Maretron TMP100 inaccurate temperature readings

[Wdeertz]

Anyone using the Maretron TMP100 temperature sensor with the ambient sensors? I'm trying to monitor my refrigerator/freezer temperatures and am getting very inaccurate temperatures which are off about 9-14f. I've extended the probes with about 10 ft of AWG18 wire which according to the resistive table shouldn't impact the reading that much. The TMP100 is measuring my refrigerator at 48f/9c while the actual temperature is 37f/3C. According to the resistive temperature table 3C should be 8415 ohms versus 9C 6264 ohms (see attached screenshot). It seems the table is backwards as if I put a resistor across the temperature probes is raises the temperature rather than drop it.

The easy answer is if Maretron had a calibration offset configured in the firmware but this would be too easy. Anyone using the Maretron TMP100 and figure out a way to get more accurate readings? I could use large size wire but not sure this will solve my problem.

 

[twistedtree]

Is it possible you are connected to one of the first two ports on the TMP100? Those are for thermocouples, not J-curve NTC thermistors.

 

[Wdeertz]

Is it possible you are connected to one of the first two ports on the TMP100? Those are for thermocouples, not J-curve NTC thermistors.

No I am connected to one of the last 4 inputs. I did mistakenly use the first input and got a really high reading (like 500f) and figured out the first two inputs are for the EGT sensors only. The resistive/temp table seems backwards to me. I tried a 8k resistor across the input terminals and this raised the temperature while the table seems to show it should lower.

 

[Greg QS]

Maratron just works better with their sensors. Go into the setup and check the max temp setting and set the max setting. Then re check. if still off lower the max setting until you get the correct reading.

 

[Wdeertz]

Maratron just works better with their sensors. Go into the setup and check the max temp setting and set the max setting. Then re check. if still off lower the max setting until you get the correct reading.

These are the Maretron supplied ambient sensors (https://www.maretron.com/support/manuals/TP-AAP-1 Installation Instructions 1.1.html). The sensors appear to be manufactured by US Sensor Corp as that is the resistive/temp table provided by Maretron (https://www.maretron.com/wp-content/phpkbv95/article.php?id=567&oid=30). The IPG100 has no setting for min/max temperatures so I'm assuming you are referring to the displays in N2KView. These settings only effect the gauges shown in N2KView not the actual sensor reading.

 

[twistedtree]

These are the Maretron supplied ambient sensors (https://www.maretron.com/support/manuals/TP-AAP-1 Installation Instructions 1.1.html). The sensors appear to be manufactured by US Sensor Corp as that is the resistive/temp table provided by Maretron (https://www.maretron.com/wp-content/phpkbv95/article.php?id=567&oid=30). The IPG100 has no setting for min/max temperatures so I'm assuming you are referring to the displays in N2KView. These settings only effect the gauges shown in N2KView not the actual sensor reading.

Maybe he means setting in the TMP100 setup screens? I don't recall what settings are available.

 

[twistedtree]

What about connection polarity? You would think it wouldn't matter, but I seem to recall discovering the opposite about one of the maretron sensor. Just can't recall if it was temp or something else.

 

[Wdeertz]

Maybe he means setting in the TMP100 setup screens? I don't recall what settings are available.

Not much available to configure in the TMP100 other than a label, instance # and PGN frequency.

 

[Wdeertz]

What about connection polarity? You would think it wouldn't matter, but I seem to recall discovering the opposite about one of the maretron sensor. Just can't recall if it was temp or something else.

I confirmed I have the correct polarity but like you don't think it matters. On the SIM100 switch module it clearly states in the manual that polarity matters but is silent in the TMP100 manual.

 

[twistedtree]

According to the resistive temperature table 3C should be 8415 ohms versus 9C 6264 ohms (see attached screenshot). It seems the table is backwards as if I put a resistor across the temperature probes is raises the temperature rather than drop it.

If you put a resistor across the TMP100 terminals in parallel with the temp probe, it lowers the resistance as seen by the TMP, and that corresponds with an increase in temperate. So that part actually sounds correct.


Have you disconnected the probe and measured its resistance with a VOM? That would help narrow down whether the probe is out of calibration, or if the TMP is off.

 

[twistedtree]

I see Maretron specifies +/- 2C temp accuracy which is pretty poor. I don't know how much of that is the probe vs the TMP100. That said, you are seeing a greater error than that.

 

[Wdeertz]

What about connection polarity? You would think it wouldn't matter, but I seem to recall discovering the opposite about one of the maretron sensor. Just can't recall if it was temp or something else.

TT, ok the mystery gets deeper. I figured I'd measure the resistance across each probe wire and as expected the polarity didn't matter. See screenshot showing the pertinent resistances measured versus what it should have been per the temperatures being displayed. I'm stumped why the large resistance differences unless the table Maretron posted is wrong.

Opening the refrigerator/freezer for a quick 15-20 seconds caused the probe temperature going up 10-15f. I have both of temperatures now using a 10 minute average which tends to eliminate the spikes from opening the refrig/freezer doors. After the temperatures settle I'm off 5-6f which may be as close as I'm going to get.

 

[Wdeertz]

If you put a resistor across the TMP100 terminals in parallel with the temp probe, it lowers the resistance as seen by the TMP, and that corresponds with an increase in temperate. So that part actually sounds correct.


Have you disconnected the probe and measured its resistance with a VOM? That would help narrow down whether the probe is out of calibration, or if the TMP is off.

Duh, what was I thinking. If I can figure out the table I could get the correct resistor and rather than put across the TMP100 terminals only put on one of the wires. This would raise the resistance and act as a temperature adjustment.

 

[Bmarler]

Duh, what was I thinking. If I can figure out the table I could get the correct resistor and rather than put across the TMP100 terminals only put on one of the wires. This would raise the resistance and act as a temperature adjustment.

Sometimes I’ll use a variable resistor so I can dial it up and down to find the best value. Then solder in a fixed one to finish.

 

[Wdeertz]

Sometimes I’ll use a variable resistor so I can dial it up and down to find the best value. Then solder in a fixed one to finish.

Good suggestion, I put a 8k ohm resistor on one of my freezer probes which was incorrectly showing 17f (correct temp 4f) and it’s now showing 3.6f , happy camper.

I suppose this is a one way fix as there’s no way to reduce the resistance (short of using a larger wire) but in my experience the probes tend to be too high.

 

[twistedtree]

Just to confirm, when you measured the probe resistance, at least one of the wires was disconnected from the TMP100, right? That could impact the measurement otherwise.


The thermistors are supposedly accurate to 0.1C, so if your are reading that far off, they would appear to be junk. Are they Maretron branded sensors, or purchased from some other source?

 

[Wdeertz]

Just to confirm, when you measured the probe resistance, at least one of the wires was disconnected from the TMP100, right? That could impact the measurement otherwise.


The thermistors are supposedly accurate to 0.1C, so if your are reading that far off, they would appear to be junk. Are they Maretron branded sensors, or purchased from some other source?

I measured with the wires connected to the TMP100 (power was off) as it was easier to get a good connection. Would it make a difference? I’ll get out my alligator clips and measure from the wires only.

As mentioned in a prior post I was able to get one of the freezer probes 99.9% accurate by adding a 8k ohm resistor. I’ll do the same on the others to get an accurate reading. I suppose I could have been happy with the freezer probe showing 20f (when actual was 5f) and set an alarm for 30f but I’m a little OCD and like accurate readings.

 

[twistedtree]

I measured with the wires connected to the TMP100 (power was off) as it was easier to get a good connection. Would it make a difference? I’ll get out my alligator clips and measure from the wires only.

As mentioned in a prior post I was able to get one of the freezer probes 99.9% accurate by adding a 8k ohm resistor. I’ll do the same on the others to get an accurate reading. I suppose I could have been happy with the freezer probe showing 20f (when actual was 5f) and set an alarm for 30f but I’m a little OCD and like accurate readings.

Isolating the sensor is important when measuring. Otherwise you are measuring the impedance of the TMP100 in combination with the sensor, not just the sensor. The difference may not be significant, or even noticeable at all, but you don’t know until you have checked.


Also, what are you comparing the sensor reading to, because the other temp sensor will have error too. In
My experience, temp sensors of all kinds can be pretty inaccurate. So just for arguments sake, how do you know it’s the sensor that’s wrong, and not your other thermometer?

 

[Greg QS]

I would first put the sensor on another channel and see if you get a deferent reading out of the sensor.
You can also move a know sensor to that channel and see if the reading changes.
The channels on the unit are independent . 1 channel can go bad.

I had a tm100 give me wacky readings . Was sent back and replaced by maretron.
I have a sim100 have a bad cannel. ( grate on my smoke alarm)

One last thing. the unit only supports 2 temperature ranges for the ambient sensors. (-20°C to 80°C or -4°F to 176°F)

Off the maretron website.
Maretron's TMP100 measures the temperature for up to 6 temperature probes and reports the information over an NMEA 2000® network. The TMP100 supports up to 4 thermistor probes and 2 high temperature thermocouple probes. Optional thermistor probes (-20°C to 80°C or -4°F to 176°F) cover a wide range of applications including cabin air temperature, engine room air temperature, refrigerator/freezer temperature, under bolt temperature (inverters, charges, pumps, motors, etc.), tank temperatures (live well bait, hot water, etc.), and air duct temperatures. The optional thermocouple probes (0°C to 900°C or 32° to 1652°F) are used to measure Exhaust Gas Temperature (EGT) as part of a comprehensive fuel management system.

 

[Wdeertz]

Isolating the sensor is important when measuring. Otherwise you are measuring the impedance of the TMP100 in combination with the sensor, not just the sensor. The difference may not be significant, or even noticeable at all, but you don’t know until you have checked.


Also, what are you comparing the sensor reading to, because the other temp sensor will have error too. In
My experience, temp sensors of all kinds can be pretty inaccurate. So just for arguments sake, how do you know it’s the sensor that’s wrong, and not your other thermometer?

TT, yes you are 100% correct, the TMP100 was effecting the measured impedance. When I measured only from the wires the temperatures/resistance pretty well lined up with the table.

I'm measuring the actual temperatures with a mercury thermometer and a digital infrared thermometer so am pretty confident on the measurements. I've been monitoring the temperatures over the past day and notice that both the refrigerator (39f-46f) and freezer (-10f-5+f) temperatures are oscillating. (see attached screenshot) The screenshot was over the past 12 hours so the middle of the night when neither door was opened so that cant be the source of the oscillation. In addition I have both set for 10 minute averaging to reduce the swings from opening the doors. The temperatures near the bottom of the oscillation are fairly accurate to the actual temperatures. Any idea what would be causing this oscillation?

Interestingly, the dometic freezer is the one I put the 8k ohm resistor on and it measures a consistent 3-4f which is pretty much the right temperature. The dometic freezer is on a separate TMP100 in the engine room.

 

[twistedtree]

TT, yes you are 100% correct, the TMP100 was effecting the measured impedance. When I measured only from the wires the temperatures/resistance pretty well lined up with the table.

I'm measuring the actual temperatures with a mercury thermometer and a digital infrared thermometer so am pretty confident on the measurements. I've been monitoring the temperatures over the past day and notice that both the refrigerator (39f-46f) and freezer (-10f-5+f) temperatures are oscillating. (see attached screenshot) The screenshot was over the past 12 hours so the middle of the night when neither door was opened so that cant be the source of the oscillation. In addition I have both set for 10 minute averaging to reduce the swings from opening the doors. The temperatures near the bottom of the oscillation are fairly accurate to the actual temperatures. Any idea what would be causing this oscillation?

Interestingly, the dometic freezer is the one I put the 8k ohm resistor on and it measures a consistent 3-4f which is pretty much the right temperature. The dometic freezer is on a separate TMP100 in the engine room.

I can't quite read the info on the screen shot, and would be interested in how long the temp swing takes for a cycle.


That aside, I think you now have two things that say the TMP100 is defective. First is that you have confirmed that the thermistor is reading the correct value. So that leaves the TMP100, which is supposed to be accurate to +/- 2C. You also have this odd oscillation in the reported temp.


Now one possibility is that the oscillation is just the compressor cycling on and off. It would be normal operation for the temp to rise to some point, the compressor kicks on and drops the temp down, then turns off and the cycle repeats. The 7F and 15F swings that you are seeing seem pretty big, but I have no idea how tightly regulated the temp is on these appliances. One thing you could try is just observing how the reported temp oscillation correlates with the compressor running. If they line up, then you are just seeing the compressor cycling, and the corresponding temps. But even if the oscillation is just compressor cycling, the reported temp from the TMP is still way off.


I'd get on the phone to Maretron. No matter how you slice it, I think you have demonstrated that the TMP100 isn't working properly/meeting specs.

 

[Wdeertz]

That aside, I think you now have two things that say the TMP100 is defective. First is that you have confirmed that the thermistor is reading the correct value. So that leaves the TMP100, which is supposed to be accurate to +/- 2C. You also have this odd oscillation in the reported temp.


Now one possibility is that the oscillation is just the compressor cycling on and off. It would be normal operation for the temp to rise to some point, the compressor kicks on and drops the temp down, then turns off and the cycle repeats. The 7F and 15F swings that you are seeing seem pretty big, but I have no idea how tightly regulated the temp is on these appliances. One thing you could try is just observing how the reported temp oscillation correlates with the compressor running. If they line up, then you are just seeing the compressor cycling, and the corresponding temps. But even if the oscillation is just compressor cycling, the reported temp from the TMP is still way off.


I'd get on the phone to Maretron. No matter how you slice it, I think you have demonstrated that the TMP100 isn't working properly/meeting specs.

I reached out to Maretron support and they had me switch the probes to other TMP100 inputs. This confirmed that its not the TMP100 but rather the two probes on the refrigerator/freezer. The two refrig/freezer probes exhibited the same temp cycling on the new inputs while the other probes flat lined on the prior refrig/freezer inputs. I let Maretron support know this but they haven't got back to me with a possible cause.

The refrigerator cycles between about 35f to 44f (actual is about 38f) over a 75 minute period. The freezer cycles between about -5f to +5f (actual is about -4f) over the same 75 minute period. I have two new still in wrapper ambient temp probes but it was a PITA to install and don't really want to go through that exercise again. I can compensate this temp cycling by using a 10 minute averaging and put a 10 minute delay on any alarms. Not ideal but short of replacing the probes not sure what other options I have. Not sure what material is in the ambient probe but its a weird phenomenon.

 

[Wdeertz]

TT, I should have also mentioned that there's no way the compressor cycling is causing a 9-10f temperature fluctuation. It's a large residential Jenn-Air refrigerator. I'll try to measure the actual temperatures when its cycling near the top of the temperature curve just to confirm 100%.

 

[Bmarler]

it wouldn't surprise me at all to have those temp swings with an analog control refrigerator/freezer. especially with a frost free one. digital control will be somewhat better but there will still be swings. the more insulation you have the longer time peak to peak.
the air inside will fluctuate more than the contents will as it takes time to conduct heat or cold into your contents. if the probes are waterproof you can put them in a glass of water and test again.

 

[Bmarler]

i did a quick google search "typical refrigerator temp swings"
there's quite a lot of posts regarding this subject. read this guys post: https://www.applianceblog.com/mainf...-is-this-normal-for-a-brand-new-fridge.45718/

brand new kenmore, big swings as shown on his graphic readout.

 

[Wdeertz]

it wouldn't surprise me at all to have those temp swings with an analog control refrigerator/freezer. especially with a frost free one. digital control will be somewhat better but there will still be swings. the more insulation you have the longer time peak to peak.
the air inside will fluctuate more than the contents will as it takes time to conduct heat or cold into your contents. if the probes are waterproof you can put them in a glass of water and test again.

Thanks for the inputs. I took your suggestion and for the refrigerator put the probe in a glass of water. See attached screenshot. The temperature cycling is down to about 3f which is almost the probe accuracy range. I think this confrms that a 9-10f air temp swing must be from the normal compressor cycling. I'm a bit surprised as I would have expected it to be about 3-4f.

I'm not sure I want to leave the probe in a glass of water permanently but will tape the probe to a can of soda to moderate the compressor cycling swings. The dometic freezer is using the same probe but not experiencing any temperature swings but I think its because this freezer is packed almost full so when the compressor kicks off there isn't much of a temp swing.

Thanks for all the inputs, I guess we can consider this mystery solved.

 

[SeaMaster101]

All this hassle to get proper temperature reading???
Next time use one of these (shameless plug):
https://acrossoceansystems.com/index.php/product/nmea2000-standard-temperature-sensor-ss-cylinder-terminal/
Easy to install and configure and also accurate. And also displays on Maretron software.

 

[Wdeertz]

All this hassle to get proper temperature reading???
Next time use one of these (shameless plug):
https://acrossoceansystems.com/index.php/product/nmea2000-standard-temperature-sensor-ss-cylinder-terminal/
Easy to install and configure and also accurate. And also displays on Maretron software.

I use a AOS N2K A/C monitoring device. i have a similar temperature device from Yacht devices. The issue with these type of devices is the place the temperature is being measured needs to be fairly close to the N2K backbone and they only measure one device. The Maretron device measures 4 ambient plus 2 high temp (such as EGT) plus the probes can be extended easily with wire. In the end the Maretron temp probes were fairly accurate, I just didn't expect a 9-10f temperature swing from the frost-free cycling. My only complaint with the Maretron is the firmware doesn't allow a calibration offset for minor differences.

 

[SeaMaster101]

I use a AOS N2K A/C monitoring device.

I'm glad that you have found these devices. How did you hear about it?

i have a similar temperature device from Yacht devices. The issue with these type of devices is the place the temperature is being measured needs to be fairly close to the N2K backbone and they only measure one device.

The drop cable for a device as defined by NMEA2000 standard could be up to 6m long so you not necessary need to be close to the NMEA2000 backbone.

The Maretron device measures 4 ambient plus 2 high temp (such as EGT) plus the probes can be extended easily with wire. In the end the Maretron temp probes were fairly accurate, I just didn't expect a 9-10f temperature swing from the frost-free cycling.

To reveal a little secret: AOS temperature devices use DS18B20 digital temperature sensors
https://www.analog.com/media/en/technical-documentation/data-sheets/ds18b20.pdf
which allow you to cut the sensor wire and terminate the three wires inside +5V Sig and GND to a terminal strip. you can extend the cable if you want this way but more importantly, this will allow you to add more sensor in parallel to the terminal strip. As they(the DS18B20) are digital sensors (each with unique ID) and the AOS software in the NMEA2000 box already is designed to handle multiple devices, you will see the temperature sensors that you add show up and you can configure each one independently in the device wifi setup via webpages.

My only complaint with the Maretron is the firmware doesn't allow a calibration offset for minor differences.

and yes, you can calibrate each one of these sensors, with either offset - one calibration point, or liner scale (not necessary for this device) with two calibration points, or nonlinear scale (most often used on the tank level measuring devices) with 3 or more calibration points.

 

[Wdeertz]

I'm glad that you have found these devices. How did you hear about it?

From a google search. Seems to be a small outfit with good customer service. Initially I had an issue with the PGN’s being distributed and their IT guy made a firmware update to get mine working. ?

Yes aware of the 6m drop cable limit. What I was getting at is these sorts of temp probes can be difficult to route to more difficult areas. Thanks for the heads up about extending using a terminal strip and being able to support multiple sensors. That chances the calculus in using these devices. Surprised AOS doesn’t advertise this functionality as it makes their unit more competitive.