Archived Forum

An update...

I took a stopwatch out the electricity meter and counted the time between the red LED flashes, it's about 16 seconds, so I'd calculate the meter thinks I'm using around 225W per hour, whereas the EmonTX thinks it's about 78W per hour...

Phil

Phil, you need to get your units right, only the Beeb are allowed to mix up power and energy!  Your meter and emonTx are both measuring power, in kW.  When multiplied by time, this gives the total amount of energy that has been used, in kWh.

You've done a good selection of tests on Day 1.  When checking the unit on an isolated rig (which is generally all that I've  done), your results look OK.  But when you superimpose the effect of your house consumption, things look different.  That's not a test that I've ever done: to measure the household's background rate and then turn on & off an extra load of a known capacity.  If your 2 kW heater measures 2 KW on a test rig with zero background flow, then I would expect it to cause an increase of 2 kW when added to any other situation.

This logic presumes that you're running within the linear part of your measurement system.  You could confirm this by running one of my test sketches such as RawSamplesTool_4ss_2.ino.  If all is well, you should see a suitably sized pair of near-sinusoidal waveforms when the max current that you want to measure is present. 

Maybe a stupid question, Phil: you have put the c.t. on the correct cable?

And your meter says it is one flash per 1 Wh? (That's what many are, but some aren't.) 225 W standing consumption sounds reasonable but let's make absolutely certain there aren't any assumptions misleading us.

It's not by any chance a 3-phase system with the a/c adaptor connected to the wrong phase is it?

(2269W + 225W) x cos(120) = 1247W which is very close to your reading

Calypso_Rae - Oops, it's normally me being pedantic about things like that. Yes, it's a bit odd. I've give it a test with the RawSamplesTool this evening and report back.

Robert - Yes, it's the correct cable, brown with little orange 'LIVE' cable ties on it. And yes, I should have mentioned that it's an Ampy meter which states "1000 imp / kWh", so 225W should be correct. Interesting, I just dug out my old OWL energy meter this morning and connected its CT next to the EmonTX CT and it's reading about 250W, which whilst inaccurate too, seems more accurate than the EmonTX. Very odd.

Martin - That's an interesting result, but no, it's a single phase supply to a residential property. It also says single phase on the Ampy meter.

Thanks again chaps!

Phil

It's got to be something weird or stupid, as the calibration on test was inside 0.5%. You have pushed the c.t. plug in all the way? And clipped the c.t. together properly with no air gap? You are measuring voltage and are you using the emonTx_CT123_Voltage.ino sketch? What are the voltage and current outputs on the serial monitor?

(After ct1.calcVI( .... ) put Serial.print("V=");Serial.print( ct1.Vrms ); Serial.print("  I=");Serial.println( ct1.Irms );  or whichever c.t. you are using.)

V ~= 240,  225 W is ~= 1 A,  2200 W ~= 9 A  so is it voltage, current or both that are wrong? If it's the current, reverse the c.t. on the cable and see what the effect is (apart from power going negative!).

Robert,

Yes, plugs are all the way in and no air gaps. Yes, I have an OEM.org supplied 9V power supply plugged in and am using the emonTx_CT123_voltage.ino sketch. The serial output of the voltage was a bit low, so I've adjusted the sketch and now it matches the digi-meter connected to mains. Connecting the CT to four way power strips live cable, as in the Calibration examples, with the outer insulation removed and pluging in lights/heaters through a power meter while watching the serial output on the laptop, everything works as it should. 100W lightbulb ~= 100W on both, fan heater ~=2200W, heater/lightbulb  ~= 2300. Reversing the CT just reverses the sign on the serial data. Voltage last night was around 245V and agreed with the digi-meter which was also plugged in and out to check.

Moving the CT back to the Live feed into the house, about 6", as all this was taking place outside in meter cupboard and it all goes pear shaped.

Hopefully, I'll get some time tonight to have a go with Calyso_rae's RawSampleTool later on and see if that shows up anything.

Phil

Chaps,

Sorry for the delay.

I've added two links below to pastebin, with the results. The first is the RawSample Tool running against the house load with a bit of solar input from the panel on the roof. The second is the same but with 100W bulb being measured and the bit of solar input. Each of the pastes has three runs of the test. They all look like sine waves to me, all be it slightly kinked.

http://pastebin.com/1Ae1PXUr

http://pastebin.com/Fy7CHDzA

Hope that throws some light on the issue.

Phil

What do you read for current and voltage when you insert that line I asked you to? With the c.t. on the main incomer of course.

A recap:

You have a single phase supply. You are in the UK? You've built an emonTx, you have a c.t. and a v.t., and on test it measures real power accurately to within a small error.

You left the v.t. where it was and you moved the c.t. 6" onto what you think is the line meter tail, and it reads power 40% low, i.e. it reads about 60% of the value you expect to see.

As I understand it, the Owl meter only measures current and estimates power, assuming a voltage and power factor. That might explain its apparent discrepancy. When you reported the Owl reading, can you clarify: was that on your test rig or on the meter tail?

I can only think of two possibilities now: either you have misidentified the cable you clipped your c.t. onto, or there's a massive magnetic field that's cancelling the current the c.t. is looking for. Can I have a photo to show your meter, consumer unit and PV infeed point etc,  so that I can see which cable the c.t. is on when it reads low, with enough context to identify each cable?

There are four main cables into your meter like (a) here - Line and Neutral on the supply side and Line and Neutral on the load side. Normally, the supply side is the left, and the neutrals are normally the central pair. You should see the same current in all four.

Hi Phil, thanks for posting those results.  They really shed some light as to what's going on (or not!).

As you say, your voltage waveform looks fine in all cases.  But the current waveforms are certainly not OK - those are the straight lines running down the centre of each display.  No sinusoids!   The graphs in the upper part of each set of results is compressed to fit with a screen's width.  The raw values appear below, but nothing seems to be happening in your case.

The most likely explanation is that you're using a different CT than the Raw Samples tool has been set up to use.  The emonTx mapping between its CTs and the Analog inputs is not very intuitive, but is shown on the board if you look closely at the port markings:

CT1 is analog 3

CT2 is analog 0

CT3 is analog 1

Vac is analog 2

For all sketches that I've posted to date, current is expected to be measured at CT3 (A1).  If you're using a different CT, you'll need to change one line of the code and run the tool again.

You won't see much effect with a light bulb, I suggest you try a 3 kW kettle of similar.  Ideally, your current sensor should be using most of the available ADC range when the maximum power that you want to measure is present.

Here's an extract of the results file that I posted along with the enhanced Raw Samples tool on 21/12/13.  At half-heat, my Bosch hot-air blower only takes power for half of each mains cycle, hence the strange looking waveform for current.

                                     v .  c                                    |
|                                       .  vc                                   |
|                                       .   c  v                                |
|                                       .   c     v                             |
|                                       .   c         v                         |
|                                       .   c            v                      |
|                                       .   c               v                   |
|                                       .   c                 v                 |
|                                       .   c                   v               |
|                                       .   c                   v               |
|                                       .  c                     v              |
|                                       .  c                     v              |
|                                       .  c                    v               |
|                                       .  c                  v                 |
|                                       .  c               v                    |
|                                       .  c            v                       |
|                                       .  c        v                           |
|                                       .  c     v                              |
|                                       .  c v                                  |
|                                       v  c                                    |
|                                   v   .c                                      |
|                               v      c.                                       |
|                           v         c .                                       |
|                        v          c   .                                       |
|                     v           c     .                                       |
|                  v             c      .                                       |
|                v              c       .                                       |
|              v               c        .                                       |
|              v               c        .                                       |
|              v               c        .                                       |
|              v               c        .                                       |
|              v                c       .                                       |
|                v               c      .                                       |
|                   v             c     .                                       |
|                      v            c   .                                       |
|                          v          c .                                       |
|                              v        c                                       |
|                                 v     . c                                     |
minVoltage 192,  maxVoltage 833,  minCurrent 404,  maxCurrent 571


Raw data from stored cycle: <Vsample>,<Isample>[cr]
226, <<< No of sample pairs
523,571
555,570

Robert,

I'll start with the re-cap...

You have a single phase supply. You are in the UK? You've built an emonTx, you have a c.t. and a v.t., and on test it measures real power accurately to within a small error. - Correct

You left the v.t. where it was and you moved the c.t. 6" onto what you think is the line meter tail, and it reads power 40% low, i.e. it reads about 60% of the value you expect to see. - Correct

As I understand it, the Owl meter only measures current and estimates power, assuming a voltage and power factor. That might explain its apparent discrepancy. When you reported the Owl reading, can you clarify: was that on your test rig or on the meter tail? - The Owl was on the meter tail. I realise it is somewhat inaccurate, however when the fan heater was connected it was more accurate than the OEM, showing 2.5kW were being used. Without the fan heater connected, measuring just the base load, I could work out the W/h being consumed by timing the flashes on the electricity meter, which also showed the OEM was reading low.

I can only think of two possibilities now: either you have misidentified the cable you clipped your c.t. onto, or there's a massive magnetic field that's cancelling the current the c.t. is looking for. Can I have a photo to show your meter, consumer unit and PV infeed point etc,  so that I can see which cable the c.t. is on when it reads low, with enough context to identify each cable?

There are four main cables into your meter like (a) here - Line and Neutral on the supply side and Line and Neutral on the load side. Normally, the supply side is the left, and the neutrals are normally the central pair. You should see the same current in all four. I haven't taken a photo, but can confirm I have the same Ampy meter as in the top/left of your photo. The Live Load is far right cable to which the CT is clipped. Approx 6-8 inches away the Owl CT is connected to the same cable. The left had pair go though a large fuse and disappear into the ground. If there was a big magnetic field, wouldn't this have affected the Owl CT too?

As for the current/voltage ready you asked for, I forgot, so I've done it just now... And got this -

http://pastebin.com/4XqET0Ve

Which is bit of turn up for the books... This is measuring the whole house load, approx 160W-ish which agrees with the timed meter flashes. Then I turned on the 2300W fan heater, a 100W light bulb and then a 2800W kettle. Bizarrely this all seems to work fine now. I removed the prints and it still works fine. 

The pessimist in me hates this type of thing, as stuff doesn't just suddenly decide to work properly, so I can only summise that something in the house was causing this effect. Usually there is more stuff is turned on, but the kids are in bed and my wife is out, so maybe something is off that was on last time? Could it be power factor related or badly suppressed, or using half cycles like Calypso Rae's heat gun? Is this a possibility? Would it have to be big like a water heater? Or could something like a bad wall wart type PSU do it?

Thanks for your time,

Phil

Hi Rae,

You're quiet correct, I just downloaded and ran the sketch, not knowing what I was looking at it seemed to run fine.

The default sketch doesn't use CT1, I adjusted my copy and ran it again...

http://pastebin.com/XEDWAD3U

First with the 2300W fan heater which didn't make much of a sine wave, then again with the fan heater and the kettle on. Still not much of a sign wave but slightly bigger. I guess it looks fine, as I mentioned above the OEM seems to be reporting the correct results at the moment. I guess I've got to try switching stuff on one at a time see how it affects the OEM readings and then run the rawsample tool again to analyse the results?

Many thanks again,

Phil

I'm not sure whether that is good news or bad! The readings show that with the low load, the power factor is indeed rubbish. (That was the point of asking for the current as well!)  V=243.54  I=0.94  W=158.09  implies 228.92 VA and so pf = 0.69 which is not unusual for a small induction motor in a fridge and/or computer/TV power supplies. Of course, the Owl (as I understand it) will show 228.92 W in this situation, but the tariff meter should be agreeing with the emonTx.

How's this for a theory: I think you read the true real power when you checked/timed the meter, by the time you'd got the Owl set up, a fridge/freezer with a poor power factor had turned itself on or off (can't remember which way it has to be - I've lost the plot!), and by coincidence the numbers - power and VA - happened to be about the same?  It's the best I can come up with. If I'm right, it's good news and hopefully, your a happy bunny now. 

Now that's sorted out, and you've already got the voltage calibration done, you can if you wish tweak the current calibration of the emonTx to get even closer to your tariff meter. Just make sure fridges etc and switched-mode power supplies are off when you set up PHASECAL!

Robert,

You could well be right about the circumstances leading to inaccurate readings. I do think though for peace of mind I will just try turning all the electrical equipment in the house on one at a time to see if any single item could be causing a problem. And yes PhaseCal is next on the list... :) But see below...

Rae,

I was thinking about this last week. While the house fuse is 100kW, I have nowhere near that capability. Using all the big stuff in the house (cooker, microwave, kettle, dishwasher, washing machine, immersion heater and vacuum cleaner ) I couldn't pull more than 21kW. Realistically, I've never seen more than 10kW being used. I've got the standard 18W burden resistors fitted to the emonTX, but I'm pretty sure I've seen finer resolution than 45W though...  And some questions if I may - Do the 18W resistors equate to 50kW full scale? What do 150W resistors equate to? My PV system is just under 4kW, what resistor do you recommend for that? I assume it'd be OK to fit different resistors for different CT's as long as they were calibrated individually?

And I thought I'd nearly finished this project :)

Phil.

I presume you mean your supply fuse is 100 A (the UK standard, not 100 kW, which would imply a 400 A fuse, which would carry 700 A for about 90 minutes without rupturing!). If you do have 400 A c.t's, I think a rethink is needed.

So I can't answer your questions about the burden etc, because I don't know which c.t. you are using. You need to either work through the calibration yourself, or tell us which c.t. and the current ratio. The standard shop c.t. is 100 A : 50 mA, so a 18 Ω burden resistor gives 900 mV per 100 A, which leaves a bit in hand. (As a rough guide, full scale is 1.1 V when you allow for tolerances.) The calibration constant turns that back into a current.

You can increase the value of the burden, but you must derate the c.t. proportionately. (It's the c.t. VA rating that counts.) 

Phil, you asked me: What do 150W resistors equate to? My PV system is just under 4kW, what resistor do you recommend for that? I assume it'd be OK to fit different resistors for different CT's as long as they were calibrated individually?

I did some trials recently in which a 1250W heater was measured with the standard YHDC CT and 150R burden.  With three coils around the CT, my results using RawSamplesTool_4ss_2.ino showed that the system was close to its limit, this being with an equivalent power of 3750W flowing:

|     c          v                      .                                       |
|  c          v                         .                                       |
| c         v                           .                                       |
| c         v                           .                                       |
|  c       v                            .                                       |
|  c       v                            .                                       |
|    c     v                            .                                       |
|         c    v                        .                                       |
|              c    v                   .                                       |
|                   c   v               .                                       |
|                        c   v          .                                       |
|                              c  v     .                                       |
|                                    cv .                                       |
minVoltage 150,  maxVoltage 872,  minCurrent 36,  maxCurrent 987
 

When a fourth coil was added, thereby resembling a power of 5000W, the ADC was very clearly hitting its limits.  Any results taken with a system in this state would under-read significantly.

|                        c         v    .                                       |
|              c               v        .                                       |
|      c                   v            .                                       |
c                       v               .                                       |
c                    v                  .                                       |
c                 v                     .                                       |
c              v                        .                                       |
c            v                          .                                       |
c            v                          .                                       |
c           v                           .                                       |
c           v                           .                                       |
c            v                          .                                       |
c               v                       .                                       |
|  c                v                   .                                       |
|          c            v               .                                       |
|                  c         v          .                                       |
|                           c    v      .                                       |
|                                    vc .                                       |
minVoltage 164,  maxVoltage 857,  minCurrent 0,  maxCurrent 1023
 

From these results, it seems that a 150R burden would not provide quite enough range for a 4 kW system.  You would be better off using a slightly smaller one, say 120R, which would allow your system to cover a greater range, albeit with slightly reduced sensitivity.

If you wanted a system which can monitor all the power that your house can consume, then you would need to drop the burden's value still further.  But if your sole purpose is to divert surplus power as efficiently as possible, then I would use as high a value of burden as you can get away with.  For a 4 kW PV system, this appears to be around 120 R.

And I thought I'd nearly finished this project :)

I know the feeling!

Robert,

I need to remember that I shouldn't write technical stuff when I'm tired and have the flu... You are correct, I meant 100 Amp fuse, not 100kW. This of course equates to 24kW (ish) of draw. My CT's are the ones (YHDC SCT-013-000) the OEM shop sells at the moment. Given that I very rarely go higher than 10kW, can I just adjust the main 18W resistor to rescale it and adjust the secondary resistor (for the PV) to rescale it to say 5kW? Or will I need to change the CT's as well?

Phil

Been there, done that, DON'T do this while you're unwell. At best it'll cost you money. At worst you might get hurt.

What worried me was the numbers you posted earlier suggested you had calibrated for a 200 A c.t., and that's why I gave you the formulae to work it out for yourself. Provided you never go above the powers you indicate (or you are prepared to accept wildly inaccurate values when you do, and for the reason look at what happens to the c.t. waveform when it is overloaded - the pictures are in the report in Building Blocks, where you'll also see a graph showing the maximum burden resistor value), then yes, you can derate the c.t. and multiply the burden value by the same number. So for 46.8 A max (~ 11.25kW), a 47 Ω burden resistor will do (and change the calibration accordingly to 42.553); and for 22 A max (~5.28 kW) a 100 Ω burden resistor will do, with a calibration constant of 20.0.

The get-out clause is YHDC haven't confirmed their c.t. will work like this, there might be more waveform distortion and I have not measured the phase shift under these conditions.

RW: YHDC haven't confirmed their c.t. will work like this, there might be more waveform distortion and I have not measured the phase shift under these conditions.

With the 150R burden that I generally use, the standard YHDC CT produces a 3.3V peak-to-peak waveform at around 4kW as shown above.  This is operating the device well outside its intended range, but it seem to be OK for my needs. 

If I were to use a lower value burden, the output signal from the CT would be correspondingly smaller, and less of the ADC's input range would be utilised.  This would result in less accurate performance at low levels of power.  Boosting the CT's output signal has always seemed to best way to go, but my initial attempts to do this were ineffective so I've not continued down this route.

When only "real power" is required, any phase shift that's introduced by the CT (or anything else) has very little effect.

"any phase shift that's introduced by the CT (or anything else) has very little effect."

But that's only true when the power factor of the load is close to unity anyway. As the load power factor gets worse, the transducer phase shift contributes increasing errors. Take two sine waves, voltage and current, displaced by 53°. The power factor is 0.6, representative of a small induction motor. If you add a 5° error taking the shift to 58°, the power factor becomes 0.53 and the real power you measure has dropped by approx 12%.

OK, Robert, I'd better stick to kettles and immersion heaters :)