Power mesure ac or dc signal

Hi. You seem to be pretty close to a good circuit here. A couple of things:

Hall effect sensors measure current. Are you saying you're using a hall effect sensor to measure the voltage? I'm not clear on this.

Power _is_ volts times amps. The difference in phase for non-resistive loads can be a problem, though. Your basic idea is a good way of solving this. At any instant, the true power is the instantaneous volts times the instantaneous amps. If you apply to a four-quadrant multiplier a voltage proportional to the instantaneous load voltage and a voltage proportional to the instantaneous load current, the multiplier output voltage will be proportional to the instantaneous real power being used by the load. If the reactive load is feeding current back into the line, it will show on the multiplier output at that instant as a negative voltage corresponding to an instantaneous negative power. No need for RMS-to_DC converters here, for AC or DC.

Some multiplier-based power meters will use a transformer to measure the load voltage. That does have a low frequency cutoff, I'm afraid. But there are other ways to do this, depending on how you're using your signal. You can just divide down the load voltage, for example. This can be of value if you're only using an isolated meter or displaying voltage on a display. You do have to be careful with line potential, but this should be doable if you're careful. If you consider this, be careful to limit the maximum voltage while minimizing the signal distortion. If you're planning on doing this to measure line based loads, you should definitely use an isolation transformer during development and testing. Be safe.

The AD632 works with 1% accuracy from 50KHz down to DC (the 10Hz part has to do with noise). I'm wondering why you need a 10Hz low pass filter. What do you want to do with the signal? Are you trying to measure average real power? As long as that's all you're doing, lose the 10Hz low pass, set your DVM on a DC voltage range higher than the maximum output voltage (say, the 20V range) and just attach the meter. You will be delighted.

Feel free to post again if you have any questions.

Good luck Chris

Hi Chris,

thanks for you answer!

My system must read the real power that a big Auxiliary converter take from a hv generator.The converter can use 3000Vcc and 1500V 50 Hz and

1000v 16 2/3 Hz.

So,for monitorig tension and current, with an DC panel, both ac and DC source generator i convert with AD637(true rms to DC) Ac or DC .And this part of circuit goes well. The tension trasducer is an Hall effect ABB transducer wich can take up to 5000v AC or DC and give 20 mA for end of scale.

Now,the other problem is reading average real power.

I multiple the V and I with AD632 .......but i must integrate.So i tought to filter the out of multiplexer with an active low pass filter with a 10 Hz(very low) Fs and read his output on digital panel meter in DC. If the supply is DC i think that there aren't poroble also without the filter........bat when the supply is AC???

You say that without filter the aout of multiplier is the real power that load take also if is supply with AC?

The value of multiplier should be about 2 or 3 volt max when load takes max power.

What do you think?

By,

MAX

Hi, Max. It sounds like you're really on the right track. Considering the voltages involved, I'm sure you're going the extra step to be safe.

It sounds like you're not positively sure about real power and apparent power.

Real power is the average of the instantaneous voltage-current product over a line cycle. Apparent power is just the product of the rms voltage and rms current. Period. You're going about this the right way by not just multiplying the outputs of your RMS-to-DC converters. It sounded like that was what you were possibly doing on your first post. If you want an analog representation of real power, you have to directly multiply the direct outputs of your transducers, and then average that over the line cycle. Don't believe me -- look it up.

Now you may have some difficulties with the digital panel meter. Most all of them are based on an ICL-7107 or ICL-7135 IC, which does dual slope integrating D-to-A conversion. Here's a reference that might give you an intro:

Now you're going to have some difficulty using a DPM by itself, because they're usually adjusted to achieve maximum noise rejection at a particular line frequency. Your 50Hz rejection will be good, but your

16-2/3Hz regection maybe not so much. You're going to have to carefully evaluate, and make compromises between noise reduction and response time. Either that, or you may also be able to have one of these meters custom-modified for the lower frequency. It's just math and component selection -- the chip can handle it. The 7107/7135 IC is capable of slower conversion rates, and has inherently great line frequency reject at that one particular frequency.

But I'm wondering, for example, if it's necessary to have a 10 Hz f(c). Possibly a lower cutoff frequency might be more useful. After all, no one is capable of visually absorbing 10 readings per second, anyway. Of course, since you're talking about a total project accuracy of a couple of percent, you should remember that an analog meter does the averaging automatically. You also don't have to worry about power. Some taut band analog meters with mirrored backing can get you pretty close, too. But if you use an op amp filter, watch for offset and drift. Get premium op amps that will stay put.

When doing this project, I would think your biggest difficulty will be offset and drift in the multiplier itself. This is going to be a particular problem if your circuitry is in an industrial environment with major changes in temperature. Five percent accuracy plus annoying offset is easy. Getting it down below two percent, and getting the offset at a minimum is going to take some looking at to do well. Something you may want to look at is periodically doing an offset null like a low frequency chopper. Many ways to get this better than a couple of percent accuracy, if that's your need.

It sounds like you've got a genuinely interesting project on your hands. Take the time to read up on things. I remember doing one of these years back with an old Motorola analog multiplier. I had a lot of fun, and learned quite a bit about practical instrumentation tricks in the process. You've got a much better IC that will try to make you look good. The old Motorola IC I used needed half a dozen tweaker pots and seemed like a spirited but rather willful pony that always wanted to drift off and roll in the grass when it felt like it (sorry, Mr. Thompson). This is the type of project that can lead you to real professional growth.

Good luck, and be as safe as you can Chris

Hi Chris!

Now i've configured the AD632 as datasheet for scale factor 1 with vso to gnd.In this condition i can't trim the offset. The problem now is that with a signal in input of multiplier below 1 volt the ad give 9 volt!! if i put signal up 1,5 volt it work well.Why? Is it probably needing offset trimming?

Another thing.When a multiplie the 2 sine wawe the frequency is the double,and this is correct.So the the line cycle at the out of multiplier can be 100 hz or 32 Hz.

For having the average of I*V on the cycle is necessary to filter?Yes the panel meter use the ICL7107.Do you think that if i must filter i can use the AD797 op amp? Last question:

if i put directly the out of miltiplier to DPM when i use sine wawe what i read? The real power? With the DC i think yes.......but AC??

thanks again.

Ciao,

MAX

Hi, Max. This IC is super easy to use for your application. If you're using the Fig. 6 circuit, you have to remember to tie off the unused inputs. The X and Y inputs are differential. So, if you're just multiplying two voltages, ground X2 and Y2, then try again. If this doesn't work, check your wiring. It's a 14-pin IC, there are 4 NCs, two power and 5 grounded. All you've got is X1 * Y1 = OUT. Hooked up this way, there's no wiggle room at all -- you'll have a DC offset, but it's gotta work, or the chip was smoked (ouch -- these are expensive). This is a basic cut here. This is so simple, if you have any electronics experience at all, you have to get this. Check your wiring, or try a new chip. I'll bet you've got it hooked up wrong.

I think it's a mistake to use unity scale factor, because that will limit the input amplitude of your signals. You should look toward the circuit in Fig. 5 as giving you better accuracy. For instance, in Fig.

6, your input signals would be limited to a product of 10V^2. That would mean you'd have to scale each of your inputs smaller because of power supply limitations. The chip is made to a scale factor of 10, which will mean your input signals can be a product of 100V^2. This will provide more accuracy.

As mentioned before, to get real power, you have to average the instantaneous product. You use filtering to get the average. I would look at using an active low pass filter (using a fast low offset op amp to minimize error) between the multiplier output and your DPM. As mentioned in a prior post, I'd try to make the filter f(c) as low as possible to get the DPM input as close to pure DC as possible. Do you need help with this?

You might want to go back and refresh your knowledge. Let's say you have a purely reactive load -- say you're driving a capacitor with AC. There will be reactive power, but no real power. If you look at the output of your multiplier, you will see a waveform that goes + and -. However, the average voltage will be zero. The DC component of the AC waveform at the multiplier output will be zero. That means that if you use a perfect low pass filter, its output will be 0.000V.

Ideally, a DC meter reading this AC waveform from the multiplier output will see no DC, and will read 0.00V. However, unless the signal happens to be at the one particular frequency (or a multiple of that frequency) that the -7107 circuit was optimized to reject, you'll probably see the reading of the DPM move around -- you won't get a stable signal.

your head. Usually that's not a problem (after all, this is a newbie/hobbyist newsgroup), but when you're dealing with several KV, that can be fatal. Do you have any experience in electronics? More particularly, have you worked with high voltage before? I'd really like to know before giving you a hand here. Haven't gotten anybody killed yet, and it's too late to start.

If you're not familiar with the high voltage end of things, I'd really recommend you stop immediately, and give your project to someone who can work confidently with HV.

Chris

Hi Chris,

no problem with HV,because i work every day with Hv.Normally i use Tektronix Hv probe(X1000).

Now,because i want to create a test bench that doesn't use Hv probe and not Aron insertion wattmeter,my idea was use this analog multiplier.

For unity scale factor,i thinked to use this configuration because X1 can reach at max 4 volt and Y1 300mV.But if you say that with 10 scale factor is more accurate i think that i'll go to use 5 figure in datasheet of AD632.

The Ad632 works well,but if i left the X1 and Y1(X2 and Y2 are grounded) wituout signal at the out i've 9 volt.When i charge the 2 input the multiplier works ok.

I've used a active low filter of 5 order with 3 AD797.

If i put into AD632 2 signal of 2 Vrms at the output i've 4,1 volt.When i filter i've 4,8 volt DC plus 100mV AC.

Is it correct?

I 'testing the circuit in low power and lov tension,with a Fonction generator wich can give me 2 volt RMS.I put this 2 volt Rms in X1 and X1 input of miltiplier.The output is 4,1 volt.But why after the filter the dc signal riseto 4,8 Volt?Is it probably related with average value?

Question: what is the ok reading after the filter(10 hz Fc) when i multiple 2 volt rms in X1 and X2???

Thanks,

ciao,

MAx