Hi!
I have a project where I need to have a isolated detector to distinguish ON/OFF state (with a detection current of 10mA) in a 10A/230VAC 50Hz circuit. The isolated part makes it much harder of course.
It interfaces with a Atttiny with 12b ADC and analog comparator.
My option list:
-
ALLEGRO CT110 Hall Sensor, consumes 1mA at 3.3V, even at the scale of +-5A the difference from 0 to 10mA is only about 3mV.
- Shunt resistor + isolated amp-op - More complex than hall sensor, consumes more energy
- Shunt resistor + optocoupler - impossible to achieve dynamic range required (10mA to 10A)
- CT + some passives - lowest power consumption, more board space
Am I forgeting something? Is there anything specific to this just ON/OFF detection application?
I had a similar problem detecting when the OFF Peak supply was available (for some specific reasons detecting voltage wouldn't work), I used a mains capacitor as a load and a CT with a load resistor, you can adjust the load resistor to set your sensitivity level quite easily, it has been very reliable, switching every day.
Because only detection is required, you can let the CT core saturate at as low as double your threshold current. This considerably reduces the core size and thus board area required, and also the secondary current that you have to clamp when 10A is passing through the primary wire.
To start a device when another one was switched on I put two diodes in antiparallel and a transformer followed by a full bridge, capacitor and relay.
I have a project where I need to have a isolated detector to distinguish ON/OFF state (with a detection current of 10mA) in a 10A/230VAC 50Hz circuit.
...
- Shunt resistor + optocoupler - impossible to achieve dynamic range required (10mA to 10A)
It's true the simplest circuit will not manage 10mA-10A, but you need only a few active parts to improve that, and opto will be smaller ands cheaper than any CT solution.
How fast does this need to react ?
The simplest active-smarts on sense-side design would only look at one half of the AC, and send low duty pulses thru opto coupler when I > threshold.
The MCU looks for a break in pulses > (eg) 1.5 mains cycles.
Any decent micropower opamps/comparator can manage that, and modern optos and ICs allow low enough operate current that bulky/hot parts can be avoided.
You can drop the sense resistor loss, by getting lower offset amplifiers.
An old fashioned way is to wind some wires around a reed relay, but getting it to switch at 10mA could be a challenge. You could modify an RCD. It can both detect 10mA, and handle 10A.
There is a company, Melexis, that makes low current hall sensors. You can just run the wire past one or a few turns and calibrate it. If you have a constant voltage that'd maybe work OK.
(I just got these Winston and Allegra parts this Friday arvo to have a play with hall sensors myself but these won't generate much current at 10mA. I'm going to try with my 32bit dac to make a current characterisation box so I'd be better off with the Mexlexis part myself).
An old fashioned way is to wind some wires around a reed relay, but getting it to switch at 10mA could be a challenge. You could modify an RCD. It can both detect 10mA, and handle 10A.
In my experience this works fine with dc but will chatter with ac. That is why I resort to using diodes as I explained above.
I have a project where I need to have a isolated detector to distinguish ON/OFF state (with a detection current of 10mA) in a 10A/230VAC 50Hz circuit.
...
- Shunt resistor + optocoupler - impossible to achieve dynamic range required (10mA to 10A)
It's true the simplest circuit will not manage 10mA-10A, but you need only a few active parts to improve that, and opto will be smaller ands cheaper than any CT solution.
How fast does this need to react ?
The simplest active-smarts on sense-side design would only look at one half of the AC, and send low duty pulses thru opto coupler when I > threshold.
The MCU looks for a break in pulses > (eg) 1.5 mains cycles.
Any decent micropower opamps/comparator can manage that, and modern optos and ICs allow low enough operate current that bulky/hot parts can be avoided.
You can drop the sense resistor loss, by getting lower offset amplifiers.
Thanks! Do you have any reference or example?
For the CT solution I found
https://www.coilcraft.com/en-us/products/transformers/power-transformers/current-sensing/cu8965/I dont see how I can power any logic at the AC side I first need to create a voltage drop, as this is one conductor only measurement.
For the CT solution I found https://www.coilcraft.com/en-us/products/transformers/power-transformers/current-sensing/cu8965/
I dont see how I can power any logic at the AC side I first need to create a voltage drop, as this is one conductor only measurement.
The CT example says :
Designed for use between 16 kHz to 1 MHz and aboveHmm... If you cannot extract power, you can use isolated power modules, or even isolated ADC;s that have power included.
eg AD9112 series or MAX22531 etc
Those push up the price, so if you have singe-wire sense, the simplest will be a pass-thru current transformer.
The CT example says : Designed for use between 16 kHz to 1 MHz and above
Yes, it saturates at 50Hz, so the maximum voltage you can extract on the secondary size is very small (~0.1V) and it's only good for detection.
I did find this post, where OP wanted 10mA sense resolutions (linked part has good data sheet)
https://www.eevblog.com/forum/projects/current-transformers-at-milliamps-reasonable/msg825111/#msg825111
and these parts seem to have wide dynamic ranges and high precision and low drift at 50Hz, tho are larger.
I'm fine if the transformer is saturated at 10mA, while in that post he needs it to be still be pretty linear at 180mA so he can measure. I only want to distinguish betweeh 1mA and 10mA, all the rest can be a dead zone.