N2K Instrumentation - Part 2 of Electronics Upgrade

[dtfeld]

So, appears it simply draws a line between points and interpolates a relationship between temp and voltage; the more points of data the more applicability to a thermistor curve and consequently the more accurate. If this is true I would recommend you remove from the engine the temperature sensor and put in a pot of water then measure resistance and water temperature concurently as you heat the water to boiling; that is what I did. Then you can calculate the voltage measurements to load into the Chetco.

Tom

To your first point, I believe this is what it is going on.

As to off boat testing and calculated voltages, I'm not sure I can do this. I believe the Checto is reading a voltage, doing some preconditioning and processing of the signal, then it is outputting a scaled 8 bit value (0-255). The unit then looks up that value in the calibration table and outputs the associated temperature (or other real world value depending on whats being measured). The benefit is you can calibrate any sensor (resistance low->high or High->Low. But I cant input a raw voltage. The downside, is you have to identify the 8 bit value via software and correlate that to your real world value in real time. My next job is to validate what being displayed on my MFD vs the temp measured at the sender. I'll graph it and see where I am.

However, If I were to install this on the engine I believe this has a switch selectable resistance mode reading the resistance of the senders, and I think you can dial in the exact resistance vs temp/pressure reading. I have not attempted this yet, but in talking to Chetco, this would be the easier and more accurate method.

Just curious, what would you consider acceptable accuracy? Considering the current analog gauges, I probably would be happy with +/-5 deg. Just reading the gauge I could be that far off. What I'm really after is an alarm where the temp/pressure quickly deviates from the norm.

 

[ttmott]

Tom

To your first point, I believe this is what it is going on.

As to off boat testing and calculated voltages, I'm not sure I can do this. I believe the Checto is reading a voltage, doing some preconditioning and processing of the signal, then it is outputting a scaled 8 bit value (0-255). The unit then looks up that value in the calibration table and outputs the associated temperature (or other real world value depending on whats being measured). The benefit is you can calibrate any sensor (resistance low->high or High->Low. But I cant input a raw voltage. The downside, is you have to identify the 8 bit value via software and correlate that to your real world value in real time. My next job is to validate what being displayed on my MFD vs the temp measured at the sender. I'll graph it and see where I am.

However, If I were to install this on the engine I believe this has a switch selectable resistance mode reading the resistance of the senders, and I think you can dial in the exact resistance vs temp/pressure reading. I have not attempted this yet, but in talking to Chetco, this would be the easier and more accurate method.

Just curious, what would you consider acceptable accuracy? Considering the current analog gauges, I probably would be happy with +/-5 deg. Just reading the gauge I could be that far off. What I'm really after is an alarm where the temp/pressure quickly deviates from the norm.

Here would be my concern; in the below graph should a point be established at 70 degrees on the blue thermistor line then another point at 220 degrees and a straight line drawn between, most of the data on that linear line would be greatly inaccurate. The more data points that can be entered representative of the actual thermistor readings will better represent the correct temperature readings. Regarding accuracy, I would be concerned more about accurate temperatures between 170 and 210 degrees as that is where monitoring is important and the reason I set linearization in that range for the Noland product which would not integrate multiple points for a curve; I would say that +/-5% accuracy is reasonable.

 

[dtfeld]

I think the beauty of this system is that you are assigning actual temperature values to data points. With "enough" data points, I should have very accurate response across the temp range. If I change over to the resistive mode, then the process would be almost identical to your process.

I agree this is a little to much black box majic, but I used 4 points as a start and 3 at the hot end as I agree that is the critical zone.

In thinking about how to do this, I think I'm going to set up a video camera and video the vDash values and a digital temp guage. I would be able to go back and add a huge number of data points. The points would correspond to the actual and correct temperature across the range, and I would expect pretty good accuracy.

We'll see. The learning curve is pretty steep and I'm making it up as I go!

 

[Blkbird]

I think the beauty of this system is that you are assigning actual temperature values to data points. With "enough" data points, I should have very accurate response across the temp range. I used 4 points as a start and 3 at the hot end as I agree that is the critical zone.

In thinking about how to do this, I think I'm going to set up a video camera and video the vDash values and a digital temp guage. I would be able to go back and add a huge number of data points. The points would correspond to the actual and correct temperature across the range, and I would expect pretty good accuracy.

We'll see. The learning curve is pretty steep and I'm making it up as I go!

David,

I have a digital temperature gauge available. Since we are both local, couldn't we go out and take a series of measurements while the boat is underway?

 

[dtfeld]

David,

I have a digital temperature gauge available. Since we are both local, couldn't we go out and take a series of measurements while the boat is underway?

That would be a huge help!

 

[Blkbird]

That would be a huge help!

Excellent. Deal is you have to teach me how to set up my own digital gauges

 

[dtfeld]

Deal. But it will be the blind leading the blind!

 

[MonacoMike]

My understanding is this can be applied to a twin gas boat just as easily....easy being a relative term.

Thanks,

MM

 

[dtfeld]

Tom

Unfortunately, due to lack of documentation, I'm not really sure what the math is. But once I measured the temps and correlated that to the G2's raw data point, one button push did the calculation. My "understanding" is this is a linearazation of the data a similar calculation to what you did (I've learned a lot from you on CSR so again thank you).

"Does it track temperature from cold to hot or is the cold inaccurate?" I think what you asking is does this lineralize the data? The answer to best of my limited knowledge is yes. If this is not what you were asking clarify and I'll get you an answer.

In talking with Skybolt, he made it more accurate with more datapoints (I think he measured every 10 deg). I also need to add the hottest point (running under load), but I need an assistant to run the boat while I measure the temps. However, because these sensors resistance curve is flatter at the hotter end, I would expect more accuracy at the hot end of the spectrum, even beyond where I measured.. My next job is to go back and validate what I have.

One thing I understand the Chetco can do and the reason I chose it over the Noland, is that the Chetco Digital unit can bias its sensors with a regulated voltage getting rid of a source of error and frustration. However, I am not setting this up that way at this time. Currently I have this connected to the existing gauges that use the variable voltage the boat produces. If I like the unit and it works well, I may move it to the engine room where I can connect directly to the senders in resistance mode (selectable on the inside of unit with a dip switch). That would give me the most accurate results, but I'm not ready to give up the gauges just yet.

So far, I like the unit but documentation leaves a lot to be desired.

Tom: I found the documentaion. It is as I described, takes raw sensor data, does some averaging, then looks it up in a table. The process I used is a quick method to fill in the entire table based on a couple data points (Low middle and high), which is probably accurate enough for my application. Theoretically, you could measure and correlate 256 data points and have the correct temp across the entire range of a sender for more accuracy. See the explanation below in the attached pdf.

This gets around one of the problems with the Albi Combi...it couldnt handle the US senders as the resistance was inverse to the European senders.

This also allows easier onsite (especially on a boat) calibration using empirical data. This was the reason I chose the Chetco over the Noland. Jury still out for me but it is working as I expected.

 

[ttmott]

Tom: I found the documentaion. It is as I described, takes raw sensor data, does some averaging, then looks it up in a table. The process I used is a quick method to fill in the entire table based on a couple data points (Low middle and high), which is probably accurate enough for my application. Theoretically, you could measure and correlate 256 data points and have the correct temp across the entire range of a sender for more accuracy. See the explanation below in the attached pdf.

This gets around one of the problems with the Albi Combi...it couldnt handle the US senders as the resistance was inverse to the European senders.

This also allows easier onsite (especially on a boat) calibration using empirical data. This was the reason I chose the Chetco over the Noland. Jury still out for me but it is working as I expected.

Very cool - ten data points with most at the higher temperatures would give good data.