Hi Russell - good to see you're back on your feet again! There is a hall effect based system that uses a feedback technique to improve accuracy. The hall effect devices I've used have tended to suffer from offset and temperature drift - and hysterisis from any core material used. The technique is to use a coil in opposition to the measurement coil so that the net field strength at the hall effect sensor is zero. This cancelling coil can be driven from an analogue amplifier, or from a PWM source. It is controlled from the hall effect sensor. The cancellation current can be measured easily using a load resistor and correcting for turns rato so you don't have a lot of power lost. Or you could use a lookup table on the pWM setting? For DC measurement you would still need to be careful regarding any magnetic cores etc as hysterisis could still be an issue. We looked at it for DC current but if driven from an H bridge I don't see why AC wouldn't work - as long as the PWM frequency was >> measured AC frequency. There are commercial sensors using the technique but I don't think there are any patent issues. ??? We didn't end up using the technique for Telecom current measurement as the PWM freqency could injected into the DC line and we had tight noise limits, but otherwise it looked quite promising. There was also a system using a saturated core, the current being related to the time it takes the core to come out of saturation (or saturate in opposite polarity). I thikyou could make an oscillator out of it and measure the frequency? Can't quite remember the details but think there was a NASA paper on it as it iwas used by them somewhere "up there". You might be able to find it. We ended up using a temperature compensated voltage drop measurement on the busbars. Accurate to only about 1% in our version and required in-situ calibration. Not ideal but could cover a very wide range of installations as the busbar size increases with current. No magnetic effects but isolation was a problem. If required it was usually done at the digital level, following ADC stages. Analogue optos had etither large offsets or gain drift and basically couldn't even meet the 1% spec. One system I looked at to isolate larger UPS systems was to just add a cheap radio link. This worked out cheaper than the specilised opto isolators otherwise required. By placing the TX and RX adjacent to each ther you didn't need to worry about antennae and only very, very low power was required. Of course, that expanded into a full RF monitored system once we thought it through with each battery or cell having its own Zigbee module. We never got past the initial design stage on that one unfortunately. Richard On 20 January 2015 at 01:24, RussellMc wrote: > This is partially to see if the list has vanished into the nothing .... > > The following descriptions make it fully obvious what I'm trying to do > while not making it in any way obvious what I'm trying to do - hopefully > :-) > Any general technical knowledge garnered along the way which is not > exceedingly application specific is liable to be freely available. (Even > Gus could be almost happy with that). > > Summary: Looking for a sanity check on my efforts to measure domestic > AC & DC > V > I > Power > all RMs > with accuracies in the 0.1% - 0.5% range (or better) calibrated > > for typically > > 110 - 230 VAC, 0-5, 0-10, 0-30, 0-50, 0-100 , ...A > > 12-60 VDC, 10's to 100's of amps, > > 100-1000 VDC, lower amps (same but different) > > Cost - the lower the better. > > An eg 230 VAC, 30A I, V, line self powered isolated serially interfaced (= to > eg USB/serial com port) measurement node can have a material only BOM cos= t > of <$20 with ease, $10-15 not too hard, <$10 doable., <$5 maybe perhaps h= i > vol corner cutting ... > > ________________ > > I am working on an application that involves measuring AC and DC power, > voltage and current at typical domestic levels at levels of accuracy that > are as good as domestic energy meters. > > This typically involves providing calibrated accuracies in the 0.1% - 0.5= % > range long term and if better than 0.1% can be achieved cost effectively = so > much the better. > > My prototyping efforts have involved the use of 16 bit ADC inputs on a > Spark Core Processor (ARM body, Arduino heart, WiFi mouth). > In some cases I've used remote differential sensing to provide a skerrick > of an attempt at accuracy and in other cases simply used a star earth > system to measurement front ends and single ended analog return to the > Spark Core. > > Needless to say, I expect the true accuracy of the prototype measurements > to be abysmal, I have made no attempts at precision calibration and so fa= r > it doesn't matter. > > "The question" is, where to from here. > > Given: > > I have discovered the world of energy meter ICs which I was only aware of > "at a distance" and find that for say under $2 in modest volume I can but > delta-sigma ICs that measure current and voltage and internally calculate > RMS Power, voltage and current and power factor and more and provide > outputs amenable to easy isolated serial and/or USB interface. Uncalibrat= ed > accuracies depend mainly on sensor accuracy nd calibrated accuracies of > power to say +/- 0/1% is "easily enough" available. > Current sensing (depending on IC) may be some or all of series shunt, CT, > Rogowski coil and voltage measurement is usually via resistive divider fr= om > whatever voltages to typically around 250 mW RMS (eg 1000:1 divide from 2= 3 > VAC mains. Floating power supplies are used to allow measurement head > isolation > > While these ICs are targeted at AC mains energy meter applications they a= re > just as happy with eg a 1000 VDC 5 kW PV array (5A, 1000VDC) or 100A at 1= 2V > or ... with sensor issues being the main difficulties. (eg a 1 milliOhm > resistive sensor at 50A dissipates 2.5W and returns a 50 mV peak signal a= nd > may involve attention to hot spotting resistance non linearity, edge weld > consistency for 4 wire sensing, thermoelectric effects and quite a bit > more. > > For DC hall sensors make life easy in some areas BUT introduce the "hidde= n > internal Vref" issues, shunt magnetisation causing DC offsets when changi= ng > polarity, ... . > CT's work "OK" but efficiencies under 1% start getting hard and Rogowski > coils look promising - but are sure to have their own challenges when I g= et > to them. > > AND > > The many plug in power-meter / energy meter devices (Killa Watt and many > more) actually tend to "do it properly" inside technically if not > construction quality wise and in many cases have "real" ICstherein that > implement the sigma-delta type RMS systems as mentioned above - ie, with > reservations, some such systems use the exact same principles as and are > potentially able to be about as good as the "real thing". > > SO: > > Energy meter ICs with Sigma Delta converters and inherent sub 0.1% > accuracy, various sensors as above, isolated interfaces and floating powe= r > supplies. Highish tech and still amazingly low cost BOM per measurent poi= nt > consdiering the potential results. > > BUT > > What have I missed? > What other niche technologies aimed at some other application exist to > superbly provide solutions? > Are their magic hall devices that blow away the annoying secondary effect= s? > Are there ways to get high accuracy resistive sensors for less than the > related measurement IC cost,? > Are there ICs that have been used by others that give fantastic results? > How ....? > Can ....? > Will / Do / Should / If / .... other ....? > > _______________________________ > > *Coming soon to a question near you:* > > Switching power levels in the <1, 1-5, 10, ... ranges AC & DC. > Electronics versus relays - pros and cons. > > > > > > Russell > -- > http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive > View/change your membership options at > http://mailman.mit.edu/mailman/listinfo/piclist > -- http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive View/change your membership options at http://mailman.mit.edu/mailman/listinfo/piclist .