Strange TVOC measurements

Components
#1

I’m getting strange TVOC measurements with my airgradient indoor monitor (latest model).

To test the TVOC sensor, I first let it run for a week under normal conditions. I got readings that fluctuated between 50 and 150, which seemed correct (given 100 is the baseline value).

Today, I opened a bottle of nail polish near the sensor, and it after a short delay (about a minute) it spiked to 500. After that, I closed the bottle and moved it to a separate room. The sensor read 500 for about 20 minutes, then slowly decreased back to 150 over the course of 7 hours. This is in a reasonably well-ventilated room, with central A/C running. Opening a window nearby did not seem to increase the speed at which the measurement fell.

I tried again about 30 minutes ago with a bottle of acetone (nail polish remover) and got the same result. It’s still at 499.

This seems like erroneous behavior to me. In each case, the concentration of nail polish/acetone in air dropped rapidly after I closed the bottle. I know this because I could no longer smell it, which means the concentration in air couldn’t be more than about 10 ppm (according to Google). It doesn’t seem correct that it would take over 7 hours for that small amount to fully dissipate, especially with a window open for some part of that time. (In contrast, the PM2.5 readings work basically as I expect, and I can get rapid drops by introducing enough ventilation.)

Is this just a quirk of the sensor? I’ve noticed that cheap PM2.5 sensors display a similar “saturation” effect, where they essentially stop functioning properly after being exposed to high (200+) PM2.5 levels. Is a similar things happening here?

Any insight would be appreciated. I would also be curious if others can replicate this behavior.

#2

Can you post the chart? It would be interesting to see the decay curve.

#3

See below. The “sawtooth” pattern overlaid on the long-term trend has always been there in previous measurements and does not surprise me. What does surprise me is the slow decay back to 100, especially when the smell disappeared shortly after the peak.

tvoc
tvoc3082×1656 254 KB

#4

I have seen similar when using alcohol inks and the decay taking much longer than expected. I accounted it more to my nose becoming acclimated to the odor and not that it actually had fully disappated in the room. But in my case, I couldnt set up good ventalatuon.

#5

In my case I left the building, went outside a while, and returned to find no odor but still a high reading. Others who entered later also did not smell anything.

Another possible explanation is that the sensor was completely accurate, and the baseline levels were so low that it was really possible to be at 5x+ the index value while remaining below the human odor detection threshold. But I’m not sure the sensor is that sensitive.

#6

How about you repeat the experiment differently.

  • Put it on a power bank.
  • Wait for the monitor to be stable in your room.
  • Take the monitor outside of to another room (far away from the other room).
  • Spike the sensor.
  • Bring it back to the original room.
#7

I did this experiment but got odd results. Does the TVOC calibration restart every time the unit powers off and on? I think so, which would explain what I saw.

I put the monitor on a mobile power source and waited 5 minutes for the TVOC readings to stabilize at 100. Then I went to a small bathroom and opened some nail polish. I couldn’t get the reading to go above 450, so then I tried acetone. It still wouldn’t go above 400 after about 10-15 minutes of this, even after I could clearly smell these substances in the room, so I capped the bottles and moved the monitor to a different room. Then it immediately read 25, then 32 for a while, and now it’s at 80 and slowly creeping back to 100.

My preliminary guess is that the sensor is functioning correctly and my baseline levels in the first experiment were just really low. The drop from ~300 to 15 suggests the levels in the bathroom were about 20x the baseline building level. The human odor threshold for acetate is about 1 to 3 ppm according to Google, while the sensor is good down to about 0.05 ppm ethanol equivalents, so the numbers seem to roughly check out. ( Other sources say 20 to 400 ppm to smell acetone. In any case, there’s a lot of room below this threshold, so the same argument applies.)

A small lingering amount of acetate in the air after the initial experiments could have caused the high readings even if it quickly dissipated below the odor threshold, if the baseline was truly very low. This is my guess for what happened. The decay curves also roughly match the exponential diffusion curves I’d expect on first principles.

#8

Yes, that is how this sensor works. It takes a baseline on boot, which sets the VOC Index to 100 and then it goes up or down based on air improving or not