Follow-up to alog88174 on the new brand of dust monitors. The colors are consistent throughout the plots, blue for the control GT521s, red for the PMS21, yellow for the PMD331 in differential, and green for the PMD331 in cumulative.
After some back and forth with the companies support, they ended up sending us 2 new devices, a PMD331 and a PMS21. I ran a few tests with the new devices, the PMD331 continues to look reasonable and similar to the GT521s. Although I have some confusion to counting methods, the company says setting the unit of CFM (cubic foot / min) sets the counting method to differential but I'm not convinced. You can't specify counting method on this devices' interface unlike the PMD331.
The first test I ran was in the front room (incorrectly labeled as MSR) with two MetOne GT521s, a PMD331, and a PMS21. The second was in the CER with both PM# and one MetOne GT521s. The TemTops had filter elements built in which the MetOnes do not, I removed these, cleaned the devices of any leftover particulate from them and ran a bunch of cycles to clear out any more of those lingering particles. The PMD331's sampling clunk which was a multi-layered mesh style, unlike the PMS21 and GT521s which is just a funnel essentially, which seemed to be increasing the counts so I swapped it out to use the same one as the GT521s. After doing these things I prepared for the test by running both the zero count and flow rate calibrations on all the devices setup side by side.
The PMS21 still seems to be showing lower counts but it's not nearly as drastic as it was before. For the first test, I applied a scale factor of 2 and 4 for the 0.3um and 0.5um particles respectively. The second test I ran which was over a much longer time span (1.5 hours vs ~20 hours) in which the PMS21 performed more as expected, the 0.3um counts were pretty much spot on with the GT521. For the larger 0.5ums the counting was low as with the last device, this time the scale factor I found to be best was 3, compared to the 40 I had with the last device. This is the model that they prefer to make a pumpless version of, so I'm curious about testing a pumpless version from them to eliminate that potential source of error.
The PMD331 still seem reasonable although I have questions about the internal counting methods. I assumed the standard counting method was cumulative so I did some bin subtraction to the counts columns to convert it to differential counting by subtracting the 0.7um from the 0.5um and the 0.5um from the 0.3ums. I assumed the raw data collection was in cumulative as if it were the opposite the counts would be far higher than the GT521. For the 0.3um particle size bin, differential counting more closely match the control but for the 0.5ums it's less clear. The control (blue) trace is in-between the two, it matches the cumulative trace a bit better.
I have also relayed this info to the support engineer I've been talking to with TemTop.
