Analog dynamic range, accuracy and number of bits

On a sunny day (Fri, 2 Nov 2012 18:57:15 -0700 (PDT)) it happened Robert Macy wrote in :

That could be illegal, as somebody may put something magnetic on the meter without any bad intent.

Le 03/11/2012 01:33, John Larkin a écrit :

I like VMEBus, i remember i have been designing VME IO's cards interface with VME a controller named VIC (Cypress) ... it was about 20 years ago !

Oh yes with a re&l 16bits we can expect the IEC tests benches.

For currents it will be CT with a ratio 1:100 and furthermore we have to design the interface to support Rogowski Transformers. VT for voltages.

Accuracy is specified the interface only. A procedure to calirate and zeroing the interface *with* the sensor (all over the range) will be done by software.

Best Regards, Habib.

I have only seen V used for voltage in anglo-saxon literature, in many other cultures (notably German) , U has been used.

n = nominal

Le 03/11/2012 01:28, rickman a écrit :

Right. U and must be synchronized of course (each sync signal every 1s) in order to re-synchronize the two ADC's)! Remember i have to compute Power/Energy. each U, I sampled *synchronously* and make the sum/mean/square/product ... over 1s.

I live in France. U (or sometimes V) is for Voltages. Un stands for V rated and In stands for I rated. Excuse my English ...

Good point. The bench for IEC6.... -21 conformance will be done by an independent agency. Some sort of test equipement (PPS400.3 and PRS ...) will be used. I don't know exactly how and what these guys will handle any test protocol :-(

Best Regards, Habib.

That would make 160 samples/cycle at 50 Hz, ie. a sample every 2.25 degrees of the mains cycle. The conduction angles can be quite short in power supplies, so one might ask, does it make sense to use sampling rates that are exact multiple of the mains frequency.

There has been a similar discussion about energy metering here in sci.electronics.design a few months ago, so check groops.google.com

As a general plea to all posting information request to this group, provide as much as possible information about the problem (limited by any NDAs) and you will get much better responses immediately.

In the worst case, some requests look like "do my home works, please":-) and the responses will reflect that.

Dnia 03-11-2012 o 11:29:19 Habib Bouaziz-Viallet =

napisa=C5=82(a):

) =

In practice, you will need to clock the two ADCs from the same source. Y= ou =

can, however, get away using a single ADC for current and voltage. When = =

designing a 3 phase energy meter I used one ADC with a single =

sample-and-hold and multiple inputs. The phase shift was corrected quite= =

easily with a linear phase FIR filter. Reached Class 0.5 (MID Class C) =

accuracy on a 12-bit ADC (internal, on a micro) with 2 external gain =

ranges on current inputs. For the old Class 0.1 (0.1% accuracy) just use= a =

pair of good dedicated 16-bit Sigma-Deltas and you'll be fine.

... and it makes perfect sense, since V is the SI unit for voltage. V =3D= 5V =

looks weird, U =3D 5V is better.

The metering guys don't care about any digital ways of reading the =

measurement data for calibration. All they want is a LED pulse output th= at =

behaves exactly the same as the rotating plate in electromechanical =

meters. All their test and calibration benches are configured for a puls= e =

output.

Regards, Andrzej Ekiert

-- =

Macy

not in the countries they deploy these meters. The govt and industry are all pwerful there.

Don't they test up to PF of 6? or is that 10? Digital has a it of trouble there since we're starting to subtract to large numbers to find the power. Those 'disk' meters have an incredibly good accuracy curve thatconverges down to 'almost' zero.

On a sunny day (Sat, 3 Nov 2012 06:04:32 -0700 (PDT)) it happened Robert Macy wrote in :

Strange, I have one, not sure about that gov thing and industry, that is the same here, but if it was to show wrong values I could sue. Would be easy to demonstrate, hook up a 1kW space heater, readout, put magnetic object on meter, if it speeds up, you win the case always. Could be a magnet as placeholder for one of those 'safety guides' you are supposed to hang next to the meter.

Most CTs, especially cheap ones, have more phase shift, up to a few degrees, at low currents. That can be fudged in software, by blending in a bit of saved quadrature current samples at lower average currents. *IF* the CT is known and characterized.

Why not just trigger two ADCs simultaneously?

There's no need to synchronize the ADC trigger times to the AC line. That falls out of the math.

PF is always in the range of -1 to +1.

Digital has a it of

A digital power meter just simultaneous-samples the voltage and current signals, multiplies, and integrates. That works for all waveforms and all power factors. No trig, no subtraction required.

It makes sense to "walk" the waveform with N samples uniformly spaced per cycle. But that doesn't require a high sample rate; one can equivalent-time sample, like a sampling oscilloscope. For example, in a 50 Hz system, the line period is 20 milliseconds. You could sample every 20.1 milliseconds, which is 361.8 degrees, which is equivalent to sampling every 1.8 degrees. Every 200 samples, every 4.2 seconds, you will have tested all 1.8 degree spots on the line waveform.

If the voltage waveform is sinusoidal, there is no power in the current harmonics, so a high sample rate isn't needed to measure power and energy.

With a 16 bit or better ADC, there will be enough noise around that dithering isn't necessary.

else certain waveforms will cause strange power

I don't understand that. Dithering the current signal just smears out the ADC quantization. It does show up in the RMS current value, if you compute that off to the side, but it will be a few ADC LSBs, not a big deal.

Le 03/11/2012 17:13, John Larkin a écrit :

In fact there are two type of daughter board, one for currents (CT, Rogowski) and the other for voltage. I have to synchronize the delta-sigma modulators of each Delta-Sigma ADC's, 1 time per second to avoid false measurements based on each local quartz drifting ... That's the meaning of synchronizing.

Does each ADC have its own oscillator? You'd need to have your two oscillators match to something like 5 PPM!

It would be simpler to use, say, two 16 bit SAR ADCs and just trigger them simultaneously. There are nice SAR ADCs with SPI interfaces that sample and digitize as you read them out, with no explicit trigger needed. ADS7866 sorts of things. No oscillator needed, no phase drift problem.

Why separate the ADCs? Just run the analog signals from the acq boards to one location, and digitize there.

What the hell does "nominal" mean? When I last looked it up it meant "in name only", or in other words, AC power line might be 120V *nominal* since the actual voltage can vary hugely over time.

How would you make use of "nominal" voltage and current in a measuring device? I can already tell you what that power level is, 120V * your current!

As to the issue of V being anglo-saxon, I thought V for voltage was SI, an *international* system of units which define not only the units and how they are measured, but also the nomenclature which includes the abbreviations. Don't you use SI?

Rick

Interesting. The name, "SI" is actually the abbreviation in French for international system. I believe the meetings and some of the actual international standards (like the old meter bar) were kept in France, so much of the language for the system came from French.

So how do the French come up with U for voltage when the SI unit is V?

What exactly does "rated" mean? Is that what you call your "measured" value, the sample? Oh, is the 'n' the index, like V1, V2, V3,... Vn?

You don't need to apologize for your English. It is tons better than my French. That's one bad thing about living in the US. Other than Spanish which is becoming more common here, we don't get exposed to many languages other than English.

Yeah, I'm sure this is a very complex area with all sorts of good and expensive test equipment. But at some point they have to verify that it all works correctly under the crazy conditions found on real circuits.

Rick

I didn't think of it that way. The SI V is the abbreviation for the unit "volts", but the symbol V or U is a math symbol and not necessarily the same and the units abbreviation. Duh!

Rick