voltage sensing of 450 VAC 60 Hz to galvanically isolated ADC

[jordon]

I'd like some guidance on how to measure 450VAC, 60Hz, and pass this to a low voltage sensing circuit (approx. 3 V going to a high-impedence ADC). The two circuits (450V and sensing) must be galvanically isolated.

1. The brute force approach would be to use a 450:3V transformer. The ADC will simply grab the voltage coming out of the transformer secondary. However, I've found that this transformer is rather large. I'd like the transformer to fit on a PCB, if possible.

2. Another approach is to use voltage divider with a 1:1 transformer. See attachment. We size the voltage divider resistors such that 3V forms across the transformer. However, two problems. 1) The resistors dissipate about 6W of power. I have to measure nine (9) voltages, so this adds up (9x6W=54W). 2) The voltage divider is directly proportional to the DC resistance of the transformer, which is in turn influenced by the ambient temperature. In other words, the hotter the environment, the higher the resistance.

3. I want to explore other alternatives. I've heard about linear optocouples or high-impedence Op-Amps sitting on the 450 V line, and draw virtually no power, but Op-Amps would require a power supply hanging on the 450 V line as well, and linear optocouples may lose linearity near the peaks and zero-crosses.

Basically, the goal is to design a voltage sensing circuit that doesn't vary with temperature (error < 1% from 0 to 80 C), does not dissipate more than, say, 1-2 Watts, and has a footprint small enough to fit on a PCB, say 3x3 in.

What's the best way to do this?

 

[ScottyUK]

If you can arrange a suitably high input impedance from the ADC then you could use a capacitive divider at the front end. You would need to carefully select the capacitors for use on the AC line, typically either Class X2 or Class Y capacitors are used for a mains application. To achieve 1% error with a capacitive divider will probably require some calibration, although once set it should be pretty stable. Once the voltage is at a manageable level you can provide your isolation using the normal methods.


----------------------------------

If we learn from our mistakes I'm getting a great education!

 

[BrianG]

Is this a line-to-line measurement in a 3phase power system or just some other 60Hz voltage? If it's in a power system I would advise against any form of direct contact with a PCB at that kind of voltage doe to PCB track and connector clearance. I would mount any voltage reduction circuit off-board and make sure it's carefully current limited and over-voltage protected.

If this is part of a 'power' system I can't believe that an isolating step-down transformer can be THAT big. I found several rail-mounted transformers that can take 450V line-line inputs and give low voltage secondaries which can be attenuated to 3V if necessary. OK, they may not be PCB-mounted but this is much safer and more controllable than dealing with a 450V 'power' feed directly to the PCB.

 

[ScottyUK]

PCBs carrying rectified line voltage are a matter of daily routine in variable speed drives up to about 15kW rating. Nothing revolutionary needed to deal with it, just good design.

I do agree that it should be possible to find a small transformer capable of 450V operation though.


----------------------------------

If we learn from our mistakes I'm getting a great education!

 

[OperaHouse]

Generally, looking linearly at 450V doesn't yeild much informarion. I hope you will offset that so you are reading something like 350-500V.

You could self power something like a TL494 and transmit duty cycle % through an opto coupler.

 

[LionelHutz]

Why must they be galvanically isolated? It's fairly common in industrial controls (VFD's, soft-starters, some potection relays) to ground the board and then use voltage dividers for sensing the line voltages without any galvanic isolation.

 

[LiteYear]

We did something vaguely similar and had good experience with a 3 stage device: resistive divider, instrumentation amplifier, isolation amplifier. The resistive divider totals over 2Mohm and divides at about 100:1, so power is very low. The instrumentation amplifier is an INA121 in its 1:1 configuration and the isolation amplifier is an ISO124 with two DC power supplies derived from a common AC mains source.

This worked very well for us. Our main application was measuring DC, but the same could be (and was at one point) used for AC too. For your wide temperature range you might need to do some compensation or maybe you could get lucky and the resisters vary in proportion to each other.

We weren't particularly space confined so the whole unit was about 150mm x 200mm for 3 channels. This gave us loads of room for creepage and clearance but you could easily miniaturise it by a factor of two or three.