Hall Effect Sensor to Measure Low Frequency Current

[LMAO]

I am using a Hall Effect Sensor (clamp) to measure motor locked rotor current fed by variable frequency drive. Obviously, as soon as the drive turns on it goes to current limit mode and the result is a large current (about 2000A) at 1 to 2Hz.
For various reasons, I am having difficulties to measure a stable current and what I read fluctuates by about 200A. Below are the reasons I can think of:

1. VFD PWM which is about 1.5kHz.
2. Shaft being locked which results in VFD error signal being fed back to PID control resulting in fluctuating current: basically, drive ramps up voltage because speed feedback never gets above zero but then decreases voltage because the current limit is exceeded and so on...
3. Interference from other phases.

I can only do something about number 3. I am thinking of placing a conductive sheet between the phase I am not measuring and current clamp. what do you all think? Dumb, genius, or ...?

 

[GoldDigger]

I am using a Hall Effect Sensor (clamp) to measure motor locked rotor current fed by variable frequency drive. Obviously, as soon as the drive turns on it goes to current limit mode and the result is a large current (about 2000A) at 1 to 2Hz.
For various reasons, I am having difficulties to measure a stable current and what I read fluctuates by about 200A. Below are the reasons I can think of:

1. VFD PWM which is about 1.5kHz.
2. Shaft being locked which results in VFD error signal being fed back to PID control resulting in fluctuating current: basically, drive ramps up voltage because speed feedback never gets above zero but then decreases voltage because the current limit is exceeded and so on...
3. Interference from other phases.

I can only do something about number 3. I am thinking of placing a conductive sheet between the phase I am not measuring and current clamp. what do you all think? Dumb, genius, or ...?

If the Hall effect sensor is not adequately shielded from electric fields, you may have a problem that can be reduced by the kind of conductive shield that you mention. There is no good way to shield against magnetic field effects except to make sure that the clamp is closed completely and that the wire and the clamp are not moving around.

The simplest test you can do is look at the signal from the current clamp when it is placed next to one or more of the energized wires but not actually surrounding any of them. Any reading you get will be electric field related noise.


A more complicated test, which may also give you a much better idea what is going on, would to to look at the clamp output on a scope rather than just reading the meter display. That will show you the results of PWM and also show you whether the VFD frequency is stable or ramping up and down too.

 

[LMAO]

If the Hall effect sensor is not adequately shielded from electric fields, you may have a problem that can be reduced by the kind of conductive shield that you mention. There is no good way to shield against magnetic field effects except to make sure that the clamp is closed completely and that the wire and the clamp are not moving around.

The simplest test you can do is look at the signal from the current clamp when it is placed next to one or more of the energized wires but not actually surrounding any of them. Any reading you get will be electric field related noise.


A more complicated test, which may also give you a much better idea what is going on, would to to look at the clamp output on a scope rather than just reading the meter display. That will show you the results of PWM and also show you whether the VFD frequency is stable or ramping up and down too.

Cables ARE vibrating violently because of the high current (+2000A) so as you said, that may be the major problem. I'll see if I can secure them to get better reading.

And yes, I AM using a scope to read the current.

 

[Besoeker]

I am using a Hall Effect Sensor (clamp) to measure motor locked rotor current fed by variable frequency drive. Obviously, as soon as the drive turns on it goes to current limit mode and the result is a large current (about 2000A) at 1 to 2Hz.
For various reasons, I am having difficulties to measure a stable current and what I read fluctuates by about 200A. Below are the reasons I can think of:

1. VFD PWM which is about 1.5kHz.
2. Shaft being locked which results in VFD error signal being fed back to PID control resulting in fluctuating current: basically, drive ramps up voltage because speed feedback never gets above zero but then decreases voltage because the current limit is exceeded and so on...
3. Interference from other phases.

I can only do something about number 3. I am thinking of placing a conductive sheet between the phase I am not measuring and current clamp. what do you all think? Dumb, genius, or ...?

At 1 to 2Hz the Hall sensor probably follows it. It is seen as a slowly changing current rather than a fixed low frequency magnitude.

 

[gar]

131106-0721 EST

LMAO:

Can you provide screen shots of several of your locked rotor current plots vs time?

You could have both electric and magnetic field coupling to your Hall sensor and its cable. Also turn-on phase angle of the cable you are looking at could be a factor, but we have no idea what your waveform looks like or what you mean by your statement about current variation.

If there is magnetic field coupling, then you could try some mu metal shielding or other high permeability material, or move the cable being sensed further away from the other cables. Search mu metal on the Internet.

Try your Hall sensor near, but not enclosing, the cables of interest, and then further away. Mentioned by GoldDigger. But you will also see effects from stray magnetic fields, not just electric fields. With my Fluke YB100 on the 20 A range and a DC meter on the 200 mV range I can see the effects of the earth's magnetic field.

.

 

[Besoeker]

131106-0721 EST

LMAO:

Can you provide screen shots of several of your locked rotor current plots vs time?

You could have both electric and magnetic field coupling to your Hall sensor and its cable. Also turn-on phase angle of the cable you are looking at could be a factor, but we have no idea what your waveform looks like or what you mean by your statement about current variation.

If there is magnetic field coupling, then you could try some mu metal shielding or other high permeability material, or move the cable being sensed further away from the other cables. Search mu metal on the Internet.

Try your Hall sensor near, but not enclosing, the cables of interest, and then further away. Mentioned by GoldDigger. But you will also see effects from stray magnetic fields, not just electric fields. With my Fluke YB100 on the 20 A range and a DC meter on the 200 mV range I can see the effects of the earth's magnetic field.

.

I don't think it's position. I think it's frequency.
Suppose, for example, it was 0.001Hz. You would expect the current to follow the current as it varies over a single cycle.

 

[Besoeker]

I don't think it's position. I think it's frequency.
Suppose, for example, it was 0.001Hz. You would expect the current to follow the current as it varies over a single cycle.

I don't think it's position. I think it's frequency.
Suppose, for example, it was 0.001Hz. You would expect the reading to follow the current as it varies over a single cycle is what I meant to post.......

 

[gar]

131106-0855 EST

Besoeker:

There seems to be the question of whether or not there is noise, an unwanted signal, coupling into the Hall current probe. My comment on positioning of the Hall sensor without enclosing the wire is to estimate what may be noise signals.

A different aspect is that, if when the Hall sensor is clamped around a cable and there is suspect of noise coupling from the other cables, then by moving the wire being measured and its Hall sensor further away from the other cables it may be possible to reduce noise coupling.

But the first test of seeing what signal is coupled into the Hall device when it does not enclose a wire and is put in various positions is probably the key to where to look for the problem.

.

 

[Besoeker]

131106-0855 EST

Besoeker:
But the first test of seeing what signal is coupled into the Hall device when it does not enclose a wire and is put in various positions is probably the key to where to look for the problem.

Gar
I don't know that there is a problem other than it being the wrong tool for the job.
If the update interval on the Hall device is comparable to the period of the waveform being measured I don't think it will read reliably.
The Hall instruments we use have an output for an oscilloscope. The oscilloscopes we mostly use have the capability of making (and displaying) numerical measurements from the input data - things like RMS, frequency, pk-pk.

Or we can send the data down a link to a spreadsheet and do the calculations there. Either would take away the likely difficulty with low frequency measurements.

 

[gar]

131106-0939 EST

Besoeker:

A plain Hall sensor has a response from DC to some reasonably high frequency, possibly a MHz. My Fluke has a moderately flat response from DC to 200 Hz +/-2%, to 1 kHz at lesser accuracy and current dependent, magnetic material being a factor.

Thus, no problems at low frequencies, only zero stability.

With my Fluke I can balance with the zero adjustment to a few millivolts on the 20 A range, or about 10 to 100 mA equivalent. This is mostly a pot sensitivity problem. After a short warm up time the zero stays within about a 0.2 mV range for at least 1 minute. Meter resolution is 0.01 mV.

.

 

[Besoeker]

131106-0939 EST

Besoeker:

A plain Hall sensor has a response from DC to some reasonably high frequency, possibly a MHz. My Fluke has a moderately flat response from DC to 200 Hz +/-2%, to 1 kHz at lesser accuracy and current dependent, magnetic material being a factor.

Thus, no problems at low frequencies, only zero stability.
.

OK. Lets go back to my extreme example of 0.001Hz sinewave and assume an RMS value of 1000A for that. Thus a peak of 1414A.
In the first 250 seconds, the current will go from 0A to 1414A as a quarter sinewave.

How do you think the Hall transducer will respond to that? What will the readout be?
Don't you think it will follow the current with a constantly changing output?

You won't get a constant 1000Arms. Yes, the sensor will sense the correct current from DC to some reasonably high frequency . But how does it get processed to read correct RMS at such low frequencies?
And that's where I think the problem lies.
As I said, the wrong tool for the job.

 

[gar]

131106-1241 EST

Besoeker:

Where was there a reference to an RMS measurement?

If there is a sub-cycle RMS measurement made, then there has to be some information on the start and end phase angles. If one measures only a very few cycles for an RMS value this will be necessary.

If one uses an unsynchronized RMS measurement of a number of cycles, then the averaging time constant has to be long compared to one cycle to reduce the variance in the reading.

.

 

[Besoeker]

131106-1241 EST

Besoeker:

Where was there a reference to an RMS measurement?

What other measurement value would you use for locked rotor current?

 

[gar]

131106-1316 EST

Besoeker:

Peak when I look at the envelop of a pulsating waveform or when I lack controlled phase information. LMAO said he was using an oscilloscope, and he made no mention of an RMS converter between the Hall sensor and the scope.

.

 

[Besoeker]

131106-1316 EST

LMAO said he was using an oscilloscope, and he made no mention of an RMS converter between the Hall sensor and the scope.

Then perhaps he can kindly tell us where the numerical readings came from.

 

[LMAO]

Then perhaps he can kindly tell us where the numerical readings came from.

Just finished my test. Current magnitude was about 2300A at 1 to 1.5 Hz. I connected the Hall Effect clamp to a scope and used very high time division scale (500ms per division) to get plenty of cycles per screen. The more cycles per screen you have (aka the higher time per division is) the more stable the RMS reading is. For every measurement, you just have to wait a few seconds until waveform fills up the screen.

 

[gar]

131106-1426 EST

LMAO:

Where is the converter from the instantaneous signal to some RMS value?

If you have 10 divisions per screen in the time axis, and 500 mS/div, then one screen scan is 5 seconds. At 1 Hz there are 5 cycles per screen width. At 1.5 Hz there are 7.5 cycles. If you are using a scope that includes an RMS calculation function that is performed over one screen width or some other period, then with the sweep rate and frequencies you are using you can be integrating over non-integral periods of the waveform and certainly can expect random variations in the result. If this is how you are getting the measurement, then can you force the integrating period to start and stop on a positive zero crossing?

Do you understand what RMS means and how it is calculated from a waveform? This knowledge is critical to your understanding what a measured number means.

.

 

[LMAO]

131106-1426 EST

LMAO:

Where is the converter from the instantaneous signal to some RMS value?

If you have 10 divisions per screen in the time axis, and 500 mS/div, then one screen scan is 5 seconds. At 1 Hz there are 5 cycles per screen width. At 1.5 Hz there are 7.5 cycles. If you are using a scope that includes an RMS calculation function that is performed over one screen width or some other period, then with the sweep rate and frequencies you are using you can be integrating over non-integral periods of the waveform and certainly can expect random variations in the result. If this is how you are getting the measurement, then can you force the integrating period to start and stop on a positive zero crossing?

Do you understand what RMS means and how it is calculated from a waveform? This knowledge is critical to your understanding what a measured number means.

.

You have a point but I don't think this "non-uniform-integrating" you are worried about is a significant problem. If scope is integrating 8 cycle of a 2 Hz signal, 100mS of a cycle cannot make whole lot of difference.
Edit: also, I am guessing scope uses multiple screen readings to calculate and average RMS.

 

[gar]

131106-1636 EST

LMAO:

Now you are at 2 Hz, and what were the sweep speeds you used prior to the 500 mS?.

Have you looked at the instantaneous waveforms of several samples from several trials. What kind of differences are you seeing?

Can you provide plots of several measurements that show this large variation of your first post?

.

 

[Besoeker]

You have a point but I don't think this "non-uniform-integrating" you are worried about is a significant problem. If scope is integrating 8 cycle of a 2 Hz signal, 100mS of a cycle cannot make whole lot of difference.
Edit: also, I am guessing scope uses multiple screen readings to calculate and average RMS.

If the scope is displaying the output from the Hall effect device as a sinewave, can't you just take the peak and divide by sqrt(2) to get RMS?