1 wire is custom interface, hard to make "own" devices. Serial port port more usable, but still will require to put some kind of application inside router. You can always buy some low power PC with 2 ethernet and 1 wifi interface and made own router with using Linux.
The metrology specs don't specifically require measurement of harmonics - this is just for test and diagnostic instruments. They do define some voltage and current waveforms containing harmonics and state the accuracy that should be achieved when measuring them. It's not difficult to meet these. Aliasing of harmonics present in both current and voltage doesn't actually affect the measurement of energy, so all that's required is a modest input filter to take off significant high frequency noise, which would otherwise spoil your determinations of V**2 and I**2 which are needed for apparent power assessment.
I was wondering about the anti-aliasing filter design problem especially in case of high value of THD at power line System .. How the did the standards describe this problem demands ( Pass band flatness , at which order of harmonics we should stop acquisition and therefore at is the best value of the cut frequency and stop frequency of the anti-aliasing filter ... I hope some expert would answer me .. thanks
My colleague Dave van Ess did some cool AC power measurements by XORing the 1-bit outputs from a couple of del-sig modulators. This enables the modestly powered PSoC1 to do a usable job in such products. The work I did was on the recently introduced PSoC3 which has a lovely quiet, clean, flexible del-sig ADC. But therein lies an important concept; it's the inherent linearity of the ADC, not its SNR-calculated ENOB, that determines power measurement accuracy. Uncorrelated random noise in the current and voltage channels produces a net zero energy contribution. My simulations indicated that you could get good metrology results just taking the raw 3-bit output from our modulator. The point is that ENOB is a misleadingly irrelevant metric for this task. You need to take much more of an audio perspective. Departures from power measurement linearity at low levels are determined by small-signal non-linearity, not either large signal non-linearity or small-signal noise.
Make vs buy - check out the Kill-A-Watt product
sold on ebay and elsewhere - I've seen them in Home Depot. Great where "close enough" is sufficient. I think "Kill" is based on the Cypress PSOC. The Cypress site has design ap notes for an ac meter application. I don't remember/believe if the PSOC ADC is 14.5 bits but what really is? Check out "Understanding the impact of digitizer noise on oscilloscope measurements" by Jit Lim of Tektronix.
One significant trend is to go away from a single-chip solution. By separating the metrology from the user interface and comms pieces, a manufacturer can get the legally-binding stuff sorted out but still alter the product for various utilities, different AMR/AMI schemes, different logging/billing requirements and so on. It's possible to achieve spectacular performance with recent devices like the Cypress PSoC3 - I spent months deep-diving into that. In that case, BTW, I eliminated the single-ADC multiplexer delay problem with a polyphase FIR fractional interpolator. We easily beat Class 0.2 and OUML R46D specs across the board and were 6x better than the 62053-23 reactive specs, in all 4 quadrants, with a single-point cal. But as you say, it's not about the specs, but cost and also comfort factor; it costs a lot to get certification done.