• State TI Thinks Resolved
  • Locked Locked
  • Replies 10 replies
  • Answers 2 answers
  • Subscribers 78 subscribers
  • Views 1075 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

MSP430F67791A: CT sample rate

dflohr
Expert 1140 points
Part Number: MSP430F67791A
Other Parts Discussed in Thread: MSP430F6779A, MSP430I2041, TINA-TI

James and others:

Interesting application possibility came up for using the MSP430F6779A meter design.  While I understand it would give up  accuracy, would likely be OK in the application.

Assuming that we always have the A/D inputs for voltage,  question is, can we bank switch in groups of CTs,  let it settle, sample, calculate, record and then disconnect the bank so that the next bank can be switched in?

Based on the switching time and sampling period, we could multiply the sample to extrapolate creating what we hope would be a close representation of the actual KWH used even with just partial data.

Possible to do with sub-second bank switching?  If so any guess on any of the timing  (bank switching frequency, sample period while the bank is switched in, etc)

Thanks

  

  • 0
    TI__Mastermind 49692 points

    Hi Dan,

    If your application doesn't require much accuracy or doesn't expect the current to vary quickly over time, then this may be possible. I'm not sure how you'd implement the switching, but perhaps you'd use a controllable XPYT switch or something similar. My concern would be the scenario where the current flow is drastically different than the interpolation/estimation and the error accumulates over time causing very inaccurate results or estimations. For the timing, you'd obviously want to split up the timing based on the number of banks. For example, if you have 3 banks, you'd want to split up one second into thirds for each bank. Out of those thirds, you'd have to do some testing with how those thirds would be measured - the more times you switch between banks, the more "dead time" where nothing is getting measured but perhaps less measurement inaccuracy because you're measuring the banks more in parallel than sequentially. I'd assume it'd need to be a balance.

    Instead, I would recommend using several MSP430i2041 devices running the Energy Measurement software library reporting to a host MCU or processor. They are less powerful than the MSP430F6779A but still support up to 4 SD ADC channels, are cheaper, and would make more sense than using multiple MSP430F6779A devices. This way, you're always measuring resulting in much higher accuracy.

  • 0 in reply to James Evans
    Expert 1140 points
    Hi James,

    For a typical US meter, 240 VAC split phase L1+L2+N we do need 2 voltage and 2 current ADC channels. While clearly supported by MSP430i2041, can a meter using this MSP430 be built and calibrated to ANSI C.12 specs for revenue grade accuracy ?

    Thanks

    Dan
  • 0 in reply to dflohr
    TI__Mastermind 49692 points

    Hi Dan,

    That's a good question. Most likely, it will depend on the level or Class of required accuracy. I've done some testing on a CT-based i20xx design for a different certification (not ANSI) and observed active energy accuracy across a temperature range from -40C to 85C. At -40C, the error was around 0.6% and at 85C, it was around -0.25% error - a total amount of approximately 0.85% error. Note that this was without temperature compensation. With compensation, the error could be reduced substantially. Other things that impact the overall system accuracy would be the temperature ratings of the passives and the CTs.

    At room temperature, I kept the voltage constant at 230V at 50Hz and observed an approximate error less than +/-0.2% for phase angles 0, +60, and -60 degrees across 0.01A to 100A. Thus, depending on the Class accuracy required, i20xx could possibly meet the requirements.

    Regards,

    James

  • 0 in reply to dflohr
    TI__Mastermind 49692 points

    Hi Dan,

    Sorry, I missed the other part of your question.

    dflohr said:
    For a typical US meter, 240 VAC split phase L1+L2+N we do need 2 voltage and 2 current ADC channels.

    Yes, typically in this configuration, 2 voltage and 2 current channels are required. If you have a load that never references neutral, you may be able to use 1 voltage channel and 1 current channel, since the currents in both L1 and L2 would be identical. Obviously, if there are multiple loads or sources, these numbers may change.

    Regards,

    James

  • 0 in reply to James Evans
    Expert 1140 points

    James:

    US meters are not connected to neutral. Just 240 VAC on the two legs that go through the meter. We are thinking  one ADC channel for 240 VAC and then 2 ADCs for the CT's
    While we could use just one CT with both L1 and L2 passing through it, we would lose the capability for net metering.  (If 1kw into the house on L1 and 1kW out of the house on L2, CT readings would cancel and we would just show 0KW which, while true,   would not be helpful )
    Can you confirm that we can configure the TI code  code for US meter operation  where, since many of the loads are 120VAC,  the current would only show up on one of the CT's and the voltage will be 1/2 of the 240 VAC across the meter?
    Examples:    Load  like 240 VAC A/C and water heater where the current on both CTs is the same  so   CT1 x 1/2  240VAC  + CT2 x 1/2 240VAC  essentially = KW   and 
                       Load like 120 VAC refrigerator  where current only on CT1   so CT1 x 1/2 240 VAC +  CT2 (0)  x 1/2 240 VAC  essentially = KW
    This correct thinking?
    Dan 
  • 0 in reply to dflohr
    TI__Mastermind 49692 points

    Hi Dan,

    Have you had a chance to access the ANSI specifications yet? I suspect that they will have plenty of detailed information to answer these types of questions.

    Specifically, ANSI seems to specify that meters meet Blondel's Theorem, which states that for an N-wire configuration, N-1 elements are required. According to this ANSI blog, it should be noted that Blondel’s Theorem is not strictly adhered to in all metering practices. For reference, non-Blondel metering installations are listed in Table 2A of ANSI C12.20-2015, and these explicitly are not covered by the standard.

    Split-phase can also be called single-phase, three-wire. So, two elements are adequate according to Blonde's Theorem. Each "element" is considered a separate watt-hour meter that measures both voltage and current. For the current measurements, two CTs can be used. If the Neutral connection is not available in the 2S meter base, the two Line voltages will have to be estimated by dividing L1 - L2 by 2. This is less accurate than actually measuring them individually with respect to Neutral, but it can be done as you described above.  However, this seems like a non-Blondel meter configuration.

    Looking at your example, it seems like this approach would work. If it was possible to measure the Neutral current, this would indicate when your 120V loads are operating, except when they are identical loads - the currents cancel so there would be no Neutral current. I don't think this scenario is likely but wanted to point it out. It gets more complicated when both 120V and 240V loads are operating, but Blondel's Theorem most likely accounts for this scenario.

    Regards,

    James

  • 0 in reply to James Evans
    Expert 1140 points
    James. Thanks Given that virtually all of the US smart meters and all of the 2S meter base meters do not have a neutral wire connection (> 70 million meters) , we are missing something here if these non-blondel meters are not covered by the ANSI standard. Something not right.

    On related note, can the TI energy software library manage , calculate and store data for the the US configuration where there will be some loads that are 240 V and others 120 V (and no neutral)?

    Thanks

    Dan
  • 0 in reply to dflohr
    Expert 1140 points
    James:

    This is really helpful. en.wikipedia.org/.../Blondel's_theorem. Talks about the 2S US meters.

    Can we set up the TI energy library and firmware to manage "The current measuring devices provide a measurement equal to one half of the actual current value."

    Thanks

    Dan
  • 0 in reply to dflohr
    TI__Mastermind 49692 points

    Hi Dan,

    Thanks for you patience. It's been busy lately, but I wanted to thoroughly dig into this configuration. Using our TINA-TI SPICE simulation software, I tried to validate the approach of the Blondel's theorem in a 2S meter with 1 voltage sensor (ADC channel) and 2 current sensors (ADC channels) for several scenarios. The error seems to be minimal (as shown in the tables below).

    The simulated currents don't have a polarity here, so I'm not sure yet how this would influence the polarities and configuration for EMDC. However, EMDC can accurately measure parameters such as voltage, current, power, and energy, so it will depend on how the application code translates the polarities from the EMDC results for this configuration.

    1. Unbalanced Loads

    Ampere Meter Current [A] 2S Supported Notes
    AM1 178.19 Yes CT1 current sensor
    AM2 186.68 Yes CT2 current sensor
    AM3 8.49 No Neutral current
    AM4 169.71 No 240V current
    AM5 8.49 No 120V branch 1 current
    AM6 16.97 No 120V branch 2 current
    Loads Resistance [Ω]
    R1 10
    R2 5
    R3 1
    Sources Voltage [V] Phase [°]
    VM1 120 0
    VM2 120 180
    Power Watt [W] Equations Notes
    Measured 43784.4 =((240/2)*B2)+((240/2)*B3) Using only CT1 and CT2
    Calculated 43785.6 =((240/2)*B6)+((240/2)*B7)+(240*B5) Using branch voltages and currents
    Error 0.0027406% =ABS(B19-B20)/B20

    2. Balanced 120V Loads

    Ampere Meter Current [A] 2S Supported Notes
    AM1 186.68 Yes CT1 current sensor
    AM2 186.68 Yes CT2 current sensor
    AM3 1.03E-14 No Neutral current
    AM4 169.71 No 240V current
    AM5 16.97 No 120V branch 1 current
    AM6 16.97 No 120V branch 2 current
    Loads Resistance [Ω]
    R1 5
    R2 5
    R3 1
    Sources Voltage [V] Phase [°]
    VM1 120 0
    VM2 120 180
    Power Watt [W] Equations Notes
    Measured 44803.2 =((240/2)*B2)+((240/2)*B3) Using only CT1 and CT2
    Calculated 44803.2 =((240/2)*B6)+((240/2)*B7)+(240*B5) Using branch voltages and currents
    Error 0.0000000% =ABS(B19-B20)/B20

    3. Balanced Loads

    Ampere Meter Current [A] 2S Supported Notes
    AM1 50.91 Yes CT1 current sensor
    AM2 50.91 Yes CT2 current sensor
    AM3 2.08E-15 No Neutral current
    AM4 33.94 No 240V current
    AM5 16.97 No 120V branch 1 current
    AM6 16.97 No 120V branch 2 current
    Loads Resistance [Ω]
    R1 5
    R2 5
    R3 5
    Sources Voltage [V] Phase [°]
    VM1 120 0
    VM2 120 180
    Power Watt [W] Equations Notes
    Measured 12218.4 =((240/2)*B2)+((240/2)*B3) Using only CT1 and CT2
    Calculated 12218.4 =((240/2)*B6)+((240/2)*B7)+(240*B5) Using branch voltages and currents
    Error 0.0000000% =ABS(B19-B20)/B20

    4. No 240V Load

    Ampere Meter Current [A] 2S Supported Notes
    AM1 8.49 Yes CT1 current sensor
    AM2 16.97 Yes CT2 current sensor
    AM3 8.49 No Neutral current
    AM4 1.70E-05 No 240V current
    AM5 8.49 No 120V branch 1 current
    AM6 16.97 No 120V branch 2 current
    Loads Resistance [Ω]
    R1 10
    R2 5
    R3 10M
    Sources Voltage [V] Phase [°]
    VM1 120 0
    VM2 120 180
    Power Watt [W] Equations Notes
    Measured 3055.2 =((240/2)*B2)+((240/2)*B3) Using only CT1 and CT2
    Calculated 3055.204073 =((240/2)*B6)+((240/2)*B7)+(240*B5) Using branch voltages and currents
    Error 0.0001333% =ABS(B19-B20)/B20

    Related Files

    Results.xlsx

    1_split_phase_1ph_3w_simulations_for_unbalanced_loads.TSC

    2_split_phase_1ph_3w_simulations_for_balanced_120V_loads.TSC

    3_split_phase_1ph_3w_simulations_for_balanced_loads.TSC

    4_split_phase_1ph_3w_simulations_for_no_240V_load.TSC

    I hope this helps. Please let me know if you have any questions.

    Regards,

    James

  • 0 in reply to dflohr
    TI__Mastermind 49692 points

    dflohr said:
    Can we set up the TI energy library and firmware to manage "The current measuring devices provide a measurement equal to one half of the actual current value."

    To achieve this, configure the external voltage divider to prevent the input voltage at 240V from exceeding the max Full-Scale Range input value. In EMDC, you can click on the voltage sensor, input the max voltage of 240V, select your resistors, and compare the calculated output voltage to this max input value. Even though there's only one voltage channel, you can still select this voltage channel for BOTH current channels under the "Library Configuration" > "Phases" tab in EMDC!

    Now, during calibration, you'll want to apply 240V to the board, but in order to halve the input voltage (and hence power) for this 2S configuration, you'd want to input "120" as the VRMS value in the "Calibration" > "Step 2. Test Setup Input Parameters" window in EMDC to adjust the scaling factors accordingly. Thus, when EMDC calculates the power, it's multiplying the measured current by 120V and not 240V.

    Regards,

    James

**Attention** This is a public forum