Here's another approach that eliminates the transformer: On the "hot" side, use a voltage-to-frequency converter, like the AD7740. It can be powered from a shunt regulator like the TL431 in this circuit. The output from the V-to-F converter can then be fed through an optocoupler to the "safe" side for further analysis and display.
-Phil
This is an interesting approach. If I could power the MCP32xx, Vref regulator, voltage comparitor and a few opto-isolators for the SPI/I2C interface then all of the ADC can be done on the "mains" side and fully isolated from the Propeller side. The only limitation would be if all of the "mains" devices could be powered with 14 mA from the TL431. The key advantage is with this one setup I could measure all of the AC signals, Line, V and I without additional isolation requirements.
I think this is what started the whole isolation discussion: the neutral of the AC line needs to be common with the ground in this circuit in order for the proper voltage to be present on the output end. However, I have seen this capacitive coupling used on some flame sensing rectification circuits. All is well until the coupling capacitor shorts, but then again the same could be said about a transformer. Seems from the comments here that a capacitor short is more likely than an transformer failure.
Have you looked into the analog devices products incorporating their chipscale transformers? Came across this in a medical device. Magazine for isolation. Maybe they have something of use for you needs. They meet UL standards for isolation to 2.5kv or so. Google chipscale.
On the "hot" side, use a voltage-to-frequency converter, like the AD7740.
Wow, even Analog devices 'forget' stuff.
Those poor little BAS16's are rated at 4A max for 1us, and they forgot to put any spike limiting on their 0.44uF, which can dump
150 Amps in 1us into the BAS16..... (and just adding a 0805 resistor will not cut it either.)
Have you looked into the analog devices products incorporating their chipscale transformers? Came across this in a medical device. Magazine for isolation. Maybe they have something of use for you needs. They meet UL standards for isolation to 2.5kv or so. Google chipscale.
FF
I traced this to an Analog AD7400/AD7401 device. It looks like it could be a very handy solution. Voltage input range is ±200 mV - perfect for AC applications and it has a built-in Vref function and opto-couplers for data transmission. Real estate is 10x10 mm. There is only one channel per device and so it ends up being a bit pricey per measurement: $4.75 (Q=1000+) or $8.75 (Q=1). However, its simplicity and size make it worthy of serious consideration for a demo board
Comments
This is an interesting approach. If I could power the MCP32xx, Vref regulator, voltage comparitor and a few opto-isolators for the SPI/I2C interface then all of the ADC can be done on the "mains" side and fully isolated from the Propeller side. The only limitation would be if all of the "mains" devices could be powered with 14 mA from the TL431. The key advantage is with this one setup I could measure all of the AC signals, Line, V and I without additional isolation requirements.
True, I thought OP wanted to simply measure a nominal 120V 60Hz signal. I suppose it would depend on other variables.
If the load is unknown then use 2 hall sensors.
I think this is what started the whole isolation discussion: the neutral of the AC line needs to be common with the ground in this circuit in order for the proper voltage to be present on the output end. However, I have seen this capacitive coupling used on some flame sensing rectification circuits. All is well until the coupling capacitor shorts, but then again the same could be said about a transformer. Seems from the comments here that a capacitor short is more likely than an transformer failure.
FF
Wow, even Analog devices 'forget' stuff.
Those poor little BAS16's are rated at 4A max for 1us, and they forgot to put any spike limiting on their 0.44uF, which can dump
150 Amps in 1us into the BAS16..... (and just adding a 0805 resistor will not cut it either.)
I traced this to an Analog AD7400/AD7401 device. It looks like it could be a very handy solution. Voltage input range is ±200 mV - perfect for AC applications and it has a built-in Vref function and opto-couplers for data transmission. Real estate is 10x10 mm. There is only one channel per device and so it ends up being a bit pricey per measurement: $4.75 (Q=1000+) or $8.75 (Q=1). However, its simplicity and size make it worthy of serious consideration for a demo board
Thank you for the tip!.
http://www.mouser.com/search/Refine.aspx?Keyword=145768005&Ns=Pricing%7c0&FS=True&Ntk=P_MarCom
Interesting device, but only 10 bits, and a Icc spec of 13mA is high - I guess it is a two-die solution.