Will this work? temp and power monitor

[BraveSirRobbin]

As another option to the 1-Wire interface is using a DataNab Ai8_R13 which has pulse inputs. This would replace the need for the GP1 Counter and the Ethernet 1-Wire Host Adapter. For Temp sensing depending on the number of probes you need the Ai8_R13 would still have 7 inputs left for Analog 10k Temp Sensors

You would need to write a driver for the 'modbus' serial interface though. Also be aware that their analog inputs only go to five volts (not a problem with this application, but not a "good" general limit for A-D inputs for various other projects). BUT, I do like the looks at those temp sensors (will have to look into their accuracy)!

Edit: Accuracy of temp sensors looks good: "Thermistor Accuracy: +/- 0.2C"

 

[toymaster458]

You would need to write a driver for the 'modbus' serial interface though. Also be aware that their analog inputs only go to five volts (not a problem with this application, but not a "good" general limit for A-D inputs for various other projects). BUT, I do like the looks at those temp sensors (will have to look into their accuracy)!

Edit: Accuracy of temp sensors looks good: "Thermistor Accuracy: +/- 0.2C"

A driver for the Ai32 and Temp sensors is all ready in beta. The Ai8_R13 pulse inputs should not be that hard to add. Here is a sample of how I am using it in CQC. The driver keeps a record of the days high and low temps and resets at midnight.

 

[hult]

I'm going to order the hardware this week. Do you have a picture of what your install looks like in the breaker box?

I'll take a picture tonight & post it.

This is what two Magnelab SCT-0750-150 split-core current transformers look like in my entrance panel:

www.ECOntrol.org/150A_CTinPanel.jpg

The label (which bears the RU/UL symbol ) needs to be facing the source of power for the output polarity to be correct. This is fortuitous at least in my case because it measn that the label can be viewed from above when the CT's are installed.

These are much smaller than the SCT-1250-400's that they replaced.

Here's a photo of the latter with a variety of other current transducers:

www.ECOntrol.org/CurrentTransformers_1.jpg

I have 30 of the LEM LA 100-TP current transducers (small blue open rectangles in upper left of photo and have designed a buffer board for them. These are precision, active devices.

Schematic is here: www.ECOntrol.org/TCBuffer_01Schematic.jpg
PCB layout is here: www.ECOntrol.org/TCBuffer_01PCB.jpg

This was done with Eagle CAD software and the boards are small enough to be designed with the free version. I can share the design if there is interest These are specific to the transducer used have more flexibility and capabilities than absolutely needed. I prefer clean robust electrical signals to noisey ones.

Thirty devices will allow me to measure most of the AC loads of interest iin the house including AC lighting in part because I have been migrating to centralized hardwired lighting. DC loads will require a shunt resistors rather than current transducers but the required electronics are much simpler. More on that if there is interest.

By buffering the current sensors, I can multiplex the inputs to one or more microcontrollers to perform the V*I calculations. This cuts overall costs by an order of magnitude or so.

... Marc

 

[toymaster458]

Also be aware that their analog inputs only go to five volts (not a problem with this application, but not a "good" general limit for A-D inputs for various other projects)

These products where designed mainly for building automation where 5 volts is the industry standard. 10 volt reading can also be read using a voltage divider or go with a Ai32 on those applications which is switchable between 0-5V or 0-10V.

 

[Michael McSharry]

I had been looking for a few years for whole-house real time power measurement and settled on the Watt Meter as an effective solution. One of the hurdles that had made this task more expensive that I wanted to invest is the 400 amp service where I have 6 cables from the utility meter to the two circuit panels. For most of the available solutions this meant my cost was twice the standard 200 amp service because each solution could only handle 2 CTs in a split-phase configuration. With the Watt Node I am able to attach 4 CTs to the single Watt Node so my total system cost is one Watt Node, one Computer Interface and 4 CTs.

I wanted to have the ability to measure small variations in power use and wanted it to be very responsive so a real-time measurement could be observed. The Watt Node delilvers pulses proportional to the power being utilized and the pulse rate can be set to fit the needs of the application. My computer interface is a DS2423 1-wire device. For this device a 100 Hz count looks to be the practical maximum. My standard polling rate for the DS2423 is every 5 seconds, but can be accelerated to at least 2 Hz if I want to make some specific meassurements. This means I will be able to resolve 16 counts, twice a second, at average utilization. 20 amps / 16 counts = 1.2 amp/count/sample or 0.6 amp/second which is the equivalent of observing a light bulb operation.

While I have a 400 Amp service, my average utilization is 20 amps. While I have not measured peak utilization, I felt that allowance for 3x the average would be the appropriate max amperage that I would need measure. Since I run 2 CTs in parallel this means each CT will have a full scale measurement of 30 amps. The stated accuracy is 1% or 0.3 amps over the range of 3 to 30 amps.

This design tries to achieve accuracy and real time response. The 100 Hz Watt Node output frequency may be at the limit of the DS2423's capability, but the 33 Hz average should not be a problem. I suspect most users setting up whole-house power monitoring will use a much lower sample rate such as 2 Hz rather than 100 Hz and forego the resolution and update rate of the feedback. The DS2423 uses a 32 bit counter which means the computer has days to capture the counter value so even if the computer fails the DS2423 will continue to accumulate power utilization so longer term (i.e. daily, monthly) values should remain accurate.

The linear range of the 30 amp CT starts at 3 amps. This is a concern during the hours when power consumption is lowest and accuracy may be lost during this period. At the same time, the 60 amp peak may also be insufficient to handle simultaneous inrush current that may occur in certain situations. The low end problem will be much worse for those that size their CTs to circuit panel capacity such as 150 amps.

 

[cactsbob]

I'm going to order the hardware this week. Do you have a picture of what your install looks like in the breaker box?

Here are the pictures

 

[jim3342]

Thanks for the pics. I just want to be clear on all the hardware pieces you are using with approx costs that I've found.

2 CT's ($90) connected into a Wattnode ($170). The output of the WattNode is interfaced to the Pulse Counter ($50) and that is connected to the Ethernet Bridge ($150) which sends all the readings to your .net service. Am I missing anything?

 

[hult]

View attachment 1296

I had been looking for a few years for whole-house real time power measurement and settled on the Watt Meter as an effective solution. One of the hurdles that had made this task more expensive that I wanted to invest is the 400 amp service where I have 6 cables from the utility meter to the two circuit panels. For most of the available solutions this meant my cost was twice the standard 200 amp service because each solution could only handle 2 CTs in a split-phase configuration. With the Watt Node I am able to attach 4 CTs to the single Watt Node so my total system cost is one Watt Node, one Computer Interface and 4 CTs.

Good find! And good idea. Sharing the V * I calculation circuitry between inputs is cost effective. The Watt Node presumably multiplexes the inputs from the CT's internally . The system I designed for my house has approximately 45 circuits. The inputs are multiplexed externally to the I * V power-calculation circuitry.

I wanted to have the ability to measure small variations in power use and wanted it to be very responsive so a real-time measurement could be observed. The Watt Node delilvers pulses proportional to the power being utilized and the pulse rate can be set to fit the needs of the application.

I designed and built a Multi-channel AC Power monitoring network about seven years ago using SAMES SA2002 IC's.

The IC's accomplish the P = V * I measurement with high accuracy and precision and have pulse outputs (although IC's with serial outputs are also available fom the manufacturer).

I have a few extras of these ICs and could provide at no cost to interested Cocooners wirth the proviso that they share their results.

Description here: www.econtrol.org/power_measurement.htm
IC data sheet here: www.sames.co.za/files/sa2002e.pdf

My computer interface is a DS2423 1-wire device. For this device a 100 Hz count looks to be the practical maximum. My standard polling rate for the DS2423 is every 5 seconds, but can be accelerated to at least 2 Hz if I want to make some specific meassurements. This means I will be able to resolve 16 counts, twice a second, at average utilization. 20 amps / 16 counts = 1.2 amp/count/sample or 0.6 amp/second which is the equivalent of observing a light bulb operation.

I am surprised by the low maximum frequency of the DS2423 input. My recollection was that it was much higher.

Here's an analog multiplexor and conditioner board that I designed that both provides for extensive analog signal conditioning and buffering and has provisions for optional on-board 1-wire devices including dual DS2423 counter, quad DS2450 ADC and DS18B20 etc temperature sensors with DS9503 ESD protection diodes on the 1-wire bus.

DC power, four analog outputs, analog ground and 1-wire bus are provided through a single RJ45 socket or screw terminals for Cat5 wiring.

www.ECOntrol.org/diy_projects.htm

Click on " 4-input Analog conditioning buffer with 1-wire support " for schematic and pcb layouts.

This PCB provides for both an analog and a digital ( 1-wire) signal from four or more local environmental sensors. This can helpfully facilitate use of sensor data by two or more different microcontrollers or PCs. One controller can concentrate on real-time response (eg conventional HA); the other can provide long-term data recording, archiving and preservation. For data sets in which data continuity is important, a simple data logger with a single-minded mission is a useful, parallel add-on to a task-oriented, real-time, PC-based HA system.

The schematic and PCB were done with Eagle CAD software and are in a size small enough that they could be revised/tweaked/re-purposed using the free version of the Eagle software http://www.cadsoft.de/freeware.htm.

These input boards have more extensive filtering and buffering options than are typically needed and many of the component locations could be left unpopulated depending on application. I can provide the source files should a Cocooner want to build some or modify for a different application.

While I have a 400 Amp service, my average utilization is 20 amps. While I have not measured peak utilization, I felt that allowance for 3x the average would be the appropriate max amperage that I would need measure. Since I run 2 CTs in parallel this means each CT will have a full scale measurement of 30 amps. The stated accuracy is 1% or 0.3 amps over the range of 3 to 30 amps.

This design tries to achieve accuracy and real time response. The 100 Hz Watt Node output frequency may be at the limit of the DS2423's capability, but the 33 Hz average should not be a problem. I suspect most users setting up whole-house power monitoring will use a much lower sample rate such as 2 Hz rather than 100 Hz and forego the resolution and update rate of the feedback. The DS2423 uses a 32 bit counter which means the computer has days to capture the counter value so even if the computer fails the DS2423 will continue to accumulate power utilization so longer term (i.e. daily, monthly) values should remain accurate.

The linear range of the 30 amp CT starts at 3 amps. This is a concern during the hours when power consumption is lowest and accuracy may be lost during this period. At the same time, the 60 amp peak may also be insufficient to handle simultaneous inrush current that may occur in certain situations. The low end problem will be much worse for those that size their CTs to circuit panel capacity such as 150 amps.

In an air conditioned house with major electric appliances such as ovens and dryers, this approach may only work well part of the time depending on intended use of the data.

The dynamic range of the linear output a such a system could be improved by using two CT's (sensors) with different maximum input on the same conductor. Someone with a single 220vac (two single-phase legs) service antrance panel could use both a high value CT (say, 150 amp) _and_ a low-value one (eg the 30-amp CT as per Michael's description) to provide 1:5 autoranging using the four inputs of the Watt Meter. As always when there are two measurements of the same quantity/signal, one has to figger out a way to deal with the inevitable difference in values in the region of overlap but there are simple, conventional ways of coping with that. The non-linearity at low and high extremes can be dealt with -- at the expense of additional programming complexity, -- by calibration using curve-fitting and(or) look-up tables if the non-linearity is consistent. T

The LA 100-TP current transducers I plan to use on most circuits (blue component in the attached photo referenced by url in my previous post) has excellent linearity owing to the built-in Hall-effect sensors. These CTs (like many other components) were obtained at near-negligible cost via eBay.

HTH ... Marc

 

[hult]

I'm going to order the hardware this week. Do you have a picture of what your install looks like in the breaker box?

Here are the pictures

Cactsbob 's picture on the left showing exposed low-voltage output wires from the CT's might provide an ... ahem ... 'unwanted visual stimulus' to an electrical inspector. Upon further inspection, s/he will find that the voltage rating is not printed on the wire per US National Electrical Code. Consider protecting those wires both with heat-shrink tubing and with sheathing as I showed in the picture I referenced by url yesterday and now attach to this post. This is also better, safer wiring practice in this geologist's (Not-An-Electrician's) opinion.

Marc

 

[hult]

I would use the "True RMS" split core current transformers with the 4-20 ma output (better for long cable runs and is more noise immune).

You should also be measuring voltage as well as WATTS = VOLTAGE * CURRENT, but again, the voltage probably doesn't change all that much and you could get away with using a standard measured value.

"True RMS split-core current transformers" are by definition _averaging_ devices (RMS = Root-Mean Squared; the mean is an average). These are useful but expensive new because they need to embed the circuitry to calculate RMS from the AC signal.

And to clarify, they are not what one wants for very accurately calculating _instantaneous_ " watts = volts * amps" because to do this with greatest accuracy, the calculations need to be performed at many times during the AC cycle especially with noisey loads such as conventional TRIAC dimmers.

One potentially less expensive route to 'complete satisfaction' that provides an upgrade path to 'instantaneous V*I calculation' is to use multiple less expensive, conventional (instantaneous) current transducers through a multiplexor and 'read' by a single RMS conversion circuit.


So numerous Current Transformers, Shunt Resistors, or a mix of CT's and SR's can share a single RMS (or other ) converter and Analog -to-Digital Converterinp

CT -- +
SR -- |
CT -- | Multiplexor ---> RMS converter ---> ADC
SR -- +


Attached is a picture of surplus RMS converter circuit board. I obtained a dozen of these on eBay for a few dollars each.
The inset photo shows the actual RMS converter IC on the PCB.

One can then measure the output of the RMS converter using (eg) Peter Anderson's nifty offerings and either measure or assume the AC voltage ( eg 115vac) needed for the V * I calculation.

An advantage of this approach in my experience is that the hardware can be improved incrementally depending on actual need experienced It's easy to imagine that you need more accuracy than you really do.

If you buy expensive "True RMS current transducers" and then decide that you need instantaneous V*I calculation, your investment in Time and Treasure is for nought..

But if you determine that using the AC voltage at the (eg) entrance panel is not accurate enough owing to (eg) voltage drop in the AC distribution conductors in your house, one can add the required local AC measurements through a small wall-wart and do the calculations with that, more accurate, data on an as-needed basis.

CT --- +
CT --- |
CT --- | Multiplexor ---> RMS converter --> Analog-to-Dgital --> Software Multiplier
CT --- |
VAC --|
VAC --|
VAC --|
VAC --+

And if you then decide that you need instantaneous V*I measurement, you can substitute a V*I calculator and have obsoleted only *one* component of your installed system, namely the RMS converter.

CT --- +
CT --- |
CT --- | Multiplexor ---> Analog-to-Dgital converter +
CT --- | |
+ --> Instantaneous V * I calculator
VAC --| |
VAC --| Multiplexor ---> Analog-to-Dgital converter +
VAC --|
VAC --+

The V*I calculator could be a specialized IC such as the SAMES ICs I reference elsewhere in this thread, a homebrew analog circuit such as what Ed Cheung has documented, the Watt Meter Michael McSharry describes, a PC with few enough tasks to be near real time, or a micontroller such as a Basic Stamp, BX-24, PIC or AVR. I am partial to teh AVr solution and will post a PCB and circuit time permitting. It uses the free (for no-commerical use) version for Eagle CAD software ( www.cadsoftusa.com)
and code that might be compact enough to be compiled with the free version of BASCOM AVR basic (www.mcselec.com ) .

I'll try to consolidate this at my www.ECOntrol.org web site over the next week or so.

HTH .. Marc

 

[BraveSirRobbin]

I was thinking more of using the RMS for the entire house (thus one would need only two) for measuring "overall" power monitoring, but; as you stated, they are more expensive.

 

[hult]

I was thinking more of using the RMS for the entire house (thus one would need only two) for measuring "overall" power monitoring, but; as you stated, they are more expensive.

Ah ... Then your "multiplexor" for the two phases could be as simple as a single-pole, double-throw, signal-quality relay or switch.

(Email me your snail-mail if you want one of the RMS converters in the picture -- gratis. Least I can do for our BSR ;-)

... Marc

 

[cactsbob]

Actually, I was a licensed electrician for some 15 years. Although, I haven't worked as an electrician in a few decades. So, things could have changed. But, a few centuries ago I installed CT's in 480 switch gear without any grief from Inspectors. Usualy, the insulation of the CT wiring was rated the same as the maximun cable that the CT would fit. If in doubt check with the manufacturer. I just checked on the CT's that I used & they are rated for below 600vac UL, CE approved. The Wattnode itself is approved to be located inside low voltage (120vac - 480vac is considered low voltage) switch gear.

Easy Installation saves you time and money.
The WATTNODE is small enough to fit entirely within a
standard electrical panel and the screw terminals unplug
for easy wiring.

http://www.ccontrolsys.com/downloads/WNPulsedsheet.pdf

With that said, it's better safe than sorry. I located my Wattnode outside of the switch gear and adding an approved conduit or raceway for the CT wiring might be worth the extra effort.

I contacted Wattnode & got this response. I don't think I'm going to change my installation. But, anyone adding a new installation might like to take the route that Mark took & add the extra conduit.

It’s better to put the CT wiring in a separate conduit. Not so much for safety, but as to prevent noise from the mains voltage coupling on to the CT wires. The CTs have less than 0.333VAC output, so noise from the mains voltage could be a problem.



Best Regards,

Steve Whitaker

Director of Marketing

Continental Control Systems, LLC

I'm going to order the hardware this week. Do you have a picture of what your install looks like in the breaker box?

Here are the pictures

Cactsbob 's picture on the left showing exposed low-voltage output wires from the CT's might provide an ... ahem ... 'unwanted visual stimulus' to an electrical inspector. Upon further inspection, s/he will find that the voltage rating is not printed on the wire per US National Electrical Code. Consider protecting those wires both with heat-shrink tubing and with sheathing as I showed in the picture I referenced by url yesterday and now attach to this post. This is also better, safer wiring practice in this geologist's (Not-An-Electrician's) opinion.

Marc