Current sensor 15A SCT-013

Vicky_

@robert-hh Hello, thanks for your reply! sorry for late reply because I was trying to digest your information. I have read up a bit about "goertzel-Algorithm" but Im not very understand the coding and the diagram. Why adc bits=9? Can I get the DC voltage using that coding? by diving the value obtained 512 and times with 3.3. like this (x/512)x3.3

I also confused with the connection of the current sensor. can I use back my own connection because this can work with Arduino.
https://learn.openenergymonitor.org/electricity-monitoring/ct-sensors/interface-with-arduino?redirected=true
just that i dont have the Emon Lib.

Sorry if I ask silly questions! I really appreciate your help!

robert-hh

@vicky_ I used adc=9 bits, because thats the useful resolution for the ESP32 ADC. You may also go for 12 bits, but all you will get is more noise, not more precision.
Yes, you can use also your circuit, as long as you connect it to 3.3V and not to 5V. The two circuits are effectively the same. The reason I prefer mine is, that one side of the sensor output is connected to GND. But that is more a flavour than technical need. I strongly reccomend the protection circuit. The sensor is a low impedance source, and saturates at +/- 10V, which, when applied to the ESP32, will most likely kill your LoPy.
Bot the sensor and the software work with AC only. The sensor if an AC transformer. With DC the output is 0. The software returns the amplitude of a fixed frequency in the input signal. Since the sensor is used for mains current, typical frequencies are 50 or 60 Hz.
maybe a note for using the sensor: You must feed only a single wire though the sensor. Loads are connected typicall with at least two wires. Only one of them must be passed through the sensor. Otherwise, the output signal of the sensor is 0.

Vicky_

@robert-hh Thanks for your clear explanation! I will try it and see how.
Yea, I only feed one single wire which is the live wire. Thanks for your remind. Thank you

Vicky_

@robert-hh Hello. The coding you shared, it works! Thanks a lot.
I use it to measure a hairdryer rated 1000w to 1200w.
when hairdryer is off I get 0.something.
when I open level 1 I get 10. something.
when I open level 2 I get 45. something.
when I open level 3 I get 71. something.
I chose the maximum value 71.something and divide 512 (9 bits ADC) times with 3.3v and 15A.

eg:
((71.22/512)*3.3)*15 = 6.86A

but the maximum current should be 1200/240v =5A, right?

or my calculation is wrong?

I used it to measure a phone charger but it has no different with when it is off, maybe the current is too small.

once again, thanks

alt text

robert-hh

@vicky_ The transfer ratio is hard to guess, because there are some unknown figures in between, like the transfer function of your sensor. That is 1Veff at 15A. So at 5 A it should deliver 0.333Veff. But i did never determine the remainder of the transfer function.
What you should do is to take the current of the hairdryer with an known good Multimeter and determine a factor from that. The ESP32 ADC is non-linear, especially at 11 dB attenuation. But the mid range values should be OK. Using 6 dB attenuation the linearity is better. But then you'll need a different interface circuit.
Yes, at low currents the ADC noise and the phase noise of the sampling steps in. If the time for reading a value is larger, you'll have somewhat less noise. This time should be a integer multiple of the wave period. For 50 Hz, a multiple of 20 is fine, for 60 Hz a multiple of 50. So taking a multiple of 100 is good for both frequencies.

Vicky_

@robert-hh Thanks for your suggestion! I understand what you trying to say until this part "If the time for reading a value is larger, you'll have somewhat less noise. This time should be a integer multiple of the wave period. For 50 Hz, a multiple of 20 is fine, for 60 Hz a multiple of 50. So taking a multiple of 100 is good for both frequencies."

do you mean that at the coding the time I should modify it
time1 = time x 100 and all time change it to time1?
sorry.

robert-hh

@vicky_ in my example, the time for reading is 1000 = 1000ms. If that is too slow, you may take any multiple of 100, like 100, 200, 300, .... instead. Besides that, you can freely select the total measuring time.

Vicky_

@robert-hh ok thank you so much!

StefanoF

@robert-hh

Hi,
sorry Robert, because this schema is different of schema di Vicky? ,,
can you kindly let me know which schema I need used?

Many thanks

robert-hh

@StefanoF They are identical in that they shift up the signal to be measured by Vdd/2. The difference:

In my schematic the sensor is directly tied to GND, here it is tied to a virtual GND at Vdd/2 with a impedance at 50 Hz of about 30 Ohm.
I added in my schematics overvoltage protection, which is missing here.

As long as there is no excess current you are try to measure, both will work. In case of a shortage in the sensed circuit, there is a high risk that the connected Pycom device will be damaged.

P.S.: To get a result for measuring, only C1, R1 and R2 are needed in my circuit. Everything else is for protection.

robert-hh

@StefanoF The schema in your post and the one of Vicky are identical, just differently sketched.

StefanoF

@robert-hh Thanks for your suggestion!

StefanoF

@robert-hh

StefanoF 2 days ago
Hi ,
I've used this the electronic schema , I've in out 1.65 volt at empty, the sensor is at 15A (STC 013-015)
Can you kindly help me with easy code,

apin = adc.channel(pin='P13', attn=ADC.ATTN_11DB)

peakVoltage = float(0)
ACCurrtntValue = float(0)
for i in range(5):
    peakVoltage += apin.voltage()
    utime.sleep_ms(1)
peakVoltage = (peakVoltage / 5)
voltageVirtualValue = (peakVoltage / 1000)
print(round(voltageVirtualValue,2))

result 1.65

Many thanks

robert-hh

This post is deleted!

robert-hh

@StefanoF You code snippet calculates the average DC voltage at the input. But what you need is the amplitude of the 50 Hz (or 60 Hz) AC voltage. So you have to do some kind of frequency analysis for that frequency. The most simple algorithm for doing that is the so called goertzel algorithm. If you scroll back in this thread, you will find an example which I posted there. I used that and it works fine, except for very small currents. But that is not the problem of the algorithm, that caused by the noise of the ADC capture, both amplitude and phase noise.

StefanoF

@robert-hh

Hi Robert,
I've used this my code e variabile ma è valido:
Thanks you

def readACCurrentValue():
    global RANGE_AMPS, ADJUST_VOLTAGE
    peakVoltage = float(0)
    ACCurrtntValue = float(0)
    for i in range(5):
        peakVoltage += apin.voltage()
        utime.sleep_ms(1)
    peakVoltage = (peakVoltage / 5) 
    voltageVirtualValue = round(peakVoltage / 1000,2) - 1.65   
    if(math.fabs(voltageVirtualValue)>0.1):
        ACCurrtntValue = (math.fabs(voltageVirtualValue) * 15)   #15 A
    return round(ACCurrtntValue,2)

robert-hh

@StefanoF I think your code is not useful. It does not determine the value of the 50HZ AC component. Instead it calculate the average value of a 5ms time window somewhere in the 20ms period of a 50 Hz wave. For a given constant current you will get a largely varying output value, if you repeat the call several time non-synchronous with the mains frequency. And even is the value is in the synchronous case seemingly constant, it may be totally wrong.
I tried similar approaches, e.g. finding the highest and lowest value and then calculating the difference, like a Vpp determination. But that was not robust at all. The only robust approach was using a FFT approach, like the mentioned Goertzel algorithm.

StefanoF

@robert-hh

Not exist a driver for this component , ex. EmonLib.h of arduino ?

Many thanks

robert-hh

@StefanoF said in Current sensor 15A SCT-013:

EmonLib.h, Emonlib.cpp

I do not know a python implementation for that algorithm. As far as I understand is, if calculates the integral over the zero compensated voltage square. That can be done and may work for strong signals. I expect it to fail for small signals.

robert-hh

@StefanoF Here is another code, which uses plain vanilla calc instead of the goertzel algorithm. That is more robust to cumulating rounding errors. I have my doubts about that for goertzel. This also tries to cater for sampling jitter.

#
import gc
import math
import array
from utime import ticks_us, ticks_ms, ticks_diff, sleep_ms
from machine import ADC, Timer, idle, enable_irq, disable_irq

#
# acquire ADC values. The paramters of the constructor are:
# 1. the mains frequency (number, typically 50 or 60)
# 2. the time (ms) for one test period. time must be an integer
#    multiple of frequency period duration and the sampling period
#    being n * 50ms for 60Hz and n * 20ms for 50 Hz. A good value is 200
# 3. the sampling period (ms). time must be an integer multiple
#    of the sampling period. The default is 2.
#
class Acquire:
    def __init__(self, freq=50, *, time=200, sample_rate=2):
        self.sample_rate = sample_rate
        self.samples = time // sample_rate
        self.adc = ADC(bits=12)
        self.freq = freq

        self.data = array.array("f", [0.0 for _ in range(self.samples)])
        pi_f = (2 * math.pi * freq * self.sample_rate) / 1000
        self.timestamp = array.array("L", [0 for _ in range(self.samples)])

    def start(self, pin):
        self.pin_adc = self.adc.channel(pin=pin, attn=ADC.ATTN_11DB)
        self.count = 0
        self.busy = True
        self.alarm = Timer.Alarm(self.read_adc, 0, ms=self.sample_rate, periodic=True)

    def stop(self):
        self.alarm.cancel()

    def read_adc(self, alarm):
        self.data[self.count] = self.pin_adc.value()
        self.timestamp[self.count] = ticks_us()
        self.count += 1
        if self.count >= self.samples:
            self.alarm.cancel()
            self.busy = False

    def result(self):
        while self.busy == True:
            sleep_ms(self.sample_rate)
        # normalize the time stamps
        for _ in range(1, self.count):
            self.timestamp[_] = ticks_diff(self.timestamp[_], self.timestamp[0])
        self.timestamp[0] = 0

        vsin = 0.0
        vcos = 0.0
        pi_f = (2 * math.pi * self.freq) / 1000000
        for _ in range(self.count):
            vsin += self.data[_] * math.sin(self.timestamp[_] * pi_f)
            vcos += self.data[_] * math.cos(self.timestamp[_] * pi_f)
        amplitude = math.sqrt(vsin * vsin + vcos * vcos) / self.count
        gc.collect() # avoids automatic gc during sampling
        return amplitude

    def reading(self, pin):
        self.start(pin)
        return self.result()

def run():

    pin = "P13"
    freq = 50
    meas_time = 1000
    acq = Acquire(freq, time=meas_time)

    try:
        for _ in range(100):
            value = acq.reading(pin)
            print(value)
    except:
        pass
    acq.stop()
    print("Handler stopped")

run()

Explore Pybytes | Official Documentation | Report a Firmware Bug/Issue | GitHub

Current sensor 15A SCT-013

Pycom on Twitter