We have a 1-Channel Wireless Current (Amp) Sensor modified with a Setra 5-500 Amp CT Clamp.
We were attempting to change the sensors delay from the default 10 min to 2 min (using Node Red), and unfortunately, the only option to change that delay also requires one to enter the new calibration factor for the sensor (side note, this should really be a separate function IMO).
Previously this has not been an issue, and we have been able to recalibrated the sensor with a new calibration factor.
However, with this one, we were not able to correctly back calculate the calibration factor based on “actual” vs “sensor amps”. See the table below. As you can see, we were trying to get the sensor to read about 83 amps, but there was a massive change in sensor readings with only a few decimal change in the calibration factor.
What is the issue here?
At 0.0185 factor, we were reporting 67 amps. But at 0.022 factor we were reporting 112 amps. No matter what we tried, we couldn’t get it to report 83 amps.
Hi Bhaskar, we attemped this calibration again with TWO different edge computers: (1) a New Robustel Gateway running the latest version of Node Red/NCD Enterprise Pallette, and (2) an older NCD Edge Computer running the Wireless Sensor Pallete.
No matter what we tried, we were not able to hone in on a good value:
1: Using Older edge Computer
Actual Amps
NCD Current Monitor
Calibration Factor
93.2
12125.856
606
88.2
9505.856
606
82
2607
0.56
88.1
2630
0.56
165.1
84.009
0.018
164
224.91
0.02
From this it appears the Cal Factor should be betwen 0.018 and 0.02 (this is similar to the original post)
2: Using Robustel Gateway (Calibration Factors seem to be 10x that of the older edge computer calibration)
Actual Amps
NCD Current Monitor
Calibration Factor
164.2
9078.272
606
165
0.872
10.96
163.7
1.375
10.96
163
1.417
10
164.5
44.467
1150
86
576.631
4254
163.5
0
634.45
163.1
0
634
From this round of testing there was no clear trend emerging.
What could be the issue be? This sensor was working perfectly until we tried changing the sensor reporting time (which resulted in the cal factor being reset, and this situation).
Hey, OK we are attempting this now, but we are a bit confused:
Do we connect the 1-channel amp sensor directly to our computer and do this? If so, where is the USB port to connect this?
Because since this is a battery powered sensor, XCTU does not “see It” as it is set to sleep and wakeup every x min. So not sure how to send/receive these commands.
Awesome, thanks for the explanation and for making a temp fork of the code.
We used the Node Red Temp Fork, and here is what we found:
Basically, when we put in a number like 50,000, we get an absurdly high amp reading.
When we put in ~1.6, we are getting close, but it is still bouncing around.
This is the behaviour we were seeing originally (Screenshot of original post)
We don’t have extra time to play with this at the moment, so we will continue tomorrow, but for now, its very clear that the calibration of this device (which is a 400 amp CT Clamp) requires a VERY low calibration value (like 1-2 vs 60,600 which is the default).
What would the reason be for this?
We will continue to play with this a bit.
Question: With this temp fork, are there are decimal constraints? In other words, how many decimal points will it accept? It seems to accept at least 1.
Hi @Bhaskar I think I figured out what is going on after conducting more tests:
Calibration Value
Real Amps
NCD Sensor Amps
New Calculated Calibration Value
Difference in Amps
Comments
500
45.6
12488.56
1.826
273.872
Reset to 500 Cal value
500
45.5
12493.955
1.821
274.592
Reset to 500 Cal Value
28927
45.6
6530.432
201.988
143.211
Tried Default Value as Per Bhaskar 28927
28927
45.6
10065.344
131.051
220.731
Tried again, no cal, just reading. Sensor is reporting inconsistent data.
28927
45.5
1659.776
792.986
36.479
Tried again, no cal, just reading. Sensor is reporting inconsistent data.
28927
45.4
8865.28
148.138
195.270
Tried again, no cal, just reading. Sensor is reporting inconsistent data.
28927
45.5
6823.104
192.900
149.958
Tried again, no cal, just reading. Sensor is reporting inconsistent data.
28927
45.6
2990.336
441.111
65.578
Tried again, no cal, just reading. Sensor is reporting inconsistent data.
1.82
45.6
32.476
2.555
0.712
Tried to goto 1.82 as per initial 2 readings.
1.82
45.5
32.456
2.551
0.713
Consistent
1.82
45.6
32.605
2.545
0.715
Consistent
1.82
45.6
32.933
2.520
0.722
Consistent
2.55
45.7
57.599
2.023
1.260
Trying 2.55
2.55
45.6
57.716
2.015
1.266
Consistent
2.55
45.7
57.883
2.013
1.267
Consistent
2
45.7
57.777
1.582
1.264
Trying 2 Now as per last calculation
2
45.7
57.311
1.595
1.254
Didn’t change calibration
2.0
45.7
57.317
1.595
1.254
Trying 2.0 instead of just 2. Didn’t change anything
1.59
45.7
32.235
2.254164728
0.70536105
Trying 1.59
1.59
45.7
32.501
2.235715824
0.711181619
Now its back to 32 amps
1.8
45.5
32.287
2.536624648
0.709604396
Just trying 1.8. didn’t seem to work. Seems to be ignoring decimals
1
45.7
32.33
1.413547788
0.707439825
Testing to see if 1 (without decimal) reports 32 AMPS as well to confirm theory about decimals. Seems to be the case!
1
45.7
32.339
1.413154396
0.707636761
Seems to report 32 amps, and ignore decimals
3
45.7
81.92
1.673583984
1.792560175
Trying without decimal to test my theory
3
45.6
83.096
1.64628863
1.822280702
Consistent reading.
10
45.7
258.767
1.766067543
5.662297593
Trying 10 to scale. No decimals
10.9
45.7
253.292
1.966623502
5.54249453
Tryingt 10.9 to see if .9 makes a diference. It doesn’t/
I believe the issue is that the Calibration is IGNORING decimals. Based on my testing, the sensor is responding linearly when using whole number calibrations:
Cal Factor
NCD Sensor Output
Actual Amps
1
32.3
45.6
2
57.7
45.6
3
83.1
45.6
10
258.7
45.6
500
12488.56
45.6
Based on the data presented, it appears that:
The optimal Calibration Factor should be between 1.5 and 1.8.
However, the system seems to be ignoring decimal places in the Calibration Factor:
Any value between 1.0 and 1.9 is treated as if it were exactly 1.0.
When the value reaches 2.0, there’s a sudden jump in the sensor’s reported output.
This behavior suggests a limitation in how the Calibration Factor is being processed:
It’s likely that the software or Node-RED package is truncating the Calibration Factor to an integer.
This prevents the use of precise calibration values needed for accurate measurements.
Proposed solution:
The issue could potentially be resolved by modifying the software or Node-RED package.
The modification should allow for decimal places in the Calibration Factor, enabling more precise calibration.
What do you think? Could this be the case? If so, can NCD provide a fix by updating the temp branch?
