I know what you mean…
This appears to be suitable for those who want a real CO detector
aliexpress.com/item/1005006255453561.html - right now it’s on promotion for less than $6 so I ordered one to test out
Did this work?
Does the EP1s addon meet these criteria?
Yes and no.
I was able to flash ESPhome on it, but right now there’s no deep_sleep support for Beken and the batteries are flattened in about 24 hours, so for all intents and purposes it’s useless unless converted to use a USB supply.
display accuracy seems to track locally purchased alarms but the holtek chipset’s output to the Tuya Wifi module is very “secret squirrel” so importing that to HA isn’t happening yet. (it contains ppm and temperature readings but appears to also contain a couple of other items). It DOES have GPIO feeds for both CO2 and low battery alarms so it can be used for that purpose…
HOWEVER: I’ll reiterate my earlier statement. These things MUST NOT be used as a substitute for a “proper” detectors (as in locally purchased and certified as calibrated by a vendor which faces liabilties if people get hurt). Carbon monoxide is lethal in one lungful at high concentrations and insidiously damaging in lower ones.
A friend of mine had a CO leak in her kitchen (water boiler) and it took me a couple of years to realise why I kept leaving their place with stinking headaches. The day after I gave the family a CO alarm it went off at 5am and the family came downstairs to find her husband trying to shut it up so he could finish his coffee before leaving for work. The chemical sensors I’d attached to the alarm as a backstop had gone completely black, indicating levels lethal in a few minutes of exposure. They’d been living with it for many years and CO poisoning really does cloud the thinking quite badly. Her kids suddenly started doing better at school after they stopped being subtley poisoned at home and the family’s number of illnesses & migraines simply plummeted
Interesting (and luckily happy ending) story…
in my country there were a couple of deaths in the past years because of CO poisoning (in wooden vacation cabins). It’s really sad how folks ignore the danger part of CO gas. It’s too expensive to buy a 30€ sensor when you build your wooden house with a fireplace, i guess 
That’s why i also wonder why there numerous discussions (and also sensors available) about CO2, but very few about CO, which is way more dangerous (and basically impossible to detect by a human) than CO2.
I have —uff-- 4 co sensors in my basement, where my central heating pellets stove is and also my workshop: one pretty new and one old (theoretically expired, but still kinda works) bought sensor and two homemade, connected to HA 
(call me paranoid…)
CO2 sensors are easier to supply as they are easier for the end user to calibrate. You need sample gas for CO calibration.
Correct.
I wonder, however, isn’t it ok for home use (or let’s say as “better than nothing”) to “calibrate” CO sensor with home air? Some guides suggest that you put CO sensor outside for a while and take that value as a reference. Ok, assuming that outside air is “healthy”, so calibrating in winter is definitely not suggested. That’s how i have my MQ7 is “calibrated”. Such a value is then more of a guide than precise value, but it will report CO increase, and that’s main goal → to warn, right?
(luckily) i can’t speak from experience, since in all these years co level in my basement never rised to alarm yet…my workshop is decently big and well ventilated.
No, that’s CO2, ~416ppm outside. CO would (should) be 0 outside. Not much use as a reference.
I have an Apollo Air-1 (edit: with MiCS-4514 gas sensor) which is useless for CO because it goes off when you use ammonia in its vincinity, use a can of spray paint, etc. So while it shows a level, you cannot bet your life on it. They are honest about this in the documentation.
I also own a Fibaro CO alarm, which I do rely on to keep me safe. It is Z-Wave so you can link it to HA. It also shows CO level in ppm. Next to it is a Honeywell, not smart, but while the battery still runs it is past expiration date. So I no longer trust that either anymore. That is why I bought the Fibaro. No false alarms yet on either of them. As for false negatives that is hard to tell. We all still live…
On esphome’s page for MQ7 there’s calibration procedure for CO:
Calibration
After long-time storage the manufacturer recommends running the sensor for 48-168h to let the readings stabilize.
Then it can be calibrated by placing it in clean air (outdoors) to determine the highest possible resistance. This resistance value can be added to the ESPHome config
So i guess after all this is some kind of a reference.
But, MQ7 is a bad sensor, i know. It’s more an indicator then a real co sensor…
i did have thoughts to buy MiCS 4514 Gas Sensor, which is supposedly better, but some guys had problems with it in esphome, so i’m not sure yet.
This is misleading at best. Many CO and other sensors do not actually measure absolute values, but register change. I do not know if that is the same for MQ7. But if so, then if the sensor changes it assumes something must be going on. If it remains high, the sensors will settle down and clear the alarm (while there still is a problem). Seen that way, placing it outside would allow to detect a difference from outside. But I would not call that calibration.
The same goes for setting a ground level. It does nothing to validate CO levels if there is no CO level known. While high CO levels are unlikely outside, the levels can vary, for instance caused by traffic.
True, but that’s the main goal after all, right? → to be warned when co value increases (above danger level). At the end i don’t care much about gas quantity in absolute numbers, but only safe - dangerous.
Totally true again. BUT…
you mentioned traffic: sure, co levels are higher there, but if someone lives in such an area there’s nothing to be done (except move) and again it’s not important if constant co level is, say, 67 or 555 or …(writing a blind value), but only if it’s “normal” (read dirty → not healthy long term) or dangerous (read fatal short term).
Also true. My point was just to say, calibration is something entirely different.
It only has a CO2 and VOC sensor, not CO?
No way Sen55 could detect CO.
This is certainly how the Tuya CO detector I’ve been testing seems to operate. After injecting CO into a sealed container (burning scrap of paper, nothing complicated), the displayed PPM level spikes, then declines to zero after a few minutes - and as there’s nowhere for the CO to go, this is clearly a false negative
As I keep emphasising, the best course for actual safety is to buy a locally certified unit. Toys like the one I’m testing can easily create a false sense of security
If it reports CO2 and CO, it usually has NEITHER sensor and is relying on calculations extrapolated from a VOC sensor
Thise calculations (eCO2 and eCO) aren’t about local ppm quantities but relate to global warming effects. Attempting to use them to report local conditions is worse than simply making figures up
This is why it’s so important to KNOW what the actual sensors are in your devices. Confidently reporting a lot of values means nothing if the input is garbage
I’m going to give you two scenarios:
1: You have a faulty gas heater which outputs a lot of CO in a short period of time. CO levels in the room spike to dangerous levels
2: You have something which emits much less CO over a longer period (a candle, or a cigarette smoker). CO levels in the room slowly rise to sub-lethal counts. If the CO buildup isn’t countered, room occupants will start suffering brain fog (cognitive impairment) and/or splitting headaches
Which do you want the alarm to sound on?
Which do you want it to display a ppm count for?
If the level of CO in the room remains steady and lower than lethal levels but high enough to cause health issues, do you want the ppm display to register it or say “zero” ?
That’s why the MQ7 and friends along with simple integrating comparator circuits aren’t suitable for safety-of-life applications - They can tell you if the level changes, but not (reliably) by how much, not an absolute level and they’re incredibly sensitive to changes in ambient temperature/humidity - which is why the integrating comparator circuits are necessary.
No matter what you “calibrate” the sensor to, things will drift by potentially massive amounts as ambient temperature/humidity conditions change.
That makes them suitable for stuff like go/no-go flue-leakage detectors in situations where the furnace is cycling on and off or for things like a parking garage where a particularly high output vehicle needs to be flagged as it enters but for long term or absolute level monitoring they’re worse than useless