Defender works... and doesn't work.

I'm trying to get a used (new to me last fall) Defender running. Every outward indication is that it's operating properly,
but it hasn't caught a single bug in days of operation.

The startup sequence appears to be normal: The light blinks, the fan runs (at various speeds), and after the warmup
period the light goes solid. I can feel (but haven't measured) slight warmth at the CO2 outlet, and the tissue test at the
vacuum inlet says that air is moving. As I mentioned in my previous post in the Pro thread, I suspected the old and
evidently badly-stored (box faded) octenol cartridges were NFG, so I just bought some new ones. Yet no skeeters in
the basket.

From my so-far cursory understanding of the controller in this thing, its only feedback is from the thermistor. But if the
controller is satisfied that it's running right, and the outward indications are good, what can be going on here?

I don't have a Defender but know the Liberty and Independence units fairly well and don't think they are that different in their fundamental operation. It sure seems like you have the bases covered. As you surmised, when the unit thinks it is in normal operating temperature (via the thermistor reading), it transitions to run mode, which is a green LED on my units, but a solid RED on the Defender. At that point, you should be producing CO2 and if you also have attractant and suction, you SHOULD be sucking in some pests. I don't know where you are but up here in New England, I haven't attracted that many mosquitoes yet this time of year but am seeing many more no-see-ums this year. If you are in the South, I believe you need Lurex as the attractant for your mosquitoes. Are you not seeing ANY pests in your net? Needless to say, check for holes or leaks where they might be escaping from the basket.



Edited 1 time(s). Last edit at 06/21/2021 07:45PM by Independence.

I'm in southern Alberta (Canada), so octenol should be the correct attractant, and the mosquitos are mos def out.
I'm testing this thing in my garage, and as I work in there they're drawn in large numbers to me and not the trap.

The net basket looks okay, so I guess the next thing to do is just to open it up and see what the air path looks like.

And just for the hell of it, since I'm in there, I was thinking of pulling the MCU off the board and checking to see if the
code protection bits were set when the part was programmed.

If there are no holes in the basket or suction path, I can't think of anything else. If there were obstructions in the propane ignition pathway and it's not mixing and burning properly, the catalytic converter won't get to temperature and the unit will error out. So it's strange that your unit is apparently reaching operating temperature and not capturing mosquitoes.

The PIC MCUs used definitely have their protection bits set. But the units are easy to reverse engineer and if you are so inclined, there are various projects here that cover it, including my Independence and Liberty versions, and Dev's NodeMCU version for the Defender and also beads Arduino version.

Let us know what you find if you solve it...



Edited 1 time(s). Last edit at 06/21/2021 08:11PM by Independence.

BTW, I guess Lurex3 was discontinued so the folks in the South have no viable attractant now. Also, I see that since MM acquired Dynatrap, they have adapted their ATRAKTA attractant for use in the Mosquito Magnet. Check out what they sell now:



The ingredients are still mostly Octenol and only lasts for 16 days:

Lactic Acid Component - 1.495 grams
Octenol Component – 1.145 grams
Ammonium Bicarbonate Component – 5 grams

Edit: After looking into it, it turns out that lactic acid on your skin attracts mosquitos, and Ammonium Bicarbonate attracts fruit flies, so this attractant may catch more bugs than plain octenol.

Independence Wrote:

> The PIC MCUs used definitely have their protection bits set.

Is that speculation/presumption, or has someone actually tried to read them?

> But the units are easy to reverse engineer and if you are so inclined, there
> are various projects here that cover it, including my Independence and
> Liberty versions, and Dev's NodeMCU version for the Defender and also
> beads Arduino version.

Yeah, I got that from the impressively-over-the-top controllers that you guys
have documented here. I've been doing embedded systems like this for 40+
years (including many AVR designs starting in the late 90s - long-predating
the ardguidos), so I get it. But through the years I've gained a greater
appreciation for not using computer control where it's not actually needed,
so I'm unlikely to join you on that path. Just want to get these things working...

debugger Wrote:
I'm testing this thing in my garage, and as I work in there they're drawn in large numbers to me and not the trap.

This is to be expected. Woodstream (once) explained that the mosquitos are drawn to the CO2 and attractant, and not finding anything to bite, rise upward until they are sucked into the trap. If you are nearby, they will bite you and ignore the trap. To catch mosquitos, deploy the unit to an area between their likely source and the area you want to protect.

... But through the years I've gained a greater appreciation for not using computer control where it's not actually needed, so I'm unlikely to join you on that path. Just want to get these things working

Right. However, in my case, the trap was old and failing in various different ways. Given the start up time is several minutes, and hanging around waiting for a start or failure resulted in many bites, I just wanted to add some instrumentation and remote monitoring so I wouldn't have to stand outside. Checking the trap for the mosquito count or tank weight, which are not instrumented, is done in the heat of the day, when there are few mosquitos around. At other times, there are typically several mosquitos hovering in the vicinity of the trap, and I give the traps a wide berth.



Edited 1 time(s). Last edit at 06/21/2021 10:17PM by dev.

That suggests that if I move it out to the door of the garage (this assumes they don't have a hidden breeding spot
and lair inside the garage, of course) it should intercept them before they get to me at the bench ~20' inside. Obviously
it was worse in the evening when it began to cool off. But I left it running for a couple of days, during most of which I
wasn't in there, and it still caught nothing.

debugger Wrote:
-------------------------------------------------------
> Is that speculation/presumption, or has someone actually tried to read them?

It's my presumption, but I'd put money on it. I haven't heard of engineers forgetting to set the bit since the early days...

> Just want to get these things working...

I think you'll find that we all started out that way but for various reasons (boards dying, inexplicable behaviour, etc), had to reverse engineer them and spin our own controllers. The main thing for me was having full control and monitoring over the functions, and adding additional functions such as wifi access, run time counter, programmable run times.

When everything is running fine, you don't even think about it. But when it is having problems, you start to wish you knew what the thermistor temperature is at, what state is it in, etc...

> It's my presumption, but I'd put money on it. I haven't heard of engineers forgetting to set the bit
> since the early days...

Sure, but sometimes they don't bother because if there's nothing particularly unique or proprietary
in the code, there isn't much point and they don't care. I mean, the operation has already been
fully enough described in the patent for anyone with a clue to go "Okay, so this MCU is implementing
not much beyond a couple of comparators, right?"

> I think you'll find that we all started out that way but for various reasons (boards dying, inexplicable
> behaviour, etc), had to reverse engineer them and spin our own controllers. The main thing for me
> was having full control and monitoring over the functions, and adding additional functions such as
> wifi access, run time counter, programmable run times.
>
> When everything is running fine, you don't even think about it. But when it is having problems, you
> start to wish you knew what the thermistor temperature is at, what state is it in, etc...

Of course. As I said, I get it. I'm just hoping not to have to go there, especially with an example
that appears to be running properly, but is just - mysteriously - failing to catch anything.

debugger Wrote:
-------------------------------------------------------

> Of course. As I said, I get it. I'm just hoping not to have to go there, especially with an example
> that appears to be running properly, but is just - mysteriously - failing to catch anything.

Ahh, then you shouldn't have acquired these Mosquito Magnets in the first place! :-)

These units have flaws and when (not if) the unit invariably fails or errors out, you're going to wish you have a better view and/or control of the unit...

Heh... well, you'll note my use of the word "hoping". I don't want to have to do it, but if necessary, I'm strapped.

debugger Wrote:

That suggests that if I move it out to the door of the garage (this assumes they don't have a hidden breeding spot and lair inside the garage, of course) it should intercept them before they get to me at the bench ~ 20' inside. Obviously it was worse in the evening when it began to cool off. But I left it running for a couple of days, during most of which I wasn't in there, and it still caught nothing.

Nothing at all. It sounds like something is wrong with the trap output of CO2 and attractant. I remember reading that unburned propane and carbon monoxide are mosquito repellents, or perhaps there is insufficient CO2 or water vapor. You can check the propane consumption to a fair degree of precision using a hand held luggage scale to weigh the tank and connected regulator and hose in a consistent manner and record the time and weight. Given that 20# of propane should be used in 21 days, verify the consumption rate. Last year, my Patriot was not catching nearly as many mosquitos as my Defender, had running and startup failures, and I noticed the tank lasted much longer than 21 days, and this year noticed that the nozzle was severely plugged (see the blog).

Regarding gas detection, a snap web search turns up potential sensors: Carbon monoxide can be measured using the $5 Analog CO Sensor or some other digital module. Propane, etc., can be detected by something like the Gas Propane / Butane / LPG / LNG Detector Sensor Module (sold out at that link). Perhaps I should order some sensors for trap comparisons.

Exhaust temperature and humidity is also important, and easily instrumented.

The other consideration is that perhaps the octenol is repelling the particular species around you. Try running without it. Although it purportedly increases the catch rate 10 times, it may not be working in your location. You could try to identify a biting mosquito for more information.

Sounds plausible. A bad combustion mix could be messing up the attraction without being far enough off of temp
spec to signal the controller. I was thinking about CO2 measurement; a few years back I built a CO2 meter/controller
for grow rooms, and if I can track down my partner on that project I should be able to get it back. (I based it on the
Senseair K30)

It would help to have some kind of reference to measure against, though. Have any of you measured CO2 concentration
at the output of your unit, e.g. at one foot or something?

I'm pretty sure octenol is appropriate for this area, but could always stand to confirm. And I've got a digital commercial shipping scale (good to a couple of decimal points) that I should be able to press into service for some consumption
measurement.

All good places to start.



Edited 1 time(s). Last edit at 06/22/2021 04:30PM by debugger.

If it helps, here are some temperature readings I took off my units. They were measured with a thermocouple stuck up here:



My Independence measures 35C, while my Liberty measures 28C, but it's a cold rainy 65F day today. Both have been catching mosquitos and no-see-ums. However, based on the two Independence that I've been running for years, I believe that the Liberty is running cold, even though it's meeting it's operating range and catching bugs.

If you have a thermocouple, see what output temperature your unit is registering....

Has anyone determined what the stock controller's acceptable temperature range is?

debugger Wrote:
-------------------------------------------------------
> Has anyone determined what the stock controller's acceptable temperature range is?

In terms of the plume exhaust, I believe they are trying to mimic human breath, so somewhere just below body temperature, which is why I believe my Liberty is running low and the Independence is closer.

In terms of the thermistor operating range, I had characterized the Independence many years ago. It's thermistor location is different from the Liberty and Defender.



The table above are the thermistor values and resultant voltage for the units. The Independence thermistor is located at the beginning of the cat converter, right after the carburetor, and thus measures the outside case of the actual combustion area. The Liberty (and Defender as well, I think) has the thermistor at the end of the cat converter, and measures via a thermal conductor in the path. As a result, the thermistor values for the Independence are higher than the Liberty.

You can see the Independence goes into Run mode (green shading) at around 130C. It goes into over-temperature mode at about 220C so you can see that there is quite a wide range of operating temperature. The Liberty goes into Run mode at about 110C, and for some reason, I either didn't characterized the over temperature, or it never went into it, but I don't seem to have a number for that. So far, I have only seen the Liberty thermistor temperature get to about 170C in 85F ambient temperature. Since I was only getting an exhaust temperature of 28C today @65F, I will have to measure it again at warmer temperatures, since it's still catching pests. It probably means that there can be quite a wide variation in operating temperature for them to work, which is why your lack of capture is such a mystery.

See, what would absolutely rock here is a catch vs. temp graph.

Waitasec... If it's this obvious, colour me stupid.

I don't have the little cap for holding the octenol cartridge. It was missing when I got it, but the cartridge that was in
it was wide enough to just wedge in place. Same with the ones I presume(d) are NFG. The new one, though, is a little
narrower, so I jamming a carpenter's pencil in alongside to keep it from dropping out.

Does having it capped matter?

debugger Wrote:
-------------------------------------------------------

> Does having it capped matter?

I would think not having the cap would allow MORE octenol to flow out. UNLESS, the pencil is covering the two holes...

If you have a 3D printer, you can print a replacement

https://www.thingiverse.com/thing:3685540



Edited 1 time(s). Last edit at 06/23/2021 12:15AM by Independence.

One would think, but...

No, I don't have a printer, but the pal who has my CO2 meter/controller does, giving me yet another reason to try to
track him down this week.

debugger Wrote: ...

I was thinking about CO2 measurement; a few years back I built a CO2 meter/controller for grow rooms... It would help to have some kind of reference to measure against, though. Have any of you measured CO2 concentration at the output of your unit, e.g. at one foot or something?

I monitor the air intake temperature and humidity to check combustion, given that the combustion, when normal, raises the gas temperature a certain amount, depending on the humidity. I have not measured the trap output, but am willing to do so with a second NodeMCU and some sensors in a configuration that could run for a few days under different weather and air intake conditions. For comparable measurements, we should agree on the sensors. The NodeMCU has 1 analog input, plus I2C for any number of other boards. I suggest temperature, humidity, propane, and carbon monoxide, and, for completeness, carbon dioxide. The first four can be measured just inside the open exhaust. The CO2 might need to be a bit lower below the exhaust to avoid overload, although it would be exposed to the weather. My brief search found multiple $5 analog sensors for propane, etc., but the boards adding an ADC for I2C were +5v, while the NodeMCU is a +3.3v device.

I'm pretty sure octenol is appropriate for this area, but could always stand to confirm.

It may be good for your area, but you might have a different breed in your immediate vicinity. A little far-fetched, perhaps.

And I've got a digital commercial shipping scale (good to a couple of decimal points) that I should be able to press into service for some consumption measurement.

Those propane tanks with their circular bottom edges are a nuisance for most scales. Perhaps yours is large enough to handle the tank. It was easier for me to carry the hand held scale from unit to unit, hook it on the top, lift, and read instead of fighting with sliding a scale underneath and balancing the scale without adequate leveling, etc.

dev Wrote:
-------------------------------------------------------
> but the boards adding an ADC for I2C were +5v, while the NodeMCU is a +3.3v device.

Although the datasheet for the esp8266 device used in the NodeMCU lists the maximum i/o pin voltage VIH as 3.6v, the CEO of the company has publicly stated that it is 5V tolerant, and I have used esp8266 modules in this mode in many projects without issue.

dev Wrote:

> Those propane tanks with their circular bottom edges are a nuisance for most scales.
> Perhaps yours is large enough to handle the tank. It was easier for me to carry the
> hand held scale from unit to unit, hook it on the top, lift, and read instead of fighting
> with sliding a scale underneath and balancing the scale without adequate leveling, etc.

That's why I mentioned the scale type. Tank base is ~7" diameter. Scale is 12"x14"
and reports 10gm resolution (forgive my mixing units - it's sort of a canuck thing).

My Patriot is running but the catch is a small fraction of that of the Defender. They are of course in different locations, the Patriot just behind the southern patio, and the Defender between the wetland area and the house. I just used the gauge to measure the tank weights for the first time to track propane consumption, so no info on that yet. There were at least two mosquitos hovering in the Patriot's vicinity, and I had to weave and dodge to avoid getting bitten, so something is happening there, but maybe not enough.

Thinking about your problem, I remembered from 2 years ago that the Defender had a problem that was hard to diagnose. I called it "abnormal combustion," and wrote about it in the blog starting around August 1, 2019 (scroll down). The trap caught a few mosquitos, but not in large quantities. After a period of investigation, and several blog entries later, I had cleaned and installed the correct nozzle, and a new regulator, and the trap has been getting large catches since.

debugger Wrote:
-------------------------------------------------------
> Has anyone determined what the stock controller's acceptable temperature range is?

Have you had a chance to measure your exhaust temperature yet? Yesterday, I added another temperature sensor to my Independence unit and was able to log ambient temperature against exhaust temperature from last evening to now. Basically, the temperature ranged from 85F exhaust @57F ambient at night, to 107F exhaust @82F ambient late this morning. Within this range, it caught it's usual amount of bugs, so I don't know which temperature is optimum but it's probably when it was closest to body temperature. However, the fact that my Liberty can still catch bugs despite it's significantly lower exhaust temperature range implies that the combustion CO2 is not the only factor, and that the octenol also plays a role.

The exhaust temperature tracked closer to the ambient temperature than I would have expected. But since they don't do anything to actively regulate the combustion, I supposed this is to be expected. In the end, the unit is designed for warm summer temperatures when the pests are most active.

If you're interested, here is the chart from this morning:



That anomaly in the ambient temperature at about 11:48am was when the sun hit the sensor. After I moved it to the shade, it resumed tracking with the exhaust temperature.



Edited 1 time(s). Last edit at 06/24/2021 07:35PM by Independence.

Independence wrote:

The exhaust temperature tracked closer to the ambient temperature than I would have expected. But since they don't do anything to actively regulate the combustion, I supposed this is to be expected.

I would think so. When the trap is working normally, the sensor temperature seems to track the ambient within a few degrees 83-93­°C or so. The propane flow rate is fixed. When 100% burned it produces a certain amount of heat (e.g., BTUs). I understand that this heat will raise the intake and propane temperature by a certain amount depending on the mass of the incoming air, which is related to the pressure and humidity. I don't know how much heat is lost from radiation from the trap case. Here is some data:

Temp=97C (0) at 5:39974 F=1 I=0 G=1 S=1 E=0 T=13.8C H=79% 	97-13.8    = 83.2	"2021-06-24 03:11:25",
Temp=96.7C (0) at 5:41967 F=1 I=0 G=1 S=1 E=0 T=13C H=78.4% 	96.7-13    = 83.7	"2021-06-24 03:46:01",
Temp=114C (-0.1) at 5:5800 F=1 I=0 G=1 S=1 E=0 T=26.3C H=37.5% 	114-26.3   = 87.7	"2021-06-23 17:18:26",
Temp=112.8C (0) at 5:83699 F=1 I=0 G=1 S=1 E=0 T=29C H=30.2%   	112.8-29   = 83.8   	"2021-06-24 15:50:19"
Temp=129.9C (0) at 5:1395 F=1 I=0 G=1 S=1 E=0 T=27.6C H=32.1% 	129.9-27.6 = 102.3	"2021-06-24 16:22:32",

Temp is the sensor temperature. T and H are the intake temperature and humidity.
All in all, the trap is fairly consistent over a 15°C ambient temperature range. The temperature difference seems to be higher when the trap is first running (low counts in the state counter, e.g., 5:5800.

I included the last entry to show an outlier, something wrong. When measuring the tank weight, I jostled the power cord, and it pulled out, having been damaged by a brutal encounter with a hose a few days ago. Of course, I have not fixed it. But this most recent startup was starting from a 112°C temperature, and cooled down to 70°C before starting a new ignition/gas cycle. This ignition cycle went into the "warmup" then "run" state after only 120 seconds, pretty quick.

I cooled the trap to < 30°C and restarted it. I hope it will work correctly this time. Wait a moment ... Wait a moment ... Nope, it's doing the same thing, over 130°C as of now. The over temperature limit is 140°C, I will have to keep an eye on the MQTT to make sure that if this is exceeded, sending the trap into a fault condition, I can remember to restart the trap for this evening's catch.

dev Wrote:
-------------------------------------------------------
When the trap is working normally, the sensor temperature seems to track the ambient within a few degrees 83-93­°C or so.

My data shows the exhaust temperature tracking to the sensor/thermistor temperature as well, but since the exhaust temperature also tracks to the ambient temperature, it would seem to agree with your data. I would have liked to see some exhaust temperatures from your unit, since that would give me something to compare my unit with. I suspect the Independence runs hotter in order to support the TEGs and this is probably borne out by the fact that the Independence has a 1.75 GPM nozzle, while the Liberty (and Defender?) has a 1.5 GPM nozzle. How both units can claim 21 days of operation on one tankful, I don't know, unless the regulators are also different. I will get a better idea with my runtime counter when the tank runs out, but at 10 hours runtime a day, it will take a while....

Temp=97C (0) at 5:39974 F=1 I=0 G=1 S=1 E=0 T=13.8C H=79% 97-13.8 = 83.2 "2021-06-24 03:11:25",
Temp=96.7C (0) at 5:41967 F=1 I=0 G=1 S=1 E=0 T=13C H=78.4% 96.7-13 = 83.7 "2021-06-24 03:46:01",
Temp=114C (-0.1) at 5:5800 F=1 I=0 G=1 S=1 E=0 T=26.3C H=37.5% 114-26.3 = 87.7 "2021-06-23 17:18:26",

Based on the first line above, your Defender is in Run mode when the thermistor is as low as 97C?

This ignition cycle went into the "warmup" then "run" state after only 120 seconds, pretty quick.

Does your code not check for a temperature threshold before an ignition cycle? My units will not enter the ignition stage until the sensor temperature drops below a certain level. Though, it's strange that a short warmup cycle can result in the higher equilibrium temperatures you are now seeing...



Edited 1 time(s). Last edit at 06/25/2021 01:01AM by Independence.

Independence Wrote:

I would have liked to see some exhaust temperatures from your unit, since that would give me something to compare my unit with.

I can add a BME280 temperature/pressures/humidity sensor to the defender:

The module should not be exposed to rain, so attached just inside the exhaust port?

I will get a better idea with my run time counter when the tank runs out, but at 10 hours runtime a day, it will take a while....

You could try a luggage scale. Mine displays to 2 hundredths of a pound (e.g, D 32.28 P 32.97 #). It takes practice and patience to wait for the tank to stop swinging, however, while dodging hovering mosquitos.

Based on the first line above, your Defender is in Run mode when the thermistor is as low as 97C?

Uh, oh. when reading this, I thought, oops, what about the lack of open circuit calibration from the new Mini D1 NodeMCU, which would invalidate my earlier posted data. However, upon further reflection, the results were not far off, as the thermistor and intake cold temperatures were fairly close with the new board. Still, here are some excerpts from last year's run from the old version in which the open circuit voltage was correctly read and applied to derive an accurate thermistor resistance reading, and then a fairly accurate temperature:

2020-09-19 09:49:52 Temp=104.4C (0) at 5:549948 F=1 I=0 G=1 S=1 E=0 T=21C H=28.9% M=8704 (338/1010:9468) R=70~140 W=-74
2020-09-19 14:04:44 Temp=110.1C (0) at 5:564239 F=1 I=0 G=1 S=1 E=0 T=28C H=19% M=8704 (301/1011:7980) R=70~140 W=-74
2020-09-21 06:54:04 Temp=92.5C (0) at 5:701592 F=1 I=0 G=1 S=1 E=0 T=7C H=64.9% M=8696 (430/1009:13979) R=70~140 W=-75
2020-09-21 08:16:47 Temp=97.5C (0.1) at 5:706231 F=1 I=0 G=1 S=1 E=0 T=13.4C H=59.8% M=8696 (392/1010:11940) R=70~140 W=-74
2020-09-21 08:41:47 Temp=100C (0) at 5:707632 F=1 I=0 G=1 S=1 E=0 T=16.2C H=52.4% M=8696 (373/1010:11022) R=70~140 W=-73
2020-09-21 10:36:13 Temp=105C (0) at 5:714050 F=1 I=0 G=1 S=1 E=0 T=22.7C H=30.3% M=8696 (334/1010:9300) R=70~140 W=-76
2020-09-21 12:47:36 Temp=109C (0.1) at 5:721415 F=1 I=0 G=1 S=1 E=0 T=26.5C H=16.7% M=8688 (308/1011:8247) R=70~140 W=-74
2020-09-21 18:17:30 Temp=100C (0) at 5:739915 F=1 I=0 G=1 S=1 E=0 T=16.8C H=48.9% M=8696 (373/1010:11022) R=70~140 W=-77
2020-09-21 23:17:02 Temp=95C (0.1) at 5:756709 F=1 I=0 G=1 S=1 E=0 T=9.5C H=64.6% M=8696 (411/1010:12916) R=70~140 W=-80
2020-09-22 06:34:38 Temp=93.3C (0) at 5:781251 F=1 I=0 G=1 S=1 E=0 T=8.2C H=63.9% M=8696 (424/1010:13620) R=70~140 W=-82
2020-09-22 09:02:04 Temp=100C (0) at 5:789518 F=1 I=0 G=1 S=1 E=0 T=16.7C H=42.3% M=8696 (373/1009:11040) R=70~140 W=-81

Subtracting the ambient from the thermistor readings above:

104.4- 21   = 83.4
110.1- 28   = 82.1
92.5 - 7    = 85.5
97.5 - 13.4 = 84.1
100  - 16.2 = 83.8
105  - 22.7 = 82.3
109  - 26.5 = 82.5
100  - 16.8 = 83.2
95   - 9.5  = 85.5
93.3 - 8.2  = 85.1
100  - 16.7 = 83.3

Does your code not check for a temperature threshold before an ignition cycle? My units will not enter the ignition stage until the sensor temperature drops below a certain level. Though, it's strange that a short warmup cycle can result in the higher equilibrium temperatures you are now seeing...

Strangeness, yes, but the short warmup apparently did not cause the higher thermistor temperature.

The code checks for and reports two temperature limits, a minimum and a maximum, here R=70-140. When restarting, the trap, just previously running at ~105°C, cooled until the minimum 70°C. The code then records the starting temperature, starts the igniter (now softly) and waits, then applies the gas, keeping the igniter on until either a +5°C temperature rise is detected, or the temperature exceeds the minimum. In this case, a minimal temperature rise caused the temperature to rise above minimum, and the trap entered the running state as expected. But then, instead of leveling off at ~105°C, the thermistor temperature continued to rise to ~130°C. Stopping and cooling the trap to near ambient (<30°C) and restarting again resulted in high temperature operation. Currently, it is running at that temperature:

Temp=129.7C (0) at 5:60415 F=1 I=0 G=1 S=1 E=0 T=22.1C H=73.2% M=26608 (201/1024:4597) R=70~140 W=-71.85 B=0.0 V=3.5

The trap restarted the day before yesterday and several times previously with no issues, so this is a mystery. The trap seems to be in a different state now, possibly broken. Perhaps the nozzle needs cleaning. I will measure it, but first, I want to replace the basket to see if it is catching mosquitos in this mode. If not, this may be one reason why debugger's trap is failing. On the other hand, perhaps the basket is chock full, and this is the correct state.

After this is new investigation, I can look into adding a BME280 to the exhaust port.

Dev wrote:

I can add a BME280 temperature/pressures/humidity sensor to the defender:
The module should not be exposed to rain, so attached just inside the exhaust port?

I had used that sensor before but found it a little bit too sensitive to moisture. I have since switched to the Dallas 18B20 sensor, which is more waterproof, but I loose the humidity. Yes, attach just inside the exhaust port, but depending on the ambient conditions, the exhaust can sometimes be subject to water condensation.

You could try a luggage scale. Mine displays to 2 hundredths of a pound (e.g, D 32.28 P 32.97 #). It takes practice and patience to wait for the tank to stop swinging, however, while dodging hovering mosquitos.

I DO weigh my tank exactly the same way, but would prefer the online runtime counter. However, in order to get a baseline representative value, I need to run an entire tank and see what number I get for both my units. I also plan on making an adapter for the schrader valve to my manometer and comparing the gas pressure between the units, given the difference in operating temperature between the Independence and Liberty.

The code checks for and reports two temperature limits, a minimum and a maximum, here R=70-140

How did you come up with 70C as the low end of the run state for your Defender? My characterization of my stock Liberty unit indicated that it does not enter Run state until the thermistor is at about 105 C. A bench test with a potentiometer in place of the thermistor validated that and also shows that once it has entered Run state, there is some hysteresis and it doesn't exit Run state until around 80-85 C, whereupon it exits with an error code of 'insufficient gas/gas is empty'. I didn't have these limits in my thermistor chart, but did I had them in the code, so I must have characterized them. The overtemperature limit was about 155 C