Ideas for inexpensive True-RMS Metering?
in [Design]
|
From: John Fields on 3 Jul 2010 16:50 On Sat, 03 Jul 2010 13:38:04 -0700, AM <thisthatandtheother(a)beherenow.org> wrote: >On Sat, 03 Jul 2010 13:11:45 -0700, John Larkin ><jjlarkin(a)highNOTlandTHIStechnologyPART.com> wrote: > >>On Sat, 3 Jul 2010 08:37:38 -0700 (PDT), EnigmaPaul >><enigmapaul(a)gmail.com> wrote: >> >>>Hi Guys, >>> >>>I'm looking for the best ideas for implementing an AC power >>>instrumentation front end for a microcontroller or FPGA. >>> >>>What I need to accomplish is to measure 50-60HZ AC in the range of >>>120V upwards of 600VAC with about 1% accuracy, TRUE RMS. I need to >>>extract: >>> >>>1. Voltage >>>2. Current (via current transformers) >>>3. Frequency >>>4. Power Factor >>>5. Phase difference >>> >>>This needs to work with reasonably distorted waveforms, like what >>>might be found in small generators driven by engines. I also need >>>something that will work for single phase or for 3phase and be >>>scalable to instrument up to two 3-Phase sources. >>> >>>There are a variety of ways to go about this but I'm looking for >>>something that is very flexible and can adapt to a variety of voltage >>>configurations (line to line, or line to neutral) and is very low >>>cost. >>> >>>I would guess that it might be similar to the best practices that >>>would be used in the front end of True RMS DMMs or DAQ cards with AC >>>inputs? I know there are also energy metering chips out there, but >>>they are rather expensive. >>> >>>Thanks for any ideas! >>> >>>PW >> >>The cheapest way to do this is to buy a small uP that has an on-board >>ADC and mux. Scale down the line voltages with resistive dividers. A >>single-phase meter or Y-connected meter can let the uP ride on >>neutral, and sense neutral current(s) with cheap shunts. Other >>configurations will generally need current transformers. >> >>A 10-bit ADC works pretty well, 12 is even better. Add some dithering >>noise to the current signals. >> >>Sample voltage and current inputs and do the math. You can even >>acquire waveforms. The trickiest math is phase angles. >> >>One of the voltage dividers can also be used to run a timer channel to >>get frequency. >> >>The algorithms to get metering-quality power measurements are a bit >>tricky. RMS volts and amps are pretty obvious. >> >>You should be able to do a pretty good 3-phase power meter with a $4 >>uP, or maybe even a $1 one. >> >>John >> > > The cheapest way to get an inexpensive true RMS multimeter is to simply >go buy an inexpensive true RMS multimeter. > > You could not possibly jump into a game such as this with such nievete. --- Naivete invites the jump; reality provides the sudden stop.
From: whit3rd on 3 Jul 2010 17:11 On Jul 3, 8:37 am, EnigmaPaul <enigmap...(a)gmail.com> wrote: > I'm looking for the best ideas for implementing an AC power > instrumentation front end for a microcontroller or FPGA. > > What I need to accomplish is to measure 50-60HZ AC in the range of > 120V upwards of 600VAC with about 1% accuracy, TRUE RMS. I need to > extract: > > 1. Voltage > 2. Current (via current transformers) > 3. Frequency > 4. Power Factor > 5. Phase difference Presumably you mean AC average voltage, AC average current, and you want those to be RMS averages? For power factor, you need power, i.e. average of (voltage * current) as well. Either you'll require three multipier circuits of some sort, or an equivalent in software. "Phase difference" only applies to sinewaves, what approximation would be appropriate if your current waveform is... awkward? To get the power value correct, you have to simultaneously sample the V and I (you can't alternate measurements with a single ADC unless you know some strong limits on slew rates). I've always wanted to use measurements of (V+k*I) and (V-k*I) so the microprocessor would just have to look up the square of the digitized values (the difference of those squares is 4*k*V*I), that gets you a power measurement without integer multiply instructions.
From: John Larkin on 3 Jul 2010 18:03 On Sat, 03 Jul 2010 13:38:04 -0700, AM <thisthatandtheother(a)beherenow.org> wrote: >On Sat, 03 Jul 2010 13:11:45 -0700, John Larkin ><jjlarkin(a)highNOTlandTHIStechnologyPART.com> wrote: > >>On Sat, 3 Jul 2010 08:37:38 -0700 (PDT), EnigmaPaul >><enigmapaul(a)gmail.com> wrote: >> >>>Hi Guys, >>> >>>I'm looking for the best ideas for implementing an AC power >>>instrumentation front end for a microcontroller or FPGA. >>> >>>What I need to accomplish is to measure 50-60HZ AC in the range of >>>120V upwards of 600VAC with about 1% accuracy, TRUE RMS. I need to >>>extract: >>> >>>1. Voltage >>>2. Current (via current transformers) >>>3. Frequency >>>4. Power Factor >>>5. Phase difference >>> >>>This needs to work with reasonably distorted waveforms, like what >>>might be found in small generators driven by engines. I also need >>>something that will work for single phase or for 3phase and be >>>scalable to instrument up to two 3-Phase sources. >>> >>>There are a variety of ways to go about this but I'm looking for >>>something that is very flexible and can adapt to a variety of voltage >>>configurations (line to line, or line to neutral) and is very low >>>cost. >>> >>>I would guess that it might be similar to the best practices that >>>would be used in the front end of True RMS DMMs or DAQ cards with AC >>>inputs? I know there are also energy metering chips out there, but >>>they are rather expensive. >>> >>>Thanks for any ideas! >>> >>>PW >> >>The cheapest way to do this is to buy a small uP that has an on-board >>ADC and mux. Scale down the line voltages with resistive dividers. A >>single-phase meter or Y-connected meter can let the uP ride on >>neutral, and sense neutral current(s) with cheap shunts. Other >>configurations will generally need current transformers. >> >>A 10-bit ADC works pretty well, 12 is even better. Add some dithering >>noise to the current signals. >> >>Sample voltage and current inputs and do the math. You can even >>acquire waveforms. The trickiest math is phase angles. >> >>One of the voltage dividers can also be used to run a timer channel to >>get frequency. >> >>The algorithms to get metering-quality power measurements are a bit >>tricky. RMS volts and amps are pretty obvious. >> >>You should be able to do a pretty good 3-phase power meter with a $4 >>uP, or maybe even a $1 one. >> >>John >> > > The cheapest way to get an inexpensive true RMS multimeter is to simply >go buy an inexpensive true RMS multimeter. > > You could not possibly jump into a game such as this with such nievete. What does that have to do with designing polyphase wattmeters? John
From: Tim Williams on 4 Jul 2010 01:41 <krw(a)att.bizzzzzzzzzzzz> wrote in message news:mnjv269firfs4rdvb5kr8pi0ql06v2pppi(a)4ax.com... > But "Nobody" is still right. The phase angle and the power factor (as defined > as W/VA) are *very* often different animals; if you calculate #4 you may well > not have any meaningful answer for #5. ...if #5 has any meaning, or is even > measurable. Not really. Take the DFT of a series of samples, find the fundamental. Compare phase components of V and I. Done. Also, cos(phi) = DPF, the classical (sine wave) power factor that you learn about in Power Systems. A complex waveform (mixed frequency, anharmonic, noise, independent V and I, etc.) may not have much if anything in common, but in that case, the DPF is pretty weird too, so you get exactly what you should expect. None of this is meaningless or unmeasurable. Tim -- Deep Friar: a very philosophical monk. Website: http://webpages.charter.net/dawill/tmoranwms
From: Paul Keinanen on 4 Jul 2010 03:37
On Sat, 03 Jul 2010 13:11:45 -0700, John Larkin <jjlarkin(a)highNOTlandTHIStechnologyPART.com> wrote: > >The cheapest way to do this is to buy a small uP that has an on-board >ADC and mux. Scale down the line voltages with resistive dividers. Remember to put sufficient number of resistors in series so that the combined resistor voltage rating is 1.5 to 2.5 kV depending on the national standards, so that voltage peaks can be handled without flashover. >A >single-phase meter or Y-connected meter can let the uP ride on >neutral, and sense neutral current(s) with cheap shunts. That is a horrible idea when used with TN-C (or measuring on the TN-C side of a TN-C-S system) wiring system, in which Protective Earth (PE) and Neutral are interconnected at the load. A test equipment wiring fault or a blown shunt resistor could cause the full phase voltage on the cases of multiple load equipment. >Other >configurations will generally need current transformers. Putting current transformers on the phases is the safe way of doing it and works both with wye and delta loads. |