# detecting on off state of any appliance using a CT sensor

Status
Not open for further replies.

#### chrischris

##### Member
Hi,
I am trying to detect any appliance being turned off or on. I am using SCT013-30 (30A/1V) sensor with 62 ohm burden resistor built in. If I measure the AC voltage with just the sensor (not connected to my circuit, I get 30mV for off and 340mV for on.
As soon as I connect the circuit to PIC or even connect it to a resistor divider as shown here , the voltage goes to some random value, even up to 16V AC. and -30mV DC. I think it's the noise from the my 12V power adaptor that's effecting it, is there any way I can actually measure that chnage from 30mV AC to 340mV AC and feed it to a PIC ADC ?
I have even tried using LM324 and BC546 to amplify but as soon as I connect CT sensor output to my circuit, it goes all funny!!

p.s. I can't use a battery to drive my circuit I have to use a 12V power adaptor so no way around to the electrical noise. Also the CT sensor has to be non invasive type(or any other sensor for that matter that can serve the purpose).

Thanks,
Chris

#### alec_t

##### Well-Known Member
I get 30mV for off and 340mV for on.
Off should give you 0V, so you must be picking up a lot of interference. Stray magnetic fields near the CT, or long/straggly wires leading from the CT to the measurement point may be the cause.

#### chrischris

##### Member
Thanks for reply, my concern is not the 30mV offset, I want to amplify the 340mV I get from CT sensor and that's where the problem is. As soon as I connect the output from CT sensor to any circuit, the output from CT sensor becomes very unstable and meaningless, it's just completely gets off the charts. Even if I have an OPAMP or BJT amplifier on a breadboard, as soon as I connect my +5V and GND from 7805 which is fed by 12V DC adaptor, line noise is everywhere and it kills the signal from CT sensor..that's what I think happens unless I am missing something. Can anybody point me to right direction ? Is it possible to achieve what I am trying to do without using a CT sensor ?

Ta,
Chris

#### MikeMl

##### Well-Known Member
I see Tip, Ring, and Sleeve on the output of the sensor. Looking it up on-line, I cannot see how to hook-up the three connections. Are you sure you have it right?

As a guess, I would start with Sleeve as the circuit common, and Tip as the output. How did you hook it up?

Do you need to back-feed DC power to the sensor, or is it purely powered from the current on the burden side? If so, I do not see how they can make it very linear.

Last edited:

#### Mosaic

##### Well-Known Member
Thanks for reply, my concern is not the 30mV offset, I want to amplify the 340mV I get from CT sensor and that's where the problem is. As soon as I connect the output from CT sensor to any circuit, the output from CT sensor becomes very unstable and meaningless, it's just completely gets off the charts. Even if I have an OPAMP or BJT amplifier on a breadboard, as soon as I connect my +5V and GND from 7805 which is fed by 12V DC adaptor, line noise is everywhere and it kills the signal from CT sensor..that's what I think happens unless I am missing something. Can anybody point me to right direction ? Is it possible to achieve what I am trying to do without using a CT sensor ?

Ta,
Chris
Make your own CT sensor with a toroid and magnet wire. Rectify the output and include a burden resistor to suit the output level and improve low impedance capability. Smooth the level with a 100uF lo ESR capacitor and you have your repeatable sensor voltages for pulse & AC currents.

Not sure what your current(s) are but I have used these little sensors in dozens of applications. There are a few versions but the CR2550 will turn on down around 3/4 amp AC. They can also be hacked by removing the LED and driving an opto-coupler or fabricate your own opto-coupler using their LED. I have seen them for under $10 USD. Ron #### MaxHeadRoom78 ##### Well-Known Member Most Helpful Member There are many ACS712 modules on ebay for a couple of$.
Also Honeywell make the CSDA line, both non-invasive, although the ASC712 has power input terminals for Hall effect measurement .
Max.

#### MrAl

##### Well-Known Member
Hi,
I am trying to detect any appliance being turned off or on. I am using SCT013-30 (30A/1V) sensor with 62 ohm burden resistor built in. If I measure the AC voltage with just the sensor (not connected to my circuit, I get 30mV for off and 340mV for on.
As soon as I connect the circuit to PIC or even connect it to a resistor divider as shown here , the voltage goes to some random value, even up to 16V AC. and -30mV DC. I think it's the noise from the my 12V power adaptor that's effecting it, is there any way I can actually measure that chnage from 30mV AC to 340mV AC and feed it to a PIC ADC ?
I have even tried using LM324 and BC546 to amplify but as soon as I connect CT sensor output to my circuit, it goes all funny!!

p.s. I can't use a battery to drive my circuit I have to use a 12V power adaptor so no way around to the electrical noise. Also the CT sensor has to be non invasive type(or any other sensor for that matter that can serve the purpose).

Thanks,
Chris

Hi,

One thing to think about is that every device that is plugged into the AC mains has SOME path back to the AC mains from every terminal on that device. That path may be of low impedance or very high impedance, but it's always present. This has various implications depending on the impedances being high or low relative to other impedances.

The first possibility is that both the current transformer and the wall wart have enough leakage to cause a small current to flow right through them both. This of course causes problems. To find out if this is the case, a simple test should help.
After you connect things the way you normally did and start seeing the wild behavior, unplug the appliance from the wall. Do not use a switch or power strip switch or wall switch as they can have leakage too. Once unplugged, see if the wild behavior goes away. Note everything else is still connected, but now the appliance power is completely disconnected from the line. If the problem goes awaythen it could be an isolation issue with the current transformer or the wall wart. Obviously you should try another wall wart next.

It could also be that the 10k resistors are too high. Lowering them to 1k each might help, but that does use more power from the 5v supply.
What this 'fix' does is forces any leakage to appear to be lower than it was before to the rest of the circuit, and so might help.

It could also be something silly like an open ground connection. If using a plug board get rid of that and solder everything.

#### chrischris

##### Member
There are many ACS712 modules on ebay for a couple of $. Also Honeywell make the CSDA line, both non-invasive, although the ASC712 has power input terminals for Hall effect measurement . Max. Thanks for reply but ACS712 is not non invasive, the CSDA seems good but the current capacity is very low, only 1A. (which is on the edge of what I am trying to do) Not sure what your current(s) are but I have used these little sensors in dozens of applications. There are a few versions but the CR2550 will turn on down around 3/4 amp AC. They can also be hacked by removing the LED and driving an opto-coupler or fabricate your own opto-coupler using their LED. I have seen them for under$10 USD.
Thanks ron, those are really nice , but costs me $116 just to get them shipped to New Zealand. It could also be that the 10k resistors are too high. Lowering them to 1k each might help, but that does use more power from the 5v supply. What this 'fix' does is forces any leakage to appear to be lower than it was before to the rest of the circuit, and so might help. It could also be something silly like an open ground connection. If using a plug board get rid of that and solder everything. Will give it a try. All components are soldered on but I will check the ground connection again. It seems I will end up spending about$100 for a better sensor in any case, unless I managed to get the one working that I have. If I use a battery the sensor works perfectly fine so it is defo a line noise issue. What is the best way to isolate the current sensor ground from the rest of the DC ground ? may be that will solve the problem ?

Thanks all

#### Daniel Wood

##### Member
I would recommend against amplifying the signal from the CT, in most cases, it isn't really nessecary and will introduce unwanted noise.
Instead, just adjust the burden resistance to get enough voltage out, so that the full range of the ADC is used.

If you're not worried about true RMS in your measurements, you can couple your CT in a different way.
When I did a similar project, I basically clamped between 0 and VCC with a rail to rail op amp to make it "microcontroller-safe".
Then took a ~20mS (1/mains_freq) sample with the ADC, collecting the peak value.

It worked rather well, I got something around ±15 to ±20 mA precision over a 30A range.

#### Attachments

• 29.6 KB Views: 44

#### Pommie

##### Well-Known Member
The CR2550 are available on ebay and should work out cheaper.

Edit, forgot to mention that the negative cycle could be being attenuated by input pin protection diodes.

Mike.

#### chrischris

##### Member
just adjust the burden resistance to get enough voltage out
I thought of that but the CT sensor has built in 62 ohm resistor. Anything in parallel will just decrease the resistance. My problem is the grounding I think, as soon as I connect one end of the CT sensor to GND , and measure AC voltage with meter, the value shoots up to 14 to 16 V AC. I will check my ground connection again and see if I can find any issue with that.

The CR2550 are available on ebay and should work out cheaper.
Checked ebay, its about $36 , don't know how much it will be with shipping though, will try. Thanks to both of you , appreciate the effort and insight ! #### Daniel Wood ##### Member I think your burden resistor is the source of most your problems. (30A/1V) coupled with a (10-bit?) ADC won't get you the precision you need. Assuming VRef = 5V. (30A*5V)/(2^10-1) = 146mA per division Put in perspective, if your CT is coupled to ground, a reading of only 7 on your ADRES register (34mV on your pin) is equivalent of 1A flowing through your CT. This may be why noise is such a big issue, a 10mV ripple from the power supply will ruin your analogue signal. Is it possible to open the thing up and remove/replace it? EDIT: If your wire is quite thin, you can loop it around the CT multiple times.. If you manage to get 10 turns, you've essentially got 10 times more voltage being generated. Last edited: #### Reloadron ##### Well-Known Member Most Helpful Member Thanks ron, those are really nice , but costs me$116 just to get them shipped to New Zealand.
Wow, I never would have thought a simple \$10 USD could cost so much to get it New Zealand. My guess would be that other companies are marketing these things, possibly someone down under in your area.

Here is a very simple roll your own design which seems to work quite well. There were several versions.

Ron

Last edited:

##### Well-Known Member
the CSDA seems good but the current capacity is very low, only 1A. (which is on the edge of what I am trying to do)

l
Honeywell data show the series is available up to 1275 AMPS!!
I have used them for presence of plasma arc detection.
Max.

#### alec_t

##### Well-Known Member
Assuming VRef = 5V.
(30A*5V)/(2^10-1) = 146mA per division
I must be interpreting the CT spec differently from you. I read "30A/1V" as meaning "30A will give a 1V output across the burden resistor".

#### Daniel Wood

##### Member
I must be interpreting the CT spec differently from you.
That's how I've interpreted it too? I'll try and break my understanding down.. correct me if I'm wrong.

If 5V/1023 = 4.88mV per division,

To work out the amount of current needed to generate to register a 1-bit change, with a 30A/1V rating its...
4.88m*30 = 146.5mA per division

To prove this, we can compare the coupling factor before and after, to see if any of the calculations got fudged.
1V/30A = 0.0333_
4.88mV/146.5mA ≈ 0.0333_

Last edited:

#### ronsimpson

##### Well-Known Member
Instead, just adjust the burden resistance to get enough voltage out, so that the full range of the ADC is used.
I did a study with that last month. I found that my CTs saturated with out a resistor. The core can only handle a certain amount of VT. (volt x seconds) It works at 0.1V for (1/60hz) just fine. It will saturate long before 5V and 8mS. I know that SPICE makes it work because SPICE does not think about core saturation.

ChrisCrhis,
Please draw a schematic of what you are doing. There is something simple wrong. Include the 3 wires from the CT.

Last edited:

#### dr pepper

##### Well-Known Member
I know you cant use a bettery, however try getting the circuit to work on a battery first, then you know its good & you can start to address the noise problems.

#### chrischris

##### Member
ChrisCrhis,
Please draw a schematic of what you are doing. There is something simple wrong. Include the 3 wires from the CT.
Not sure what do you mean by three wires, there are only two from CT sensor

I know you cant use a bettery, however try getting the circuit to work on a battery first, then you know its good & you can start to address the noise problems.
I have tried a similar circuit with battery and it works so this circuit will work with battery I know.

It seem I just have to bite the bullet and spend some money on a better sensor. But I would like to get to the bottom of this. Most of the projects I find with such sensors monitors more than 10A or so , while my appliance takes only 1.2A , this much small current generates very little voltage and that's my problem. I thought this would be thee easiest part of the project !!

Thanks everyone. I will have to wait until weekend to get back at this and investigate further.

EDIT: Can anyone tell me what happens to this http://www.farnell.com/datasheets/1...59.1387307022.1494922120-577118603.1494920332 sensor if current is more than 1A ? And how would I wire this up ? There is no connection diagram in the datasheet or on google.
P.S. 1A are the only ones available on element14 NewZealand

Last edited:
Status
Not open for further replies.