Forum Replies Created
-
AuthorPosts
-
Ken JamrogowiczParticipant
I have not replaced the sensor all this time and, old as it is, seems to be working. My conclusion is there must be some source out there that is producing intermittent bursts of VOC’s. Also, I have recently seen that, by refreshing the browser page, the spikes go away.
You know that competing sensors (of which there are several) all use 10-minute averages for the default plots. This would flatten out spikes and would greatly improve the appearance of the PM graphs. This would be done at the server end of things I suppose.
This summer, the formaldehyde sensor is – for the first time ever – starting to produce some readings. I wonder if you have changed some software? or if my air is dirtier this year than ever before?
The VOC and formaldehyde readings tend to track each other (mirror image) indicating a common gas is being detected. Probably not actually formaldehyde. Both sensors respond to a a variety of gasses.
Attachments:
Ken JamrogowiczParticipantI have been watching the O3 with the new power supply and it’s quite the same. It is just that – during the daytime – there is a lot of activity that covers up the tendency to “stick” around 120 ppb. Since this sensor is designed to measure PPM, it is probably unrealistic to ask it for readings in PPB range. The resolution is 10 ppb and I think the zero point could easily drift by 100 PPB.
I would be nice if there was a way to set the zero calibration in software. In my case, zero gas is giving a reading of around 100 PPB (and has always been so).
If replacing the sensor with another one from China would be certain to fix the problem, I would be willing to give that a try.
Ken
Ken JamrogowiczParticipantSorry I should say rarely below 50 – which is the “green” zone. It has always been this way.
Ken
Ken JamrogowiczParticipantThe unit is fed from a 24 volt supply and POE injector/splitter.
Actual voltage arriving at uRad is about 20 volts when using the 24 volt supply.The center portion of this graph is with the 18-volt battery supply (3×6.3).
The right hand side is back on the 24 volt supply with some added filtering – I can do more.
I can also try lowering the voltage, as I found everything else on the system works at 18 volts.
You can see the readings average about the same values with either supply.I an quite sure the O3 sensor is responding to other compounds in the air. I found it is quite sensitive to alcohol and the forest here emits natural compounds that are chemically similar to alcohol. The second photo shows the month of September. Readings are rarely below 100 ppb and tend to show a peak right after it rains, when soil bacteria naturally emit a compound that the O3 sensor seems to detect quite well.
Ken
Attachments:
Ken JamrogowiczParticipantOK – I tried running the unit on batteries and that problem went away. I will need to filter my DC supply better! I am surprised that the internal voltage regulator in the unit does not remove this AC ripple.
Kind of a different question: why does this unit show ozone levels in the warning zone when it is in some really fresh air – near a forest and away from major highways, etc.?
Ken
Ken JamrogowiczParticipantYou are missing the point. The reason the graph looks like that is because the unit is producing readings of 30Meg to 50Meg for short periods of time – perhaps only 1 or 2 samples. Historically, the “sensible readings” have been in the range of 300K. The ratio of 30M to 300K is 100:1. According the specs the Winsen MP503, the range from quite polluted to quite clean is on the order of 10:1. So such big readings cannot be coming from a healthy sensor.
see the graph within this document https://www.winsen-sensor.com/d/files/ZP07-MP503.pdf
This sensor gives lower readings with more VOC. These >10Meg “spikes” would therefor represent sudden spikes of amazingly clean air amongst a background of almost continuous high pollution. This is nonsensical.
These very big readings of >>10M have only appeared in recent months (coinciding with colder weather) that is why I suspect it is the start of a failure. I see that the device has a small built-in heater, perhaps that is what is failing. I suppose time will tell.
Ken
Ken JamrogowiczParticipantPlease note the VOC sensor operates in reverse. Detected gas gives the low reading. Clean Air gives a high reading. When the unit was new, it looked like this curve upside down.
Ken
Ken JamrogowiczParticipantYes.
But your picture is out of date. The sensor you can buy today does not fit. And whoever assembled my unit originally simply jammed the sensor on top of the board.
Modifications I show in my photos will allow using the new sensor without forcing it.
But, basically, you have a problem with the current production run.
Ken
Ken JamrogowiczParticipantOK – answering my own question. The B unit has a bigger fan than the OEM “A” unit. The attached photo shows the old one on the left and the new one (already connected) on the right.
As it was, the OEM did not fit very well either. It was *very* difficult to get it out of the case.
To get the new one in I decided to cut off piece of the circuit board, remove two of the screws holding the fan on and file down one corner of the PM unit.
The result is shown in the last photo where some white space is left so you can see how it fits. It is quite impossible to mount the PM unit on top of the PCB, as both the -A and B versions are the same height as the enclosures. I have no idea how uRad was able to get the PM unit in there originally.
It has been running for a bit over ah hour now and looks like it will be OK.
Regards
KenAttachments:
Ken JamrogowiczParticipantThat is another possibility (need to make sure they are not on the same ISP). But before doing that I am waiting for the replacement PM sensor to arrive here – seems to be taking a long time (?)
Ken
Ken JamrogowiczParticipantWhen I take uRad down to replace the PM sensor, I think I can arrange a different connection. It will use a point-to-point wireless link to a different service provider, with the connection point about 50 km from here. So it will really be different. We’ll see.
BTW: today my service provider upgraded my connection to their “gigabit” level (actually 944/800 Mbps). But, no help for the rebooting. All they really changed was the modem connection on my end (and my monthly bill!).
Ken
August 24, 2020 at 9:04 pm in reply to: A3 Ethernet can’t establish link with enterprise class network switch #28509Ken JamrogowiczParticipantQuite some time ago I changed my router to supply a static connection for the uRad monitor, but it obviously has not helped my particular unit.
= =
The problem at the top of this chain of notes was one of the unit having trouble recognizing that it had a valid PHY connection. So – quite a bit before the issue of DHCP comes up. I actually ran into this sort of problem with an earthquake sensor that I run here. Until I added the extra length of cable, there was all kind of noise on the earthquake signal. This only happened when it was connected to a gigabit switch rated for 100 meters of cable …August 1, 2020 at 10:41 pm in reply to: A3 Ethernet can’t establish link with enterprise class network switch #28478Ken JamrogowiczParticipantI would like to explain my suggestion to add a length of cable between the A3 and the “enterprise class” switch. I have observed that some of the commercial switches that are designed to comfortably drive 1000M through 100m of cable are producing a rather powerful transmit signal. If you connect that to an Ethernet interface that is designed for “hobby” type application, the little transformers in the adapter can get saturated and make garbage out of the signal that the micro sees. The conclusion would be that this is not a valid network. It might barely work, if you force the connection to 10M because then you have some nanoseconds for the transformers to come out of saturation. But not for all data strings. Inserting a length of lossy cable between the switch and the A3 would ameliorate this particular problem (if it is the problem).
Ken
Ken JamrogowiczParticipantWell – my power supply feeds two other units at the same location and they do not have problems (it is a battery-backed supply). As posted elsewhere, what is happening is that the A3 simply stops communicating over the ethernet port – even to local hosts – so nothing to do with my router. That ethernet failure apparently is in both directions so, not hearing anything coming in either, the program reboots. I have my fingers crossed now because it has run up to 410K seconds which is some kind of record for my unit….
But the topic is CO2 calibration. Update there: after about 2 weeks the A3 spontaneously decided it was time to reset the calibration so that the minimum values are about 400 PPM. This did not coincide with a reboot. I have read in another post where someone else’s unit did the something similar. I posted the month of CO2 readings so you can see what it did. The two upward steps were when I adjusted the height of the unit. No other sensor jumps coincided with the reset to 400. It was at 9:22 AM July 26. I will be watching on August 26th to see if anything interesting happens 🙂
Attachments:
Ken JamrogowiczParticipantOk – I have been able to do some analysis of the rebooting shown above. I am running a small script that does a “curl” to the jason page on the local LAN and time-stamps it. It’s clear from this that the A3 is simply stopping to send data when requested. (see attached text file). Sometimes it stumbles and restarts. Other times it simply stops until a timeout causes a restart. The timing and duration seem to be random but one thing is certain – this has nothing to do with my router or ISP. The A3 has got an internal problem of some sort.
Ken
Attachments:
-
AuthorPosts