Archived Forum

I'd like to find a safe way to ditch the transformer completely and use a resistive divider. That gets rid of SO many problems (phase errors, wave distortion, accurate zero detection with non-symetric waveforms, etc.)

There's just the nagging problem of lack of galvanic isolation, which many find important!

Is there another way to make a resistive divider safe?

P.

Paul, some months ago, I spent a while playing around with a 240VAC resistive divider and have posted a few results using this arrangement.  Albeit rather scary (no earths, otherwise the house would trip!), the rig worked fine but its performance was found to be no better than when using a transformer.  For calculating/diverting real power, I now feel confident that it's not worth going down this route.  PM me if you'd like to exchange ideas about this.

When dealing with more complex aspects such as Power Factor, or the current signatures from individual appliances, it may be more beneficial to use some kind of direct measurement.  But that's only my unofficial opinion, not that of any panel :-)

For Mk2a type applications I'm sure a transformer is fine, but OpenEnergyMonitor is about much more than that (sorry!). At its heart it's a measurement system, and as such accuracy is important. 

P.

Sorry - that sounded harsh and a bit grumpy. Wasn't meant that way!

P.

At its heart it's a measurement system, and as such accuracy is important.

 
I understood that the initiators of this project wanted it to be non-invasive and safe (for amateurs to install). If you drop these requirements, then you are free to use resistive dividers and shunt current measurement.

And there might also be legal problems (liability questions and loss of insurance) if you as an amateur 'work' on the electrical wiring or if you connect DIY-equipment to your installation (this would at least be a problem here in Germany).

Using a ready-made plug in AC-adapter (with isolating transformer!) and CT clamps is 100% safe regarding these issues.

For my own project, I definitely need galvanic isolation.

Robert ran some simulations of using a capacitive probe to detect voltage (instead of an AC-AC adapter).  Robert reported that: "Simulation shows that the phase angle [for a capacitance probe] is quite possibly better than the AC-AC adapter - certainly it doesn't change with the applied voltage"

So a capacitive probe might provide better measurement of phase angle.  (A phase shift isn't too much of a problem if it's a constant shift across the voltage range so we can easily compensate for it).

Are there "better" transformers that we could consider?  (by "better" I mean minimum phase shift and maximum frequency response)?  e.g. a toroidal transformer?

That's what the Eco-Eye device does. I think it assumes a fixed voltage, which is kind-of fine I guess because for most of us the voltage won't be fluctuating by more than about 1%. The change in phase angle (detectable by the foil) is much more important, and helps to convert the measured VAs into the Ws that our electricity meters (and bills!) understand.

Don't get me wrong - if you CAN measure voltage too then great, but I reckon phase angle CAN be more important, and doesn't need a transformer.

P.

The Eco-eye monitor requires the user to input the value of their mains voltage; this value is then used for all power calculations. 

While developing my Mk2a code, it became very evident that our mains voltage can vary considerably during the day.  If I calibrate my system correctly at 3pm, its power values are likely to have shifted by several percent by the evening.  Supply meters take voltage variation into account, so for accurate measurements of real power I feel sure that voltage should be measured as well as current.

The effect of introducing time-shift into the voltage waveform was surprisingly small  when I measured it a few months ago (between the 4 and 8 minute points on the video).  Although PHASECAL has a very minor effect when calculating real power, it will more directly affect any measurements that are made of Power Factor and the like.

I expect Robert will reply soon himself but his simulations suggest that a capacitive probe should be able to measure voltage (as well as phase).

OK, sorry I've been otherwise occupied so far today....

The capacitive "transformer" (as it is called by the Power Transmission people) is generally and I think exclusively used on high voltage systems, so measuring the voltage with it is no problem so long as the physical arrangement remains unchanged, that is it is dimensionally stable and it is properly screened to remove changes in capacitance due to the proximity of other objects. (The high voltage capacitor is normally an integral part of the transformer or switchgear bushings - look up the links in the post that Jack quoted).

While I don't have an Eco-Eye and I don't know their design philosophy, I'd guess that they don't use it to measure the voltage because they have no control over the mechanical arrangement of the foil capacitor.

I don't see why a toroidal transformer should necessarily be better than a conventional E-I one. As I've said many times, the designer's art is to get the maximum output of acceptable quality for minimum cost. If you're rectifying the current to get d.c, the waveform isn't at all important, neither is phase shift. Neither (usually) are losses - they're the customer's problem, so the designer goes for the minimum amount of copper and iron to give him the best profit.

The only alternative that I can think of is that you could have a custom wound transformer designed and made. I have one - I believe it was the phase reference transformer out of a thyristor drive, and it has significantly less phase shift and distortion than any of the adapter-type transformers that I've seen. And for its VA rating, it is positively huge. Clearly, it is working way down the B-H curve where non-linearities are small. It is undoubtedly British made, dating from the late 1960's, but has no maker's name on it. I don't know of a firm that might be prepared to do this, nor the likely cost.

Jorg's point about safety is spot on. I understand that many emonTx kits go into schools, and the thought of a bunch of school-children messing about with anything not 101% safe and foolproof doesn't bear thinking about. Which pretty much rules out anything not sealed in a box.

Just a quick note: Texas Instruments have a 7-page PDF on "current-transformer phase-shift compensation and calibration" using a digital finite impulse response (FIR) filter: http://www.ti.com/general/docs/lit/getliterature.tsp?literatureNumber=sl...

I presume this could be easily adapted for AC-AC adapter phase shift compensation.  (I know emonLib does some phase-shift compensation using linear interpolation)

I've quickly scanned through the note but it doesn't appear to offer a solution to the problem of the phase shift changing with load current (in the case of the c.t.); or system voltage (in the case of the v.t.). The inherent problem is that on the grounds of cost and availability, we're not specifying and using measurement grade parts, but we are expecting measurement-grade results.

Hi!

I'm interested too, on a way to avoid the AC-AC adapter. Yes, I know is very dangerous for -not electrician- people. I've found this isolated linear sensing ic: HCPL-7520

http://www.avagotech.com/docs/AV02-0956EN

Thinking of a "PRO" Oem Version with 3 current inputs and 3 voltage inputs.. will be amazing. It's possible?

Regards

For those confident and competent enough to build one, I think that I may have found the answer.  Not that it is cheaper than the Mascot or any safer than the Mascot but it should be as safe as the Mascot transformer solution when in use, if correctly and carefully built.  It will be a design based on the Vishay IL300.

This is my first stab at the circuit.   I intend to order the IL300 shortly and I think it will be appropriate to change the circuit in order to get the output dc biased in the right place.  To do this I have decided that the drive to the LED should be inverted then the 'secondary' can be inverted too. None of the secondary side circuitry needs (or arguably should)  be included in the same enclosure - it belongs with the cpu. 

Last week I thought I would try the idea 'in-principle' by using two conventional opto-isolators with LED's in series. It worked but the linearity was poor (to say the least) and the dc drifted with temperature !

I probably should not be suggesting that anyone tries this but I will let you know how I get on in a week or two.

Edit:  I seem to be having some difficulty presenting that .pdf file

I've just noticed that the Mascot adapter is no longer stocked in the shop, it's been replaced by a different brand for which no test results appear to be available :(

Did you tick the "List" box after attaching your .pdf file?  If not, you can always Edit your post and do so later.

Thanks Robin.  Its there now.

"I've just noticed that the Mascot adapter is no longer stocked in the shop, it's been replaced by a different brand for which no test results appear to be available :("

If somebody would like to send me one, I'll be quite happy to run the tests on it and write another report.

Later this weekend I hope to publish on a new thread my solution for a transformerless Galvanically Isolated Voltage Monitor

Think I will call it  'GIVmon'   :-)

Rob, I await your test results with interest.  In my limited experience, the quality of the AC/AC adaptor makes minimal difference to the ability of an energy-bucket type router to divert surplus power correctly.  The one I use for heating our water is the worst of any that I've tested; its distortion is far worse than the Mascot. 

The only reason I'm still using it is that I don't believe any other one would give better performance.  Tests a few months ago with the same adaptor showed a good degree of import/export balance when I temporarily fitted a disc-style meter, and that's all that really matters for this application.

My No 1 rig has a pair of Rich's PCBs, along with a smart (donated) pcb transformer.  In due course, that rig will hopefully end up in a box on the garage wall.  In the meantime, it does an excellent job of checking out any new builds that I post on the forum.  It's only our DHW that has to manage with the original breadboard lash-up!

The GIVmon is all very well, I know it's possible to do better in a variety of ways, but it's not a viable option for many, for all the obvious reasons.

Yes, test results are everything.  AND My MFT1502 is calling for another 8 AA batteries ....

Hmm...  I was hoping to do a 1KV insulation test between input and output but my model won't do it  even with new batteries for some reason.  Will have to be satisfied with 500v dc for the time being. Yep, that's OK.

BTW   I will update my profile later this evening since for an item like this, folk will need to be satisfied with my credentials!

I think it would also be appropriate for me to send my prototype to Robert Wall if he would like to give it a try?  (Maybe in a few days time you will PM me with your address.)  Always good to have a second opinion on these things.  I shall be uncommonly fortunate if nothing needs more attention.  

Enough for the moment I'm sure many of you would like to see it.  Will have to make do with stills for the time being.  I've never done utube(?) unlike you Robin :-)

YouTube is no problem.  It's the concept of "credentials" that's worrying me!