Bosch Sensortec Community

    cancel
    Showing results for 
    Search instead for 
    Did you mean: 

    BME680 + BSEC hugely inconsistent measurements

    BME680 + BSEC hugely inconsistent measurements

    Richard_R
    New Poster

    Hello all,

    We're working on a project where we want to measure some volatile substances such as ethanol and acetone in ranges up to a couple of hundred ppm under indoor, reasonably predictable conditions (no big changes in temperature or humidity).

    Based on the datasheet, we selected the BME680 + BSEC software as a likely candidate. We built half a dozen sensor PCB's with the actual BME680 devices on tiny separate PCB's, connected through 200 mm FFC cables for easy experimenting, i.e. dunking it in bottles containing reference gas dilutions.

    The BSEC software appears to work; when we expose sensors to clean air for several hours on end in continuous mode, we get VOC readings between 0.7 and 1.4 ppm, and accuracy slowly increases from L(ow) to M(edium) to H(igh). We observed that this progression can be speeded up significantly by temporarily exposing the sensor to a slight increase in volatile substances, to a VOC reading of 20-30 ppm or thereabouts.

    When the accuracy reaches the High level, we use save_state to save a snapshot of the calibration state, enabling us to shorten the otherwise lengthy calibration process upon next power-on. Then we power down the device, and power it up again, using load_state to restore the previously saved calibration state. Measurements still show Unstable for 5 minutes, but after that, the accuracy almost immediately shows the High level. We let 3 devices sit overnight, still working in continuous mode, for purposes of burn-in and (hopefully) more accurate results.

    The next day, departing from this calibration state, we tried to get coherent gas measurements of known gas concentrations, using 100 ppm acetone vapor in a closed bottle as a reference gas. Unfortunately, this does not produce any coherent results across different sensor devices. Device 1 may produce a VOC reading of 4.13 ppm, Device 2 shows something in the order of 30 PPM, and device 3 shoots up all the way to a reading of 1000 ppm (the maximum value).

    Results are consistent and repeatable per sensor, so when these measurements are repeated, we get results close to the orignal values of 4, 30, and 1000 ppm, respectively. The problem is that we can't get two individual sensors producing values even remotely alike. Now we don't need single-number-percentage accuracy and matching; something like 20% or 30% deviation between measured values is still acceptable. But being off by one or two orders of magnitude is of course no good at all.

    All three devices have gone through the exact same self-calibration process in the exact same environment, and their sensors are all placed in the exact same vapor dilution at the same time.

    Does anyone have a clue as to what we may be doing wrong?

    Thanks in advance for any thoughts!

    Richard

    4 REPLIES 4

    BSTRobin
    Community Moderator
    Community Moderator

    Hi huytyskland,

    For your test "The next day, departing from this calibration state, we tried to get coherent gas measurements of known gas concentrations, using 100 ppm acetone vapor in a closed bottle as a reference gas. Unfortunately, this does not produce any coherent results across different sensor devices. Device 1 may produce a VOC reading of 4.13 ppm, Device 2 shows something in the order of 30 PPM, and device 3 shoots up all the way to a reading of 1000 ppm (the maximum value).", how about the accuracy value duaring that time?

    [snip strongly diverging measurements between separate devices]
    @BSTRobin wrote:

    Hi huytyskland,

    how about the accuracy value during that time?


    First of all thank you for your quick reply!

    With rising VOC reading the accuracy may drop to Medium for relatively short periods of time. However, it never reverts to Low or Unstable.

    At the moment, I have two devices running side by side (the nrs. 1 and 3 above), and after giving them another night, readings seem rather better (accuracy given between brackets [H/M/L/U])

    Baseline, clean air:

    #1: VOC: 0.61 ppm [H], IAQ 38.35 [H]
    #2: VOC: 0.69 ppm [H], IAQ 39.41 [H]

    -> this seems pretty good.

    Exposure to 100 ppm acetone, after 20 seconds:

    #1: VOC: 10.61 ppm [H], IAQ 219.16 [H]
    #2: VOC: 299.61 ppm [M], IAQ 282.22 [M]

    -> device 2 shows a reading that is 30 times the value of device 1, with Medium accuracy.The switch from [H] to [M] happens somewhere around a reading of 130 ppm. Another noteworthy detail is the fact that IAQ values are far less divergent.

    Then we tested a dilution of ~30 ppm acetone, after first returning to baseline levels:

    #1: VOC: 4.58 ppm [H], IAQ 175.64 [H]
    #2: VOC: 62.11 ppm [H], IAQ 216.88 [H]

    -> both devices now report High accuracy, yet there is still a discrepancy by a factor of 10 to 15 - way too much to be of any use.

    After removing the sensors from the acetone vapor, readings revert to slightly above baseline levels:

    #1: VOC: 0.90 ppm [H], IAQ 63.75 [H]
    #2: VOC: 1.00 ppm [H], IAQ 52.93 [H]

    The overall values are slightly higher than baseline probably because some acetone is absorbed in the materials and slowly released.

    We also have some questions that you may be able to shed some light on:

    • What is the best calibration procedure for these devices?
    • Are there ways to make these devices 'learn' about gas levels apart from the current BSEC automatic calibration?
    • Can we somehow prevent measurements from reverting from [H] to [M] or lower?

    At the moment, I'm working on a dual-channel logging application for logging most internal parameters second by second, so that responses of the two sensors can be examined and compared with the highest possible resolution. If you think that this information could be useful, I will post (a link to) the log files here as soon as I have this finished.

    I also experimented with 'dumb' hot-plate sensors such as the (now obsolete) CCS801, so I fully appreciate the huge difficulties in calibrating and using these devices; still, we had hoped that these Bosch sensors would be a bit easier to deploy (also based on accuracy figures given in the datasheet). As you can see above, we have not yet succeeded in monitoring vapor concentrations with any accuracy at all; even 'High accuracy' measurements can differ by well over an order of magnitude.

    Anyway, thanks already for any suggestions on how to improve performance here, because at the moment, things are not usable. And if it turns out that these sensors are not supposed to measure ppm's with any accuracy at all, then please let us know, so we can go look for other devices.

    Best regards,

    Richard

    I hate to nudge up this thread by asking for replies, but we are still not making progress on the problems with calibration and highly inconsistent measurements between individual BME680 sensors.

    Could someone point us to a calibration procedure using BSEC that has been found more reliable that what we observe? Our current strongly divergent results suggest that we must be doing something wrong, but we have no idea what that might be.

    Thanks once again,

    BSTRobin
    Community Moderator
    Community Moderator

    Hi Richard_R,

    Unfortunately, BME688 do not support measuring accurate gas ppm.

    Icon--AD-black-48x48Icon--address-consumer-data-black-48x48Icon--appointment-black-48x48Icon--back-left-black-48x48Icon--calendar-black-48x48Icon--center-alignedIcon--Checkbox-checkIcon--clock-black-48x48Icon--close-black-48x48Icon--compare-black-48x48Icon--confirmation-black-48x48Icon--dealer-details-black-48x48Icon--delete-black-48x48Icon--delivery-black-48x48Icon--down-black-48x48Icon--download-black-48x48Ic-OverlayAlertIcon--externallink-black-48x48Icon-Filledforward-right_adjustedIcon--grid-view-black-48x48IC_gd_Check-Circle170821_Icons_Community170823_Bosch_Icons170823_Bosch_Icons170821_Icons_CommunityIC-logout170821_Icons_Community170825_Bosch_Icons170821_Icons_CommunityIC-shopping-cart2170821_Icons_CommunityIC-upIC_UserIcon--imageIcon--info-i-black-48x48Icon--left-alignedIcon--Less-minimize-black-48x48Icon-FilledIcon--List-Check-grennIcon--List-Check-blackIcon--List-Cross-blackIcon--list-view-mobile-black-48x48Icon--list-view-black-48x48Icon--More-Maximize-black-48x48Icon--my-product-black-48x48Icon--newsletter-black-48x48Icon--payment-black-48x48Icon--print-black-48x48Icon--promotion-black-48x48Icon--registration-black-48x48Icon--Reset-black-48x48Icon--right-alignedshare-circle1Icon--share-black-48x48Icon--shopping-bag-black-48x48Icon-shopping-cartIcon--start-play-black-48x48Icon--store-locator-black-48x48Ic-OverlayAlertIcon--summary-black-48x48tumblrIcon-FilledvineIc-OverlayAlertwhishlist