Measuring phase angle of 240Vac, transformer or isolation amplifier

I have been tasked with designing a circuit to measure the phase angle between the legs of a 3-phase 208V system.

• I will need to determine the phase angle with ± 2 degree resolution / accuracy, aiming for ± 1 degree
• The processor has an integral 12bit A/D with 0-3Vdc signal range, and can can easily sample and measure the discrete points every 20us
• Isolation will be required between the hazardous 208V system and the A/D
• This device must be designed to pass UL safety requirements, so the isolation device must be certified/recognized by UL

So I either use 3 isolation transformers or 3 isolation amplifiers. The isolation amplifiers seems to be in the $5-10 range, the transformers I find when searching Digi-Key tend to be about the same price but much larger and overkill in the VA capability.

Can anyone offer real world suggestions on using a transformer vs. an isolation amplifier?

Another even lower cost way might be with three opto isolators.

Minimum cost transformers usually end up at about the 5va size, anything smaller than that, and the price usually goes up due to increased difficulty of manufacture.

Isolation transformers can also supply dc power for the rest of your circuit which is very convenient.

Your biggest problem doing this to high accuracy, might be harmonic distortion on the three phases.
The usual "simple" method is to just measure and compare time intervals between the three zero crossings, and you don't need an a/d converter to do that.

I presume "phase angle" means angle of the fundamental and thus involves an analog measurement.

You might get off which cheap transformers if you only need an accurate phase angle measurement, relying on a certain degree of transformer matching. Generally I would prefer modern "digital" isolation amplifiers if galvanic isolation is a must.

In most cases, a low leakage current (< 1 mA) can be a tolerated and high ohmic differential voltage divider + differential amplifiers can be used.

You should also consider the technique used by energy meters: Analog front and digital signal processing residing on the high voltage side and an isolated serial interface to the safe low voltage.

If you are just looking at the phase of the voltage, I would use three opto couplers fed via resistors from each phase and an inverse diode across the LED. A capacitor should be put across each opto to make a low pass filter. each opto coupler output should then be squared with a high gain opamp then taken as pairs fed into an exclusive OR gate. This will then give a pulse output whose width is proportional to the time difference between its two inputs. If these pulses are then integrated they will give a mean voltage DC proportional to the time difference. These voltages can be read with a meter.
Frank

chuckey said:

then taken as pairs fed into an exclusive OR gate. This will then give a pulse output whose width is proportional to the time difference between its two inputs. If these pulses are then integrated they will give a mean voltage DC proportional to the time difference. These voltages can be read with a meter.
Frank

Click to expand...

In fact this exact method is used in many high class phase meters.
If the amplitude of the square waves can be accurately regulated to 3.6 volts, a normal digital voltmeter can very easily read the averaged and smoothed resultant dc, reading out directly in degrees.
For example example 1.200 volts = 120.0 degrees 2.400 volts 240.0 degrees.and so on.

Warpspeed said:

In fact this exact method is used in many high class phase meters.
If the amplitude of the square waves can be accurately regulated to 3.6 volts, a normal digital voltmeter can very easily read the averaged and smoothed resultant dc, reading out directly in degrees.
For example example 1.200 volts = 120.0 degrees 2.400 volts 240.0 degrees.and so on.

Click to expand...

Can you provide a link to the circuit your describing?

The digital method works as long as there are no significant harmonics that affect the zero-crossing point of the fundamental.
If there are harmonics then you may need to filter the waveform (either by analog or digital filtering) so that you are measuring the phase of only the fundamental.

Jester said:

Can you provide a link to the circuit your describing?

Click to expand...

I have two instruments right here that work on this principle, a Dranetz TM-101378A phase meter, and a Hewlett Packard 3575A gain - phase meter.

The complete service and repair manuals for both are available on the internet "somewhere".
I did print both out a very long time ago, both run to several hundred pages.
There is a description, block diagram, waveforms, and all the schematics.

Here is a link to the HP 3575A manual, theory of operation starts at page 34.
**broken link removed**

Cannot now find the Dranetz, but its very similar in concept.

Warspeed:

By browsing at your linked manual, brought me fond memories when HP was THE WORLD'S BEST instrumentation and electronics engineering company.
Just look at the manual....the amount of man-hours just to write and document it is astonishing!

Thank you. I used a Dranetz many years back (blue box 314 I think), that was a pretty decent unit. I will do a search for the schematics

These commercial instruments are made to operate over a very wide bandwidth, with huge variations of input amplitude.
They are necessarily complex to offer those features and performance.

Measuring mains voltage at only mains frequency, and at a pretty constant input level, is vastly easier to do.
So don't become discouraged by the complexity of these instruments.
The basic idea behind them is quite simple, and a few logic gates and an averaging filter, plus a digital voltmeter are all you really need.