Question about digital clamp ammeter reading technique/philosophy/principle

[ArchieMedes]

Hi All,

I would like to ask about the principle on how a clamp ammeter reads the current. This is not about EMF / induction that is sensed by the ammeter. This is about the current digital reading shown by the ammeter.
I understand that there are some ammeter which uses averaging and RMS value of the current flowing. I am guessing this is based on the assumption that the AC current it is reading is a pure/perfect sinusoidal waveform with a frequency of, say, 50 or 60 Hz. So, I am guessing, it will just take ONE cycle of the AC current, computes its average or RMS value and show this value in the digital display. IS THIS THE CASE???
Now, if I have an AC current continuously flowing being measured by the clamp ammeter and it has perfect sinusoidal waveforms (no distortion whatsoever) BUT the current cycles are turning 1 cycle ON and 1 cycle OFF of the usual 50 or 60 Hz (in this case it is half or 25 or 30 Hz)...Will it still SAMPLE ONE cycle and measure its average or RMS value? In this case, it will be the same reading as the former''s case reading BUT I believe this is NOT ACCURATE anymore as the current reading SHOULD be effectively only HALF of the former's reading because it is turning ON and OFF at 1 cycle each.
Are there any clamp meters which can read the average value NOT of the single cycle of the AC waveform but the average of the effective current, say, for 1 second timeframe???

Please enlighten me on this. Sorry for the very long narration. Just want to be clear and concise on this question. Thanks in advance!

Archie

 

[gar]

120126-2357 EST

A few quick comments:

An average reading meter and an RMS meter both perform an averaging function, but the signal averaged is different.

For either method the measurement may or may not include the DC component of the waveform.

In general, unless the meter is special, the averaging period will be over a number of cycles of the waveform dependent upon the AC frequency and the design of the meter. This averaging time may be synchronized with an integral number of cycles, or the start and end times may be random relative to the signal being measured.

For a standard current transformer type of meter, clamp-on or otherwise, the DC component is automatically removed. Thus, the measurement is of a full wave average of the waveform stripped of DC, or the RMS value of the waveform with DC removed. The easy way to see what the waveform is without DC is to use capacitive coupling to an oscilloscope.

Many digital meter have an averaging time of 0.1 to 0.3 seconds. Not much different for most mechanical meter movements.

If you have had calculus then you may have been exposed to means to calculate average and RMS values of waveforms.

With this small amount of background ask further specific questions.

.

 

[ArchieMedes]

Gar,

Thanks for the reply. Based from your reply that the DC component is automatically removed from the meter reading, does it mean that a cycling current (like the attached) will give the same reading to the meter like the stable 50/60Hz system with continuous train of current cycle?
I am talking about the difference of a "snapshot" measurement and average of one cycle reading and one cycle off. Will the attached (with one cycle ON and one cycle OFF) should have only half of the value/reading compared to the continuous cycle like that of a standard 50/60Hz.
I would like to equate it to a kW and kWh reading but I don''t know if it is the case here.

 

[gar]

120127-0820 EST

First consider what happens if we use a resistive shunt to make the measurement, and disregard your switching transients.

A DC meter will read zero because the average is zero. A Simpson 260 in the AC position and assuming it could measure current would read ( 0.636 * I_peak / 2 ) * a scaling fudge factor. The scaling fudge factor is 0.707/0.636 and used to convert from average to RMS for a sine wave. The / 2 is because you eliminate 1/2 of the cycles on a sufficiently uniform and fast basis vs the averaging time constant of the meter movement.

An electrodynamometer type AC ammeter will read an RMS value of 1/2 the continuous sine wave because only 1/2 of the power is available. Same for a thermocouple type AC ammeter.

There is no DC component to your waveform so other than for phase shift and bandwidth problems a current transformer type of meter will perform the same as a resistive shunt meter.

Then the question is how does an electronic meter differ from the above meters? Probably not much different. Special meters can be made to do many different things.

Is this a theoretical question or do you have this actual waveform? If the dead spots get longer, and sufficiently long then the averaging time constant in the meter will become important. Much like what happens if you put an AC voltmeter on audio oscillator and gradually lower the frequency toward 0 Hz.

.

 

[gar]

120127-0909 EST

Continuing about kW and kWh.

Suppose the load is a pure resistance, then each small increment of i²*R*delta-t is the energy to the load for that increment. Where i is the average current for the very short time delta-t. Taken to the limit delta-t is zero.

So assume the sine wave is continuous and the one cycle period is 1/60 second and you measure a current of 1 A RMS to a 1 ohm resistor, then average power is 1 W. The energy to the load is 1 W times whatever time period is of interest where time is moderately long compared to one cycle. A 1 hour duration would be 1 Wh.

Remove 1/2 of the cycles, then energy used is 1/2 Wh, and average power is 1/2 W.

In your sketch you have 4 on cycles and 3 off. Thus, for this example the energy used is 4/7 of the energy as that for a continuous sine wave over a time of 7 cycles. Add one more off cycle and the value is 1/2.

.

 

[PetrosA]

You can't average one on/one off cycle without measuring the next cycle since you won't know how long the off cycle is till you measure the next on cycle. Every clamp meter will have a minimum frequency rating and minimum voltage requirement for that. My clamp needs a minimum 10 Hz signal at 30V to get a TRMS reading. Other clamps will be different. If your signal is within the range of your meter, you should get a reading within the specified accuracy of your meter.

These limitations are based on sampling rate. A clamp meter samples a constant number of times per second so depending on the frequency of the signal, it will be sampling different points on the waveform at each sample. The sampling rate of a clamp meter is MUCH lower than that of an oscilloscope, so it can't actually see a complete waveform. Instead, it builds an image of the waveform over a large number of samples. Some clamp meters have a peak function which will sample faster (1 ms) but it's still only looking for a peak value. If you've ever tried measuring inrush current to a motor with a clamp meter, you'll have noticed that each time you do it you get a different reading. That's because the motor may start at a different point on the waveform each time it starts which will affect how the current peaks as it saturates the winding.

It sounds to me that if you need the ability to analyze single waveforms you are going to need an oscilloscope. If you just need an accurate current reading of the type of waveform you illustrated, look for a clamp with a specified range that covers your conditions (frequency and voltage level) and you should be fine. I don't think I've seen a pulsed sine wave signal before, but I imagine that a clamp meter designed to measure PWM systems should be able to handle it and may also have a duty cycle function to tell you how long the off pulse lasts.

 

[ArchieMedes]

Hi All,

Thanks for all the reply. This is actually true and not just a theoretical question. The resultant waveform is an output of an SCR which has a special trigger which detects zero voltage crossing and turns on and off per cycle depending on how much frequency it needs. From the example above, the SCR is being triggered to give half of the frequency.
The problem is different types of clamp ammeter that I have used, gives different readings. I am confused as to how the ammeter measures the current and why it gives different readings. One particular clamp ammeter even gives the same reading as the normal frequency (60Hz). It didn''t display half of the intended current (load resistance is fix) which I believe should be the correct case. It seems it just got a snapshot on ONE of the cycle and measures its RMS or average and did not consider the time it was OFF on the next cycle.
I am leaning towards the sampling rate explanation but the problem is still not solve. How can it be measured accurately? Can you please give a recommendation of a digital clamp ammeter that can give the most accurate reading on this type of waveform with ON and OFF cycling or varied frequency? I am not sure if it can be dealt as a PWM as the waveform is still a perfect sine wave unlike the PWM with square waves.

 

[gar]

120127-1147 EST

ArchieMedes:

The one that reads half for your waveform is correct. I would expect a Fluke 27 or 87 to read about 1/2. If you measure voltage across the resistive load with a Simpson 260 I would expect 1/2.

If you have two off cycles per one full on cycle, then I expect 1/3. By the time you get to 15 off cycles per one on cycle, then you are probably in trouble with the above instruments.

Whether the measurement technique uses sampling or not is not what causes a problem.


If you could use either a Kill-A-Watt or a TED 1000 instrument you could work with much greater ratios than 15 to 1 in energy measurement. I have not tried the experiment, but this comment is based on my understanding of how these instruments work in an energy mode. Print out a copy of the Cirrus CS5461A manual and study it. Their normal suggested sampling rate is 4000 per second or 66.7 samples per cycle. If you adjusted the input clock frequency from 4.096 MHz you could make an integral number of samples per cycle. But if you make the averaging time 1 second instead of 1 cycle, then you get an integral number of samples, 4000, per output measurement. The TED 1000 system uses this particular chip.

.

 

[Speedskater]

I suspect that the waveform in post #3 is just a snap-shot of a continuous current.

You could double check the old fashioned way.

Use a power resistor with an attached thermocouple as the load, allow the temp. to stabilize. Then using a normal AC or DC valuable power supply adjust the voltage to get the same temperature. Then read the current or calculate the current by measuring the voltage and resistance.

 

[Speedskater]

In the thread "power analysis...... " ( http://forums.mikeholt.com/showthread.php?t=142550 )
I wrote about how
On the hi-fi forums, I often read complaints about their equipment having transformer humming caused by DC on the AC power line. Really it might be harmonic distortion on the power line causing DC asymmetry or a DC offset.
and then
Or sometimes these hi-fi discussions use the primitive example of DC offset caused by a hair-dryer switching to low power by inserting a diode is series with the heating element and half-wave rectifying the current.

This is exactly that kind of nasty primitive device. I hope that you keep it securely locked in the laboratory. In any case, I don't think that it could pass a EMC compliance test.

 

[PetrosA]

Hi All,

The problem is different types of clamp ammeter that I have used, gives different readings. I am confused as to how the ammeter measures the current and why it gives different readings. One particular clamp ammeter even gives the same reading as the normal frequency (60Hz). It didn''t display half of the intended current (load resistance is fix) which I believe should be the correct case. It seems it just got a snapshot on ONE of the cycle and measures its RMS or average and did not consider the time it was OFF on the next cycle.
I am leaning towards the sampling rate explanation but the problem is still not solve. How can it be measured accurately? Can you please give a recommendation of a digital clamp ammeter that can give the most accurate reading on this type of waveform with ON and OFF cycling or varied frequency? I am not sure if it can be dealt as a PWM as the waveform is still a perfect sine wave unlike the PWM with square waves.

A few thoughts:

If you are not changing the frequency of the sine wave (ex. 60 Hz), then you still have a 60 Hz sine, regardless of the on and off cycles. I don't know if eliminating every other cycle would be considered a 50% duty cycle or not, as it would with a square wave.

Out of curiosity, which clamp meters have you tried, and what were the results? Did any of them come close to what you calculate should be the correct reading?

If you'd like I can forward this thread on to a few people I know at Agilent and Extech to see if their engineers can shed any light on the matter. They may want to contact you directly.

 

[ron]

I think this is how some lighting dimmers work. They chop the sin wave so it is only on for part of the cycle and the average light http://www.caseydiers.com/sinewavedimming.htm

 

[Speedskater]

I think that most light dimmers use triacs that work on both half's of the sine way. What they do is delay the turn-on point until almost the peak of the sine.

 

[gar]

120127-1716 EST

What ArchieMedes's equipment does is to gate one or more contiguous full sine wave cycles of X amount of current with 0 or more full cycle periods of zero current. This results in a modulation of the energy to the load from maximum to some lesser value.

Since the load is a resistance of apparently constant value there is no need to measure current. Simply measure the voltage across the resistor. To measure energy to the load the averaging time constant has to be long enough to avoid excessive measurement ripple.

If ripple is a problem, then an energy instrument is needed such as I mentioned above.

There will be other frequencies generated because of the on-off modulation of the sine wave. Is this a problem probably not.

.

 

[Speedskater]

Darn, your correct. Don't know what I was thinking about in post #10
Sorry

 

[ArchieMedes]

120127-1716 EST

What ArchieMedes's equipment does is to gate one or more contiguous full sine wave cycles of X amount of current with 0 or more full cycle periods of zero current. This results in a modulation of the energy to the load from maximum to some lesser value.

Since the load is a resistance of apparently constant value there is no need to measure current. Simply measure the voltage across the resistor. To measure energy to the load the averaging time constant has to be long enough to avoid excessive measurement ripple.

If ripple is a problem, then an energy instrument is needed such as I mentioned above.

There will be other frequencies generated because of the on-off modulation of the sine wave. Is this a problem probably not.

.

You got it spot on...
I did not get the part where I dont have to measure the current and just measure the voltage. Because this is SCR turning ON and OFF, I am sure the voltage also has the same waveform as the current (also cycling ON and OFF) because the SCR is switching ON and OFF or open and close. Probably I would also get different readings of voltage value as my the first problem with the current measurement reading accuracy.
With regards to the frequencies generated, you are also correct on this. Sometimes harmonics can become high enough to affect other equipment such as UPS and small generators..I know as I have seen it in action.The UPS and generators are "dancing" to the tune of the SCR tune.

 

[ArchieMedes]

I think this is how some lighting dimmers work. They chop the sin wave so it is only on for part of the cycle and the average light http://www.caseydiers.com/sinewavedimming.htm

Light dimmers uses phase angle which chops the sine wave form so a distorted sine waves are the result but in this case a zero voltage crossing detection makes sure every sine wave output is clean or whole.

 

[ArchieMedes]

A few thoughts:

If you are not changing the frequency of the sine wave (ex. 60 Hz), then you still have a 60 Hz sine, regardless of the on and off cycles. I don't know if eliminating every other cycle would be considered a 50% duty cycle or not, as it would with a square wave.

Out of curiosity, which clamp meters have you tried, and what were the results? Did any of them come close to what you calculate should be the correct reading?

If you'd like I can forward this thread on to a few people I know at Agilent and Extech to see if their engineers can shed any light on the matter. They may want to contact you directly.

I am sure that the frequency has changed as I have used a voltmeter with a frequency function and the readings of the frequency are changed with respect to the SCR triggering demand.
i have tried well known brands like fluke but I cant remember the make/model, what I remember is it doesnt have the RMS or true RMS function.
If you can forward this to those people who can shed light on this matter please do. I would appreciate it. They can drop me a PM here. I would really like to close this issue with our client. I appreciate every reply here. Thanks...

 

[ArchieMedes]

120127-1147 EST

ArchieMedes:

The one that reads half for your waveform is correct. I would expect a Fluke 27 or 87 to read about 1/2. If you measure voltage across the resistive load with a Simpson 260 I would expect 1/2.

If you have two off cycles per one full on cycle, then I expect 1/3. By the time you get to 15 off cycles per one on cycle, then you are probably in trouble with the above instruments.

Whether the measurement technique uses sampling or not is not what causes a problem.


If you could use either a Kill-A-Watt or a TED 1000 instrument you could work with much greater ratios than 15 to 1 in energy measurement. I have not tried the experiment, but this comment is based on my understanding of how these instruments work in an energy mode. Print out a copy of the Cirrus CS5461A manual and study it. Their normal suggested sampling rate is 4000 per second or 66.7 samples per cycle. If you adjusted the input clock frequency from 4.096 MHz you could make an integral number of samples per cycle. But if you make the averaging time 1 second instead of 1 cycle, then you get an integral number of samples, 4000, per output measurement. The TED 1000 system uses this particular chip.

.

Can you recommend a clamp ammeter? It has to be clamp as the current is high probably 400A. It has to be an ammeter because we would like to see the current being controlled as the SCR switches or cycles on and off. I have checked the fluke you recommended and these are all not capable of measuring current.

 

[gar]

120127-2351 EST

ArchieMedes:

You have stated that your load is a resistance. Thus, the voltage across the load is equivalent to measuring the current. Just a scaling factor. I = V/R. The SCR controlling the current with a resistive load has nothing to do with whether you determine the current my measuring the voltage across the load or you directly measure the current. Measuring the voltage across a resistance as a shunt in series with the load or across the load is probably better than using a current transformer.

What a frequency measurement might tell you is the relation of no current to current, but it might not work well based on how the frequency meter worked. If positive crossings in a given time period are counted then it could work well.

The advantage of using a voltmeter to measure current is that greater overload capability exists. Also for a current probe that is rugged you could use a Hall device probe and it can work from DC to high frequencies with less phase shift problems.

I do not see any particular need to prefer average reading or RMS in your application.

What is the maximum on duration, how many full cycles if more than one, and maximum off duration in full cycles?

From any information you have provided so far you never half cycle, there is no phase shift control of the turn on time within a cycle (turn on is only at a positive zero crossing). Automatically SCRs or Triacs turn off at a zero crossing, and your implication is that this always occurs on a positive zero crossing.

.