers an "interpretation library" that you also can train yourself specifically, that can be used for the successor BME688 [1]...
yah, i think a lot of these IAQ (indoor air quality) products are ok at relative differences (which is mostly what lay consumers care about). the article mainly points out that they're not good at absolute measurements, and for something like VOCs, it's likely using a CO₂ sensor and a lookup table that's static and only really valid for the test room it was created in.
> for something like VOCs, it's likely using a CO₂ sensor and a lookup table that's static and only really valid for the test room it was created in.
It's the other way around, usually. CO₂ is more typically the estimated value, since its direct detection involves a relatively expensive NDIR value. VOCs as a category need some sort of modeling and estimation, but sensors can go fairly far by directly measuring a few species (hydrogen, for example), and hoping.
For sure. I have a couple of Winix air quality sensors / purifiers and they power up (with red LEDs and rushing fans) almost immediately if anyone has a little gas anywhere in the room.
Maybe a voltage drop? Most sensors just return a voltage. And then you have to calibrate that value and convert it into some sort of measurement using an equation. So an increase in VOC might result in a drop in voltage on the sensor.
I have a few of the Airthings View Plus units. Can’t speak objectively but when there are fires or something degrading air quality, PM2.5 rises, when all windows and doors are closed, co2 rises and then falls when I open the windows and turn on the fan blowing the air out. VOCs rise can be correlated directly with the nest turning on our old AC unit and then with it turning off and fanning the air out the windows. Radon seems to rise with outdoor temperature when it gets hot out. One per floor of the house and they are all correlated with the things above, and the VOCs in the room with the most air conditioning blowing rise the most, so I feel pretty happy about their efficacy.
I’ve been happy with these units so far after six months.
Happy user here. I started with 2 of the BT connected widgets, that used my phone to upload to the cloud. I then added to the Airthing View Plus which does wifi, BT, and if you plug a cable in will act as a gateway from bluetooth to the airthings web service.
Seems like a cool company, and have seen things like forest fires, using a gas stove, and burning things while cooking all correlate with readings. In particular I noticed that 3am in my bedroom has a pretty big CO2 spike. I also monitored radon before, during, and after some radon mitigation.
I really like that at least some of their units allow exporting their data to graphana, without any cloud involvment. I'm unsure if the view plus does the same though.
So for the HN crowd, lots of sensors, can use a raspberry pi as a gateway, your phone, or a turn key view plus. Graphana support is a big plus and you don't have to use any cloud service if you don't want to.
I have these for a year and I'm rather happy with them. The view plus acts as a hub for the minis and the values measured correlate with smells and stale air feel of the rooms. They did help me significantly improve my sleep quality.
The fact that they have a good API made them worth the fairly steep entry price.
We got at least one which seems faulty, but to their credit they immediately got back to us and asked for more data for their engineering to look at. The numbers produced are obviously very very wrong. They seem solid otherwise and are accessible with open source tools.
Not OP, but I think the View Plus is the only one with a screen. I have the Wave Plus that is BT. I simply use a Home assistant and ESPHome to integrate everything network wide.
The pressure is constantly less than 1000 hPa (after 1 year of having the device, not once was higher (I live at 100 m msl) so it must be higher. When I reported the problem, the technical support said "we use Bosch Sensors, so it is impossible that they are wrong".
Not only the service is trash, but also the whole interface, which requires permanent internet connection. You can access from your phone the status of the air at your home from anywhere in the world... my fear is if anybody else also can. Notably, after 1 year, there was no SW update... I really doubt the device is safe.
I only have it because of Radon measurements. And I hope they are at least more or less ok.
We launched our indoor air quality monitor AirGradient ONE at the beginning of this year. We measure PM, CO2, TVOCs, temperature and humidity. When we started the development we would have never thought that achieving an accurate temperature measurement was by far the most difficult thing.
It took us 8 hardware iterations to get it right! We experienced with different enclsoure shapes, orientations and PCBs (component placement).
Our experience is similar to the author of this article:
Your biggest enemy is internal heat produced by active components, mainly the MCU but also by other sensors like the PM sensor and CO2 (NDIR).
How we solved it:
1) Minimize heat from energy by using components that have low energy consumption and make extensive use of sleep functions of the MCU as well as sensor components (e.g. PM sensors can be put to sleep).
2) Make use of physics. Our monitor is wall mounted and uses the convection mechanism. So nearly all components that heat up like LEDs, MCU etc. are located at the very top inside the enclosure. Optimize the vents at the top and bottom of the enclosure and create an unobstructed air flow through the enclosure based on natural convection.
3) Plan for enough space. To get temperature right, you need a lot of air around the temperature sensor that is free from radiation from other heat sources. So we specifically designed the PCB to have the temperature sensor at a 'finger' at the bottom with a lot of space to other components and ensured there is a good air flow around.
If you are interested to see our design, have a look at our open-source, open-hardware DIY Pro monitor [1].
We integrated a lot of the things we learned designing the commercial monitor mentioned above into these build instructions [2] that allow you to build your own highly accurate monitor and then also be able to change the firmware and adjust to your needs. In fact, our kit contains exactly the same plastic injected enclosure we use for our commercial monitor (but you can also just 3D print the one from the build instructions and get the other parts by yourself).
In our journey to get the temperature right, we made extensive use of thermal imaging and compared our prototypes with competitors.
The images showed that one of these sensors got extremely hot during operation but the temperature was nearly identical to the real ambient temperature. We couldn't believe this.
So we made a test and put a strong fan onto this sensor. The temperature then dropped by more than two degrees Celcius BELOW ambient which showed clearly that they made a software offset.
It can be debated if such an offset is a legitimate design choice (and I personally believe it has a lot of risks) but it was something we wanted to avoid in our design.
Guilty as charged. Based on design validation testing, I ended up making the design choice to partially rely on a software offset, a strong fan blowing into the sensor not being a supported scenario.
But the placement / air flow / multiple openings definitely took us more design iterations than we had planned for.
Looks like non-professionals have easier time with this. :) I did put DHT22 into a breadboard, some atmel mcu on the other side to convert from DHT's protocol to UART and half a meter of wires to UART of Orange Pi SBC which serves as a logger/controller. Open air design, lot of space, far away from anything that heats up air.
Sensors hanging off wires may look ugly, but it has benefits.
Your product looks interesting but I'm curious why you have a "contact for pricing" for your AirGradient One, are you targeting B2B sales with that device?
I'm looking at your build kits which are pretty cool. I'd probably get the "PRO Pre-soldered" kit. Couple questions about them:
> PCB has slots for TVOC, pressure or other sensors (sensors not included)
Do you sell these sensors separately? or have recommendations on sensors you trust and work with this kit? Your AirGradient One comes with TVOC sensing built in so you definitely have something
> 24 months AirGradient data platform (or connect it to your own server, e.g. home assistant)
Can I use the device without connecting to a server?
I live in an area prone to frequent fog for many months out of the year.
I learned that the fog will thoroughly disrupt the accuracy of pm2.5 sensors.
After some research I found that running the air through a heated pitot or similar before taking a measurement will almost entirely correct this for somewhat obvious reasons; the fog is evaporated.
As far as I can tell, none of the mid to high range weather stations have such a system and they will all be corrupted by a sight fog.
I’ve resigned myself to making one… Does such a sensor even exist?
> most of my reviews would be worth dirt a month after publishing when either the items go out of stock or get their insides silently changed
Very sad reality, from clothes to electronics. One wonders whether this run for cheaper builds while remaining relatively conservative in prices, unrestrained by a - missing - critical mass of discerning, demanding consumers, will end through some fading of the economic causes that triggered it. Or if this culturally polluting trend will stick now that it has been installed as a "normalcy".
(Since I see that this is this quite alive: Update: real time: I have just been told in a shoestore, "No, forget about the good quality you could find years ago: it is either "meh" shoes or shoeless". I mean, I have good hope that if one made of finding good products a job in itself, with extensive effort, some reservoir may eventually be found, but really this whole situation is /very/ suboptimal.)
I'm just looking for an indoor air quality monitor for temperature, humidity, co2 and ideally pm2.5 and pressure. Can't believe such trivial product is so hard to find.
There is the AirGradient DIY but I don't like the design, I'm worried about temperature accuracy and response time and I wish it was pluggable into a wall plug.
In addition I want to buy an outdoor water-proof temperature and humidity sensor running on a battery and communicating over WiFi or BT. Also can't find anything.
>I'm just looking for an indoor air quality monitor for temperature, humidity, co2 and ideally pm2.5 and pressure.
That's because it's $$$$. CO2 is hard to measure (it's inert!), and only recently advances in NDIR allow us to measure it directly. Those are still $50, and that's just component cost.
PM2.5 is also around $50 for decent parts.
Another $10 for T/H/P. Another $20 for cheapest LCD you can find.
Ok, so we are up to $130, but now you gotta pay for manufacturing ($$$), PCB, hardware development, calibration, testing, mechanical design... Probably $200-$300 total depending on quantity.
So, let's say you can make the whole device for $250 just in costs. Now you gotta add software (BT\Wifi code is not fun!), shipping, overhead, profits. So, you gotta sell this thing for $350+.
I cannot personally attest to the build of this particular product, but we have used a number of omega sensors in the past and they are not a fly-by-night outfit. The cost aligns with your estimates.
I would if it was built well. There may be a bunch of companies offering something similar but without being in the industry I have to come to places like this to get real world anecdotes on whether it's reliable or not. That inherit distrust makes it hard to buy something like this - to differentiate between marketing and quality.
Thank you for mentioning us. Since a few months we have an improved AirGradient PRO kit [1] that has a much better design, temperature accuracy and enclosure. It is fully open-source, open-hardware including 3D enclosure files.
I run 4 AirGradients with ESPhome hooked up to Home Assistant and couldn't be happier.
In case you're not aware, the kit was upgraded this year to move the temperature sensor a little further away from heat sources, they now bundle enclosures with the kit (if requested), and also sell [0] pre-soldered kits.
> I'm just looking for an indoor air quality monitor for temperature, humidity, co2 and ideally pm2.5 and pressure.
If you don't want it to be "connected," try a TemTop P1000. I've torn one apart, and I have it in my office right now - it seems to do what it says on the tin sanely. Temperature, humidity, PM2.5, PM10, and CO2 levels. I have another TemTop unit and they generally agree within some reasonable margin of error, and they do seem to independently measure things like VOC vs CO2 - I can see one rise without the other, depending on what I'm doing.
Indeed! Although it seems the point of the article is that while these are very basic measurements, building the technology to reliably provide correct readings, especially from the same board/housing as a WiFi board, is not trivial. Sure, it's not as difficult as more complex or obscure devices, but it is a lot more than cut & paste some sensors onto a board with some WiFi & slap it in a box... which seems to be the usual level of effort
Does anyone know where we could get such devices that 1) are actually designed to be reliable, and 2) don't need to exfiltrate information to the seller's servers in order for us to access it?
This one has everything listed but pressure and it’s possible to replace the sensors if you think they’ve gone bad, or migrate them to a new project. Since it’s obvious what the sensors are, it’s also possible to read the actual data sheet for each one and decide if they individually meet your needs.
Our old basic kit that you link above has also temperature inaccuracies due to its small footprint but we have a new kit that addresses these issues and is also available as a pre-soldered version for easy assembly: https://www.airgradient.com/open-airgradient/instructions/di...
I'm still planning on buying a CO2 monitor, but I'm really not looking forward to it. My plan is to buy them three at a time and if I don't get the same reading (or damn near it) on all three returning them as defective and buying another three from some other random no-name Chinese manufacturer. I'm guessing the whole process is going to take many months and a lot of my time, and in the end it still can't prevent a company from just giving identical but incorrect readings.
I built a CO2 meter around a SCD30 nondispersive infrared sensor a few years ago. (~$60) Gives readings with PPM resolution, which is good enough, since you're only going to start noticing physiological effects above 1000 PPM.
As the article mentions, there is substantial sensor self-heating. (The SCD30 uses a little incandescent bulb to produce 4.26um light, and the whole package draws 375mW during measurement) It wants to be very well ventilated, or polled very infrequently, in order to remain anywhere near ambient.
Claims to be 3.3V compatible, (Rpi voltage) which I haven't verified, since I ran it off an Arduino. (Which is 5V)
From the Seeed docs: "When activated for the first time a period of minimum 7 days is needed so that the algorithm can find its initial parameter set for ASC. The sensor has to be exposed to fresh air for at least 1 hour every day."
Disregard that entirely. When I tried to use the built in autocalibrate, it would set the floor value to 390ppm at some random time of day, not even close to diurnal minimum. Not to mention outside air is more like 420ppm now. (The problem with burning a minimum CO2 number into ROM is that the number is always going up...) I just disabled autocalibrate, stuck it outside for an hour, and hardcoded an offset in user code.
The Seeed carrier board puts an ornamental cover over the sensor to make it look prettier. You can remove it if you want to reduce self-heating.
The SCD30 uses i2c address 0x61, so it should coexist on the same bus as the BME280 temperature sensor, (0x76) if you want more accurate room temp: https://www.seeedstudio.com/Grove-BME280-Environmental-Senso... (Seeed will not warn about address incompatibilities during checkout if you buy two sensors with the same address. Ask me how I know.) If you enjoy random number generators, you can also pick up one of the bad hot-wire "air quality" sensors that GP dislikes. https://www.seeedstudio.com/Grove-Air-Quality-Sensor-v1-3-Ar...
Disclosure, I used to work for a company that resold all this stuff, before it was bankrupted by the semiconductor crisis.
Try buying the CO2 monitors intended for agricultural/greenhouse use from a agricultural specialty store that's within your country. They won't be as inexpensive as the consumer units but at least you'll have some confidence that the readings are accurate.
NDIR CO2 sensors are reasonably accurate. You can look up if the device is using an NDIR sensor before you buy it. Monitors that estimate CO2 based on a VOC sensor like mentioned in the article aren't using an NDIR sensor.
You can just call outdoors 400ish, it doesn't vary hugely, or just take a bunch of measurements outdoors. Some sensors tend to zero offsets/drift periodically when they "detect" they are outside.
I bought a couple of expensive remote weather sensors because they were the only ones I could find that included sensing atmospheric pressure and wet bulb temperature along with the ordinary metrics like temperature and humidity. Though air quality and particulate count is not included, I have done without for two reasons, which is that I live in the country surrounded by water and forest and cleaner air than I have ever known, and in contrast, the small house is old, dusty, I can easily see the particulates floating in the air, and because I know it is bad, I just don't need to know any precise measurements. My only complaint concerning the devices, which seem to be rugged and accurate, though with an almost absurd accompanying price tag, is that the required app and its interface design is comically bad in every measure that could matter. If there are interested developers, please take a look at the product[1] and the software[2], and see if you're interested in developing and selling me and other customers, and perhaps even the manufacturer, an application that collects and displays data from this family of devices that doesn't completely and utterly suck.
Indoor air quality can be greatly affected by all kinds of things in your home, from the building materials to the paints, coverings, veneers, veneer glues, fabrics, etc. etc. etc.
So in other words, while your outside air may be quite good such that it doesn't negatively affect your indoor air quality, that does not at all suggest that your indoor air quality will be good.
I know, and I know it's bad in there, but knowing precisely how bad is not useful information. A week away and there's a film of dust on everything. Most places, the dust is human skin, but not here. I think it's the old veneer covering the cinderblock walls and the cinderblock. Place was built in the 1940's, and it is slowly deteriorating particle by particle. But the rent is cheap. I have a HEPA filter on a box fan for the bedroom that runs constantly, which I hope helps, and I spend my days outside.
I have a 1940s place w/o cinderblock that does this heavy dust thing too. I'm also not fantastic about making it sparkle so as a result I do get dust bunnies. I've examined a few up close, and they look like laundry lint, a generally uniform light colour and texture. It's weird, I installed the exhaust duct myself, it's sealed and directly vented to the outdoors, my furnace fan runs every 20 minutes or so and has a merv 11 filter on it I believe, so there is some minor filtration. And yet, all this dust..
My place is not at all airtight, but I am skeptical that the outdoors air has this degree of lint fibers floating around at all times.
I get what you describe, grey dust bunnies, quickly accumulating within a week after cleaning the floor under my bed. Because it is only around my bed that I see those, I assumed it was coming from my bed clothes, the blankets in particular. idk what the material is called, but generally when new it's described as "scratchy" by children, probably wool or polyester. Shaking these outside produces a lot of junk flying off. But the dust that I see accumulate elsewhere and floating through the air turns the off-white base-boards a sickly dark yellow, and the dust is black on a paper towel when wiped. Maybe wash your bed clothes and rinse thoroughly, and skip the dryer sheet. I've found too much detergent drys or sticks on clothes and produces a puff of particles when I take off my socks too quickly, in particular.
AirSensEUR is an open framework focused on air quality monitoring using low cost sensors. The project started on 2014 from a group of passionate researchers and engineers. The framework is composed by a set of electronic boards, firmware, and several software applications.
I don’t currently have air quality measurements, but I do have Aqara (not Wifi but Zigbee) temperature and humidity sensors. Zigbee uses less energy (afaik?) than WiFi or BT, and additionally, while there is no arrow, the symbol shows you which way is up, and it fits with the description (ventilation holes at the top).
From what I’m reading, the fact that several sensors measure almost exactly the same values when mounted next to each other, is actually a good sign, correct?
I have Aqara WSDCGQ11LM sensors which only have small holes for the sensor at the bottom but no ventilation holes for the main part of the board.
At the bottom of the product page [1] they have an explosion graphic rendering which looks like (assuming it is somewhat correct) the sensor is below other chips and moved away from them. The sensor also seems to be shielded by having a separate area of it's own in the plastic enclosure.
I haven't tested them yet but I think they should do fine, by the criteria of the article.
It's not just air quality sensors. Got some wifi thermostats some years back. Oops--they draw too much power. At least it's consistent, we set them two degrees above what we actually want and and they work fine.
And we also have an example of the problem of air circulation. We have a wireless thermometer (a quick check if it's cool enough to open the windows) that's in a weather enclosure that admits a decent amount of air and sitting under a tree (keeping it out of the sun and away from big masses of concrete--it's simply impossible to properly place it in our yard) and it can be up to two degrees off because of lag.
> Monitoring my home's air quality (CO2, PM2.5, Temp/Humidity) with AirGradient's DIY sensor September 8, 2021
> A few months ago, I found this Hacker News post about the AirGradient DIY Air Quality Monitor. I had already been considering buying an AirThings Wave Plus sensor to monitor my home's CO2 levels, but the high price and limited 'ownership' of the data coming from it turned me off.
also
> AirGradient: DIY Air Quality Monitoring with Open-Source Hardware & Firmware
> I once saw a sensor chip that claimed to measure both CO2 and TVOC (total volatile organic compounds). After finding some further documents on the part I discovered it was a total scam. It only actually measured one of those things, then it used this measurement to look up an expected value of the other in a table the manufacturer made based off “common indoor measurements”.
I remember similar thing with some UVC sensors, they just returned expected UVC based on UVA level
I purchased a 'Yvelines' brand air quality monitor. The company appears to be one of those random word Chinese generic product companies. Some of the readings spike when I take the unit into the garage where we park the cars, keep the lawnmower, gas can, etc. Most go down if I take the unit outside. I don't know how accurate the numbers actually are but it certainly can distinguish between areas with different levels of pollutants.
I own an IQAir AirVisual Pro, it has identified when the air from ovens affect air quality. And, I bought a Honeywell HPA300 True HEPA Air Purifier and you can see air quality improve significantly when it is on. Also, when a large group of guests come over the CO2 monitor accurately tracks the large increase. And, it detected when Elon's factories had their fires. In addition, anytime there were forest fires in California it detected the increased pollution from those as well. The air quality several states away is affected by those fires severely btw, pretty interesting. The smoke plumes travel all the way to Maine. I've been pretty happy with both products. When the air quality drops outside, I close the doors and windows and let the air purifier do its magic. I leave the HPA300 on "Germ" mode, it is pretty quiet and still gets the job done.
What a useless article. How about some evidence and pointers to things that aren't?
Humidity/pressure/temperature are actually the three measurements that probably aren't a scam. The sensor chips you can get are generally well-calibrated and do what they say. Do read the datasheet, though: there are lots of ways you can screw them up (soldering them wrong, for example).
The CO2, VOC and particulate sensors seem to be absolute snake oil, though. I have yet to buy any device that "senses" these that has any correlation to anything resembling reality. Which seems kind of stupid because CO2 and particulate sensors shouldn't be that difficult. Volatile organics, though, tend to be difficult as they have a lot of confounding factors.
I'd really love to have a recommendation for something that's been validated that isn't $10K+.
I agree... I have a bunch of cheap indoor air quality measuring devices, and they all consistently read temp, pressure, and humidity close together. My CO2 meter measures exactly what I'd expect -high levels with windows closed, correct outdoor levels with windows open. My cheap PM2.5 air quality meter reads almost exactly what I see in purple air when I take it outside. Basically, I purchased a bunch of diverse cheap air quality sensor products that I have verified all read correctly, and are not scams. I don't think this article is accurate.
Granted, they're not meant for standalone use but at least you're not wondering whether the vendor is reputable or whether the sensor does what it says it does. For example, at a casual glance, the Siemens QPA2002D VOC and CO2 sensor seems to be available for <$500 and comes with a fairly comprehensive datasheet.
Seems like the best solution would be a two part enclosure... one for the sensor, with just enough to power it (e.g. a coin battery) and a Bluetooth LE transmitter (Or maybe even something entirely passive like RFID could work?) and then have all the rest of the electronics sealed off and powered seperately.
Sadly you can't skip to engineering a solution. If you're designing a device getting quality requires rigorous specification and testing, in a feedback loop with design. What you describe might be a solution, but you can't know it's worked without that testing, and it's premature/risks over-engineering unless the testing shows you need it.
It's the time and money of testing that is usually being skipped, not the engineering itself. (Although many manufacturers might sacrifice quality to save the cost of what you describe anyway :( )
Consumer Reports meets iFixit meets Twitch hybrid would be awesome. Plus the objective repeatability of Jepsen and TechEmpower.
Some magical combination of authenticity, expertise, tenacity and personal narrative.
As a noob DIY remodeler, I learned SO MUCH from channels like Project Farm, Matt Risling's The Build Show, Vancouver Carpenter, and others.
For whatever reason, I didn't get nearly as much from content tailored to prior mediums, like This Old House and misc handy person magazines. (Probably some McLuhanian "medium is the message" insight applies...)
Intentionally manufacturing something with tolerances wide enough to be nearly useless and then selling it as if it was useful sounds malicious enough to me!
Maybe so, but for the consumer is there any difference?
If you say your product does a thing, and it definitely does NOT do that thing (but you fully believe it does that thing perhaps on the basis of believing really really really hard), and you sell it on the basis of that belief, especially when said belief is easily disproven with a tiny bit of due diligence…
under those conditions I think it’s perfectly reasonable to call it a scam. Especially colloquially, when intent is near impossible to know, and the average consumer does not have the means to launch an investigation to prove intent.
The "NOT do that thing" is the questionable part, lots of devices that measure things can claim to do so when they aren't "accurate". The difference is how inaccurate is the device and is the margin of error something that should be communicated.
The classic example are scales, you can buy scales and weigh yourself on them and they might not be 100% accurate. But you can still sell a product that claims you can weigh yourself. Should we require standards that require companies to provide a +/- variance against exact measurements? Some products do such as power meters.
If a product is outside of a particular variance then could we call those scam products?
I recently bought a sensor that detects temp, humidity, PM2.5, PM10, and CO2, and the variance from the equivalent report from the airport's weather station that's only a few minutes away is discouraging even when it's set outdoors. Maybe I should bring it into my alma mater and borrow some lab equipment and test it out.
At least the CO2 says 410 when set outdoors, but the temperature it varies by a few degrees (Celcius). The humidity is even worse, I don't even bother looking at that number anymore.
The sensors I bought from Microcenter were garbage in the ways you describe. I couldn’t even take readings from some sensors because they were so ridiculous.
I see "wi fi connected" seems like a default choice now, not even questioning. I had one such CO2 detector, but then I realized that anyone gaining access to it (any employee of the manufacturer, for example) can know exactly when we are at home, when we are not, when we go to bed, when we go on vacation etc.
So I disconnected it, bought a $70 device that measures everything pretty accurately and am happy ever since.
It's called Dienmern model DM72B, from Aliexpress, I compared its temperature, humidity and CO2 measurements with other devices and found them correct; as for PM2.5, HCHO and TVOC, I didn't calibrate, but they seem reasonable and detect smoke/farts immediately.
I have seen some comments about air quality and particle counts here that I simply did not understand..
I live in a remote spot in the country in Ireland. I am surrounded by dairy farms (fields of grass with cows) I always have my windows open, all day all year round, though my room is super messy. People who know about this stuff, do you reckon I have good air quality?
Depends on how you/your neighbors produce heat if they are anywhere near. Villages' air can be pretty disgusting especially in colder months. Paradoxically, small cities when heated by remote heat can be quite enjoyable air quality wise all year round if you don't live near the main road.
PM2.5 have decent accuracy and are certainly useful at low price points, RH/temp work fine, particularly with separate sensors. Pressure works fine but is pretty useless. NDIR based CO2 (ex. Sensirion SCD31 or 41) also work fine.
TOC is a bit of a mystery, it can probably detect relative change. I've used them in a lab to monitor solvent vapor and you can detect changes, hard to quantify but don't need to most of the time.
I guess, yes and no. Technically the scd41 emits IR, it just doesn't detect transmitted light, instead sound produced by the CO2 that absorbed the light.
i have a couple of no-name air quality sensors and they react to the environment fine, but they read high (like 500) CO2 outdoors. I got them in case CO2 was the reason for my spouse's insomnia. I've tested them by putting them in a large bowl with dry ice, and it certainly reacts to that. Also cooking makes the AQI number change dramatically. right now the powered one says my AQI is 2, CO2 is 515, CO is 9, HCHO is 18, and TVOC is 0.2. Neither of mine mention temp or humidity, and i've never taken them apart.
I certainly won't rely on them to tell me if i actually have TVOC or HCHO issues. And i wish they had SD card slots that stored the values, like my 8 remote temp/humidity sensors and their base station.
I bought an aranet4 which is excellent for co2 but it doesn't do pm or vo2. I can however infer that if the co2 is high I need to ventilate anyways. Crazy to see how high co2 rises on flights that are supposedly "refreshing air" often.
I bought ~5 brands of air temperature sensor, and was surprised to find that when I left them in a sealed box overnight, every single one showed the same temperature to 0.1C.
I was expecting them to have at least 10x that error!
I got an Aprotii "7-in-1" sensor from Amazon UK for about €120 ("currently unavailable"). I haven't seen the same device from other vendors, which is unusual for this type of device on Amazon. Claims to be NDIR, can't seem to open it without trashing the glued down screen cover, so I haven't. I'm no expert on this type of device, I just figured anything less that €50 was going to be pointless.
Lots of ventilation holes on 5 sides, a screen you can read from 8m away. Agrees (within tolerance) of a nearby temp/humid sensor. CO2 and VOC are independent. CO2 alone goes up during the day if I don't crack a window, VOC varies with how hot and close beverages are. PMs/VOC are mostly a function of how badly my cooking went in the adjacent kitchen :)
I can't be certain about absolute accuracy of values, temp/humid are good, CO2 seems plausible: low is 400ppm, I start to get cranky while working by about 1200ppm.
It's wifi (Tuya) if you want, the screen and controls alone are fine though. It also has RS485 and a bunch of relays to drive A/C, Fan, (de)humdifier too.
There's a Spanish report that links Kkmoon/Kecheer/Curconsa/Aprotii device vendors:
F. Villanueva, E. Jiménez, J. M. Felisi, T. Garrido, J. L. Jiménez, M. Ródenas, and A. Muñoz. Guide about affordable CO2 detectors for COVID-19 Prevention, versión 1.4, 17/04/2021. https://bit.ly/monitorsCO2
I wonder when will personal CO2 scrubbers become a norm. With atmospheric CO2 growing ever higher and everyone having a CO2 detector (and an air "purifier") at home it's the next step.
Plants scrubbing co2 indoors are pure placebo. You would need to convert half of your apartament with a tropical jungle to offset what you breathe out.
I agree with you but for most things it honestly doesn't matter. The time you wasted researching is worth more than whatever difference there might be. I also think we don't need any of this junk in the first place.
Economics is based on trade-offs, and a trade-off of "opening windows" can be "lowering indoor temperature". Incidentally: some parts of the world are in an energy crisis.
Edit: also, that «whatever difference» clashes pretty heavily with
> It is hilariously easy to inadvertently design a device that measures temperature with an error of 5degC, humidity error of 50% and pressure error of equivalent to 1km in altitude
It is also difficult to avoid noting that we "look into things" for many reasons and also indirect gains; that there exist "silent killers" and, more importantly to some, silent harmers; that some "principle of inquiry" (of "knowing the world", of testing, like Galileo's "Saggiatore") and a "principle of wanting to get things done properly" are there...
I remember looking around for a Radon sensor for way too long. An electronic device, a litmus-like strip test, some analog crystal ball -- hell a mail-in lab kit bag of carbon -- but they all seemed inaccurate. The few people that "bought two" always showed heavy variance while the majority that bought and got a number *loved it*. Even buying two that agree often just means the products are consistently and predictably wrong, but most products don't even check this box ffs.
Reviewers that try and apples-to-apples all products by comparing cost, size, advertised features and ergonomics are doing no better than you or I opening up a package and appraising it. For awhile this Google dev was putting up extremely detailed amazon reviews for USBC cables based on their purported features v. reality and it was wonderful, but tucked away and undiscoverable.
The world needs multiple parties providing unbiased in-depth reviews, but there's very little will to pay for even one group to do something like this. Ironically, paying for information in the information age seems almost comical to today's consumers and while there's demand for good information, no one's willing to open their wallets. Labdoor moved from subscription-only memberships to "certified products" and affiliate links because everyone wants to know what's in their multivitamin, but no one wants to pay for an indepth review. We aren't cultivating a supply-side that works for us, but against us.
The result of this is shilling and fake reviews running wild. I mean hell, consumers will buy a product knowing full well they might get a shoddy knock off and yet it's $5 less and they can get it tomorrow so they'll roll the dice. Sadly, we don't live in a world where people are intimately concerned with quality and accuracy in their purchases.
edit: USBC not USB3