Post Go back to editing

Investigating Accuracy Drift at Negative Temperatures with ADE9000

Category: Hardware
Product Number: ADE9000

Hello,

We are currently working on an energy monitoring device which is implemented with an ADE9000.

During the qualification of the product under different temperature conditions, we experienced precision drifts at negative temperatures for V, I, and consequently P, as shown below:

injected -5°C Errors
V1 (V) I1 (A) PF P1 th (W) measured V1  (V) Measured Current I1 (A) measured Power P1 (W) absolute drift V1 (V) Relative Err V1   absolute drift I1 (mA) Relative Error I1  Absolute error P1 (W) Relative Power Error
230 10,00 1 2300,000 229,62 9,9907 2293,01 -0,38 -0,17% -9,3000 -0,09% -6,99 -0,30%
230 5,00 1 1150,000 229,62 4,9957 1146,59 -0,38 -0,17% -4,3000 -0,09% -3,41 -0,30%
230 2,50 1 575,000 229,63 2,4981 573,36 -0,37 -0,16% -1,9000 -0,08% -1,64 -0,29%
230 1,00 1 230,000 229,63 0,9992 229,34 -0,37 -0,16% -0,8000 -0,08% -0,66 -0,29%
230 0,50 1 115,000 229,63 0,4996 114,66 -0,37 -0,16% -0,4000 -0,08% -0,34 -0,30%
230 0,25 1 57,500 229,63 0,2499 57,34 -0,37 -0,16% -0,1000 -0,04% -0,16 -0,28%

-> we can see what looks like an offset in the voltage measurement, do you have any explanation for this? 

1- As a first step, we mounted 25 ppm resistors and then 10 ppm resistors for voltage channels, without improvement.

2- After that, we monitored the input voltage at the entry of the ADE9000 on channel V2, and we didn't get a real difference or drift after the resistor divider stage.

3- Also, we measured the internal 1.2V Vref, and we didn't see a large drift: Vref = 1.2556V @ 23°C and Vref = 1.2561V @ -5°C -> 11 ppm/°C.

Now we aim to integrate drift and gain compensation over temperature, as mentioned in the datasheet.

Is it possible to determine the drift profile of the ADE9000 in our hardware implementation?

Can we determine a general compensation factor based on experimental tests, rather than requiring calibration for each product?

Thanks,

  • Can I see the same data for 25C ? 

    If I assume the voltage channel was calibrated to 0% at 25C

    I see the errors on the voltage channel of 0.16% but power of 0.38%.  This is why I need the 25C data. 

    Can you share the voltage channel input circuitry? Are you using diodes for voltage clamping? If so, can you remove them, recalibrate and test over temp.

    What type of capacitor are they using in the anti-alias filter? Can you provide the part number?

    Dave

  • Hello Dave, 

    Below are the measurements at 23 °C and -5 °C, along with their variation:

    23°C -5°C Variation
    measured V1  (V) Measured Current I1 (A) measured Power P1 (W) measured V1  (V) Measured Current I1 (A) measured Power P1 (W) absolute variation V1 (V) absolute Variation I1 (mA) Absolute vraiation P1 (W)
    230,08 10,0054 2300,97 229,62 9,9907 2293,01 -0,46 -14,7000 -7,96
    230,08 5,0027 1150,48 229,62 4,9957 1146,59 -0,46 -7,0000 -3,89
    230,08 2,5015 575,27 229,63 2,4981 573,36 -0,45 -3,4000 -1,91
    230,07 1,0007 230,13 229,63 0,9992 229,34 -0,44 -1,5000 -0,79
    230,08 0,5005 115,09 229,63 0,4996 114,66 -0,45 -0,9000 -0,43
    230,07 0,2502 57,53 229,63 0,2499 57,34 -0,44 -0,3000 -0,19

    the voltage channel input circuitry:

    RV1: Varistor (PN: B72220S0461K101)
    L1:ferrite bead from Murata 
    R,R2,R3: with 25ppm/°C then tested with 10ppm/°C
    R4,R5: with 10ppm/°C
    C1,C2: 22nF 5% 50V X7R from Yageo (CC0603JRX7R9BB223)

    I hope I have provided the necessary information 

    Thanks,

  • Could you remove the MOV , ferrite and the 22nf caps and retest at 23c and -5c. If this see improvement add back each component to see which is the cause. 

    Seems to be the voltage channel as you indicated. 

    How is neutral tied to the system? from neutral to AGND.

    23 measured V1  (V) Measured Current I1 (A) measured Power P1 (W) V I W
    230 10 2300 230.08 10.005 2301 0.035% 0.054% 0.042%
    230 5 1150 230.08 5.003 1150.5 0.035% 0.054% 0.042%
    230 2.5 575 230.08 2.502 575.27 0.035% 0.060% 0.047%
    230 1 230 230.07 1.001 230.13 0.030% 0.070% 0.057%
    230 0.5 115 230.08 0.501 115.09 0.035% 0.100% 0.078%
    230 0.25 57.5 230.07 0.250 57.53 0.030% 0.080% 0.052%
    -5 measured V1  (V) Measured Current I1 (A) measured Power P1 (W) V I W
    230 10 2300 229.62 9.9907 2293 -0.165% -0.093% -0.304%
    230 5 1150 229.62 4.9957 1146.6 -0.165% -0.086% -0.297%
    230 2.5 575 229.63 2.4981 573.36 -0.161% -0.076% -0.285%
    230 1 230 229.63 0.9992 229.34 -0.161% -0.080% -0.287%
    230 0.5 115 229.63 0.4996 114.66 -0.161% -0.080% -0.296%
    230 0.25 57.5 229.63 0.2499 57.34 -0.161% -0.040% -0.278%

    Dave

  • Hello Dave,

    We have already measured the input voltage at -5 °C at the ADE9000 input (V1_ADE) on the “Voltage” channel, and we did not observe any difference or drift after the resistive divider stage.

    The neutral is connected to the system via two Murata ferrite beads, like ones mounted on the voltage channel.

    We’ll run some tests after removing the MOV, the ferrite bead, and the 22 nF capacitors, and I will keep you posted.

    If this doesn’t improve the measurements, could the issue be with the ADE itself, which might require drift and gain compensation over temperature, as indicated in the datasheet?

    Thanks,

  • Do you have any data on the test I recommended? 

    Dave

Before You Switch


Switching languages will make ADI Explorer unavailable. Resume your session by switching back to English and reopening ADI Explorer.