Fail of the Week: Inaugural Edition

fail-of-the-week-inaugural-edition

We’re sure you’ve all been waiting on the edge of your seats to see whose project makes it as the first Hackaday Fail of the Week. Wait no longer, it’s [Mobile Will] with his woeful tale about monitoring AC current usage.

He had been working on a microcontroller actuated mains outlet project and wanted an accurate way to measure the AC current being used by the device connected to it. The ADE7753 energy metering IC was perfect for this so he designed the board above and ordered it up from OSH Park. After populating the components he hooked it up to his Arduino for a test run, and poof! Magic blue smoke arose from the board. As you’ve probably guessed — this also fried the Arduino, actually melting the plastic housing of the jumper wire that carried the rampant current. Thanks to the designers of the USB portion of his motherboard he didn’t lose the computer to as the current protection kicked in, requiring a reboot to reset it.

We can’t wait to hear the conversation in the comments. But as this is our first FotW post we’d like to remind you: [Mobile Will] already knows he screwed up, so no ripping on his skills or other non-productive dibble. Let’s keep this conversation productive, like what caused this? He still isn’t completely sure and that would be useful information for designing future iterations. Update: here’s the schematic and board artwork.

We’ve got a bit more to share in this post so keep reading after the break.

There were a ton of submissions, which we’ve filed away for future use. Our plan is to pick one a week (the selection process is pretty arbitrary, sorry!). It was hard to pass up fried circuits for the first post. But fails can include projects that just didn’t turn out like you wanted and much more. What are you waiting for? Write up your fail and send it our way.

We’d also like to mention that there are a couple of niche websites that already specialize in these types of engineering fails. [Daniel] wrote in to tell us about what he calls his “new-ish” site Mistake Engine. Another regular reader whose projects have been featured many times here at Hackaday also chimed in. But [MS3FGX] so frequently writes about his efforts that didn’t make it that he named his own blog DigiFAIL.

That’s it until next Thursday. We’ll see you then!

Comments

  1. syntroniks says:

    Without schematics, it may be difficult to diagnose his issues.

  2. Neoxy says:

    Did the same… but was working with MSP430F2013 and I fried the chip, programmer and my whole computer… Only thing that survived was the hard drive. Everything else just went out with a loud pop :D

  3. Raidcore says:

    Thank you Mobile Will for sharing with us. I await the solution!

  4. Dylan says:

    I’m proud of him for trying, it doesn’t say anything about a fire or other dangerous situation. I’d call it a learning experience. You can’t learn if you don’t make mistakes. Hopefully it’ll work at some point if he tried again. I’ve been considering something like this to see where all my power goes and what I can unplug when not in use to save it, or even better a remote circuit or something to control my house for me.

  5. Bob says:

    No clue as to what was wrong with it, no details for others to solve, no moral to learn. What exactly is the point of this feature?

  6. mobilewill says:

    Hi all,

    I will post the schematics tonight on my github http://www.github.com/friedcircuits

  7. Luke says:

    I suspect improper grounding and handling of the mains neutral line.

  8. D. Schultz says:

    The measurement circuit is connected directly to the AC line which means that there must be some isolation between it and everything else. Optocouplers on the digital data lines for example.

    Without a schematic for the circuit in question all I have to go on is the example cited which has a jumper (JP3) which connects the AC neutral to digital ground.

    It seems likely that the computer power supply has a connection between safety ground and its output power ground. Now there are two possibilities when you connect the measurement circuit to AC:

    1) You connect the neutral to digital ground: Not much exciting happens here but you still have a dangerous circuit.

    2) You swap the wires and connect AC hot to digital ground. This is exciting because you have short circuited the AC power line. All that current flowing in places that weren’t designed for it. Bang.

    • Starbuck says:

      #2 gets my vote. Simple “dumb” mistakes are often the most painful.

    • mobilewill says:

      I triple checked that I didn’t swap the lines. My thought is I just isolation. The ADE7753 wants a <1v(can't remember exactly) or so AC signel to see the sine wave.

      • D. Schultz says:

        I simplified the first case a bit. If you have any load at all on the circuit (even in the next room) it doesn’t apply.

        Current normally returns to the circuit breaker panel through the neutral wire. But you have provided an alternate path by connecting neutral to safety ground. (neutral and safety ground are connected to the same bus in the circuit breaker panel) So the return current will take both paths with the split being determined by the relative impedances.

      • A Phillips says:

        If you are sure you didn’t swap the lines I’m guessing its either bad grounding… or its C2 and C3. What kind of voltage are those caps rated for?

      • A Phillips says:

        The absolute maximum ratings for V1P and V1N are +/- 6V to ground, thats most likely the problem.

      • Mystick says:

        Mains AC usually isn’t a clean sine wave… through transmission mediums it picks up all kinds of harmonics, spikes, and irregularities…. it’s what happens when you have an initially clean signal distributed and connected in parallel to millions of random noise sources, variable-current-draw loads, long transmission lines with varying impedance,..

        That’s why they have power conditioners and phase converters.

      • t-bone says:

        Would swapping hot and neutral have mattered in your case? Half of the sockets in my house were wired in reverse like that.

    • Rollyn01 says:

      I’ll agree somewhat with #2 as well, but for a different reason. I see a voltage divider in his design but I can’t figure what it’s there for unless one of those resistors are really a diode to add some form of rectification. Other than that, and if the ratio on the transformer is as stated on the data sheet, he bump down the voltage without properly adjusting his circuit to handle the incoming spike in current. That fast-rising current would definitely fry everything. Maybe he should have added some caps to help add some impedance to match the impedance of the transformer. Just a thought though.

    • A Phillips says:

      I agree with the improper grounding now, here is an application note on the ADE7753, there is an example application on page 12.

      http://www.analog.com/static/imported-files/application_notes/an564.pdf

  9. mobilewill says:

    Schematic is coming, give a few minutes to upload.

  10. nah! says:

    why is this rubrik calles hack a day fail and not hack a fail

  11. Hi all! I work with this kind of solution and it’s FULLY NECESSARY to use galvanic isolation when you are connected to the mains and then plugged into the USB.

  12. bsg says:

    Somewhat off-topic, but is there some sort of USB isolator, that would sit between one’s project and the computer to prevent damage propagation? Maybe a board with 4 optoisolators or somesuch? While [Mobile Will]‘s computer protection kicked in, i’d rather know that there’s something we can do, as opposed leaving things to blind luck. Thanks!

  13. Scuffles says:

    Hmm I have never really played with mains power (partly because I’m *Bawwwk Baaaaawwwk* chicken) but doesn’t look to me like any of those components are really designed to handle serious (mains) current. I would have thought there would have been some beefier non SMT components involved? maybe some optoisolation?

    But I am speaking from complete inexperience and truth be told I avoid SMT like the plague because I have big clumsy hands :(

    I am glad to hear no one was hurt and your computer survived. As I have a project coming up that will involve voltages and currents outside my comfort zone I am quite interested in finding out what went wrong. Just remember we learn nothing from success :)

    • No problem to use SMD components with high current. The point is the isolation itself. I’ve designed a little box using one 1:1 transformer to isolate and protect my PC. I’ve used some SMD shunt resistors that holds 80amps or even more!

  14. mobilewill says:
    • Roger Wolff says:

      Dear Mobilewill,

      Your schematic indicates that you’ve connected mains voltages to the low-voltage lines on your breakout board.

      At best you have the power-N line connected to the GND of your board. This should be OK, provided your system (PC/arduino/sound-system/whatever you have connected) doesn’t connect the low-voltage ground anywhere. As things DID blow up, I’m guessing someone somewhere had something grounded. In theory no current will flow as the N line should be at the same voltage as your GROUND. But in practise, having something connected that draws some current, will result in a voltage difference of a few volts with a very low impedance. i.e. a large current will flow.

      Worst case, you inverted the power line. That means you connected the “live” pin of the socket to the ground of your PC/arduino/whatever. Again, things blow up if one of the connected systems grounds the low-power. In this case, blowing up is the “good” outcome: If things are NOT grounded, you now have a live 110V (or 230 if you’re on this side of the water) on your low-voltage system that you expect to be safe-to-touch.

      I’ve tested a PC powersupply here and it grounds the low-voltage GND line. (I’m using GND for the negative line of your PC and arduino. I’m using GROUND for the protective ground in your wall outlet.)

      So… in conclusion: The ADE7753 does not provide galvanic isolation between the mains and the low-voltage system that you might have expected.

  15. ino says:

    Isn’t it more simplier to post a PDF of the schematics instead of an obscur link on github ? I don’t even know where (or what) to look for.

  16. tempestion says:

    i’ve seen something similar its generally something like coupling neutral to the DGND. Some supplies are allowed to float on neutral however the digital interface must be isolated.

  17. Buddha Jones says:

    i dig the good vibe and attitude of this tail of fail

  18. Mark says:

    What kind of current sensor was this coupled to?

  19. ILAMtitan says:

    So, what happened here is that you connected the mains neutral directly to your PC’s ground via the USB port. Neutral is considered a high voltage. The ground on your PC was generated from the power supply and when connected to the mains created about a 170V difference between the two. This just happened to be through your board, the arduino, and the USB ports protection circuits.

    If you wanted to talk to this, you would need a separate power supply on your daughter card, and digital isolators in between it and the Arduino.

    • ADIDAIllini says:

      Isn’t the GND on the on computer power supply bonded to supply ground through the plug (and possibly the case)? The main neutral should be bonded to ground at the breaker box. If all the grounds in the home wiring are intact then there should be no potential difference, right?

      • Alez says:

        This is correct. The neutral is at a very close potential to the ground (save for a few volts voltage drop).

        However I suspect that he swapped the hot and neutral wires. This way the hot wire was directly connected to his digital ground. Which would be fine (as in dangerous but blue smoke-less) if not for the USB connection. The ground of the USB is mains earth referenced. Therefore he made a direct connection from the hot wire over the arduino, over the USB to the earth ground. BOOM!

        • ILAMtitan says:

          It’s highly likely that the live and neutral lines were confused, but by no means should neutral be tied to the ground on your computer. There is too much potential for wiring mishaps and stray voltages on neutral.

      • richms says:

        Yes, but any other connection between neutral and the protective ground is expressly prohibited as the neutral will be a volt or 2 off ground due to voltage drops, and also any other path between the 2 will cause a current imbalance, tripping out the ground fault breaker or whatever they call them around your way.

        Basically, if swapping live and neutral around makes your design explode, your design is fundamentally dangerous.

    • Ralph says:

      You are wrong, do not post when you are so ignorant, especially when you could trivially go and measure such a thing.

      The voltage between the neutral wire and PC ground is within a couple volts, though that shouldn’t be relied upon.

    • Harry says:

      Unbalanced loads on 3 phase can cause high voltages to appear between neutral and earth ground despite proper bonding. I got zapped installing some signs and despite the power being shut off at the breaker and 0 volts between hot and neutral there were 270VAC between neutral and ground! Isolation is absolutely necessary when working with mains power.

  20. nrrdzilla says:

    Am I looking at the wrong board layout? I’m seeing a bunch of what looks like missing ground connections at every capacitor, and more alarmingly at R5…

  21. Douglas Bouttell says:

    I’ve seen a similar problem before, however this system was protected by a top notch power system. Basically, he earthed it in a very round about way.

    USB is Earth referenced, it is not a floating ground. So when you connect your ground to the cold/neutral line on AC mains. POP! Suddenly your AC has a path to Earth and pushes several Amps through a flimsy wire and it fries everything.

    Your “Ground” plane of your board is actually a neutral plane. Which might explain the massive charring on the test point shown in the image. You connect that up to the “Ground” (or Earth) of the Arduino and sparks will fly!

    Moral: When working with mains, ALWAYS ISOLATE! Use optocouplers or transformers. You can also get some pretty awesome current transducers which are completely isolated and need a couple of resistors

  22. Simbo says:

    First off, I’m no electronics engineer, for me this is just a hobby!

    My best guess is after looking at the datasheet for the ADE7753. pins 6 & 7 (V2N & V2P) are for use with a voltage transducer, from MobileWill’s schematic I can see 110v AC was connected through a divider to pin 7.
    from the datasheet: V2N, V2P: Analog Inputs for Channel 2. This channel is intended for use with the voltage transducer. These inputs are fully differential voltage inputs with a maximum differential signal level of ±0.5 V. Channel 2 also has a PGA with gain selections of 1, 2, 4, 8, or 16.
    The maximum signal level at these pins with respect to AGND is ±0.5 V. Both inputs have internal ESD protection circuitry, and an overvoltage of ±6 V can be sustained on these inputs without risk of permanent damage.

    BUT then i saw their test setup and the only major thing different is H- is also connected to ground

    I’d really like to find out why this failed and think this is a great new feature of HaD

    Thanks for sharing MobileWill

  23. popeiler says:

    Several of the others have mentioned that the issue is likely grounding related. In cases like this, I usually recommend using one of the ADUMxxxx isolators that Analog makes. I have successfully designed this into several industrial temperature sensors.

    http://www.analog.com/en/interface-isolation/digital-isolators/products/index.html

    They give you something like 2500v of isolation, are faster, and use less current than optical isolators. The key when using them is to put a bigass isolation gap in your board on all layers. The isolation gap goes right under the ADUM chip.

  24. Dan from Mistake Engine:
    Thanks for the mention! I noticed it when my google analytics said that most of my referrals for today come from hackaday.com.

  25. Douglas Bouttell says:

    Furthermore to my comment, you seem to misunderstand HOW this circuit is supposed to measure current. As a lot of you know, a wire generates an electrical field around it when current flows through it, the current transducer on this board is designed to have a wire poke through the hole and the sensor detect the current that way.

    http://www.analog.com/static/imported-files/eval_boards/ADE7753EB.pdf

    Figure 3 of the Eval Board Datasheet will show you how this input stage will work. AT NO POINT SHOULD AC TOUCH THIS BOARD. You take a wire, and poke it through the hole in the transducer.

    • mobilewill says:

      I read over the Eval Board Datasheet for hours. What about the part that the IC needs to “see” the AC sine wave? Is there where an isolator should be?

      I had the wire through the CT sensor for current, but the IC still needs the sine wave imput.

      • Douglas Bouttell says:

        Having the wire through the CT sensor will cause a sine wave to be induced within it (i.e. the magnetic field caused by the sine wave in the wire, will cause the coils of the CT to pick that up). On your board you are connecting your Neutral/Cold line to Ground, which you don’t need to do.

        I recommend that if you get another board, have your Hot/Live connected to the arduino’s 5V, your cold/neutral line connected to the arduino’s ground, and a single core wire (no two/three core flexes) going through the CT to measure current. That should be your problem fixed, and will definitely prevent another bang from happening.

    • Matt C says:

      If I read the data sheet correctly, it wants to measure the current AND the voltage, so that it can properly calculate power usage for resistive and inductive loads. The V1N/V1P channel is for hooking up to the CT sensor, and the V2N/V2P channel is for measuring the voltage – the data sheet talks about doing it via a “voltage transducer”, but it seems like most of the other designs just use a resistor voltage divider.

      I think the real issue is that “neutral” is not “ground”, and connecting the two together through the PC’s USB port created a ground loop: https://en.wikipedia.org/wiki/Ground_loop_(electricity)

      • Douglas Bouttell says:

        I’d try that first, its late and I dont have my EEE head on right.

      • MRE says:

        Voltage transducer = stepdown transformer. For example a 3 volt AC wall wart. The key is that it be ac and unregulated. The transformer offers both isolation and voltage reduction.

        It kills me when I see people trying to follow the guides in these various current monitor ICs. There is the cheap parts version wherein they illustrate directly connecting ac for both voltage (via divider) and current (via shunt), and the practical and safe version using CTs for current and a transformer for voltage. Always ALWAYS use the later.

    • ILAMtitan says:

      This is incorrect. The chip measures power, not just current. In order to do this, it must measure the mains voltage as well. If you look at fig 6, you’ll see the voltage channel inputs with a sufficient attenuation network. The issue on this board comes down to “what is ground.” It’s perfectly acceptable to have AC come into this board with the proper attenuation and isolation to any external devices.

  26. Matt C says:

    I agree with others that it has to do with neutral and grounding. The “neutral conductor” is not necessarily at 0V relative to “ground” (which is what your USB port’s ground will be at). It may not be very far off voltage wise, but putting a short between the “neutral conductor” and “ground” will try to pass a lot of current.

    Try plugging a light in, and measuring the voltage difference between neutral and ground at the outlet where the light is plugged in. With a reasonably beefy ammeter, try measuring how much current will flow between neutral and ground here. Note that if this is a GFCI protected outlet, this latter experiment may cause it to trip. Obviously be very careful when doing these experiments.

    • spike says:

      Years ago, i friend of mine was testing his new cheap multimeter (like what you can find at harbor freight) he stuck it in the wall and measured 110v, then he switched it to amps and when he stuck the probes back in the meter exploded. His response was pretty much “I don’t know what i was expecting.”

  27. Jrdoty says:

    Check to make sure the outlet is wired correctly. Sometimes the hot and neutral are inadvertantly swapped when wired. That situation would put mains voltage on the board ground causing all that magical smoke to appear.

  28. Chris C. says:

    Seems the ADE7753 has been more trouble in the long run. It was apparently used because:

    “Using an Hall Effect sensor with an Atmega 328p didn’t work out so well. It was very difficult to get accurate calculations and it tied up the microcontroller from doing other tasks.”

    But periodically processing an interrupt to get ADC readings for 60hz current doesn’t tie up the MCU. Nor is it even particularly expensive in terms of CPU time. As for accuracy, I suspect a large part of that is because a 30A sensor was used – far more than necessary, and range/accuracy is a trade-off.

  29. Tom the Brat says:

    Well, no matter how competent or careful you are in your engineering, there will always be some mistake somewhere that causes disaster the first time you design for mains. Strobe light kit laying on a pair of pliers anyone? BOOM!

  30. Tech in Oz says:

    Hi, I didn’t read all the comments but I did look at the schematic and the specs for the ADE7753. It doesn’t have ANY internal isolation!!! LOL

    So yeah the circuit is “tuned for maximum smoke”.

    There must be absolutely NO path or connection between the high voltage side and the digital.

    (V1P, V1N) (V2P, V2N) are differential inputs so that cannot have ANY path to anything except the sensor. (For the way you are doing this).

    The sensor needs to be a voltage to voltage transformer or a current to voltage tranducer depending on what you want to measure.

    So here’s what to do –

    Case 1) Measuring Voltage
    Use a step down transformer to bring the high voltage down to within the chips input specs. I didn’t read enough to say if it needs to be rectified. Draw a line across the circuit board. Put the primary on one side and the secondary on the digital side. No PCB trace or power plane should cross this line. Keep the the traces (tracks) on the high voltage side as far apart as possible and also far away from the digital traces including any ground/power planes.

    Case 2) Measuring Current
    The same as case but use a current transducer.

    A simple mains transformer will act as a voltage transducer. A current transducer is a current shunt (series link or wire) that is not looped and has no inductance with a sensor coil around it’s axis.

    In the application notes it shows a circuit that has connections between the live side and the chip. What is not made clear there is that the power supplies for the chip are derived from the same live side and so can share a common ground. In these examples an opto couple is used for the digital output.

    What I think went wrong …
    Well first this chip wasn’t designed to be used the way you wanted to use it.
    If you have something like a fridge compressor start in your house it will briefly drop the line voltage to all devices on the same circuit. The voltage drop is a result of the resistance and inductance of the wiring in the house. Both (A)ctive and (N)eutral have the same resistance and inductance. This means that both A and N will change with respect to earth. The change in N is likely to be much higher than the specs for this chip … about 7 Volts max. So pop goes the chip and all of a sudden you have a circuit from the live side to the USB.

    Let me know if you want more info.

  31. Dumb Geezer says:

    The current transformer is designed for switching power supplies. It has a frequency range of 20kHz-200kHz. I don’t think the mains wherever you are operate quite that way, so it might not work just the way you want. Your load resistor is higher than specified, so you would get more output voltage, if your signal was in spec. There are many parts from Pulse that are specd to operate at 50/60Hz. Might want to try one.
    And do be careful with the grounding, I think a voltage transformer would be a safer choice.

  32. exHomunculus says:

    On JP3 on your pcb design, is neutral tied directly to the hot line?

  33. Steve Marple says:

    Your schematic worries me. You’re mixing mains voltages and 5V on one PCB and I want to see clearly where and how your separate them. All I see is that LIVE and GND disappear into JP1. Even if you get the circuit right you need to ensure mains voltages can’t jump to the 5V side. How about laying our your schematic with mains voltages on one side and 5V on the other? Indicate your separation boundary with a dashed line. Then layout your board so that line becomes your physical isolation boundary. There’s plenty of examples of cheap and dangerous power supplies which fail to do this (eg see http://www.righto.com/2012/03/inside-cheap-phone-charger-and-why-you.html) and I bet your board is the same.

    In the UK our plugs are keyed, with correct wiring live and neutral cannot be reversed but AFAIK US and Euro plugs are not keyed, so live and neutral can be reversed. Your circuit needs to be able to handle that possibility.

    • Reg says:

      Current US plugs and outlets are keyed, but it’s not uncommon to find hot and neutral swapped. They sell testers w/ 3 neon lamps for quickly checking outlets. Older plugs and outlets were not keyed and typically lacked the safety ground. I routinely encounter miswired outlets. Worst was a 220 which had each leg on a separate breaker. Ever since I check for voltage on both wires AFTER I drop the breaker before I touch anything. I didn’t get shocked, but dumping a 30A arc with your hand inside the box will give you quite a start.

      There are several issues to address:

      transformer isolation of all sense signals from the mains

      optoisolation of sense circuits from logic circuits

      physical separation of sense and logic sections

      use a GFID when testing things like this. It might not save the device, but it will help protect you.

      Note: There are lots of application notes for smart meter designs. You should read a bunch of them.

      • Frode says:

        In some places of the world, the mains outlets don’t even have any Neutral. Where I live, we have 110VAC on both pins compared to earth-ground, where one line is offset by 180 degrees (making the difference 220VAC, as expected). Otherwise, I’ve heard that in some areas where the mains do operate with Neutral/Live lines, the Neutral isn’t always completely earth-ground either. The dedicated Ground line on the outlet, when present, should on the other hand always be earth-ground.

        Worst case of misswiering an outlet I’ve heard about was when my parents house was buildt. The electrician mixed one of the live contacts with ground, so 110VAC was constantly fed to the ground terminals throughout the entire house! They got it fixed quite fast after it was discovered.

    • Gerald says:

      I recently noticed that even with keyed plugs, a lot (most?) of outlet multipliers have some of their outputs reversed. I even found a simple extension cord (i mean prebuilt one) with reversed output! I thought that respecting live/neutral polarity in such commercial products should be mandatory, but it seems it’s not. And not only with some kind of chinese crap, but also with notorious brand like “Legrand” in France (which does have a “NF” security certification).

      I’m wondering what are possible risks of having reversed live/neutral polarity, except of course having live present on the wrong line and not cut off by switches).

  34. rascalangel says:

    You should have connected the neutral line to ground of R5 only. While all the rest of the GND’s should have been connected to USB’s GND.Then it would have worked. Since you shorted these two GND’s during negative cycle the GND for the IC’s etc was at 156V!!!

  35. Beau says:

    I noticed you used +5V to AVDD and DVDD for the first schematic, but in version 2 u have 5+ to AVDD but for the DVDD is is going through a 10ohm resistor first (your VCC source). You are gonna have up to 500mA going straight into DVDD. What’s the max current allowed at this pin?

  36. bootfetch says:

    I think the 10Ohm there is for isolation

    I can still remember the pain those Current transformers caused me when I made my undergraduate thesis

    • Beau says:

      I just checked the datasheet only supposed to be 4mA Max to this pin. He should increase the resistor to a more suitable size >1.25k, at the moment up 500mA is going to the pin. Also, that resistor will cause an issue if the internal resistance is not much, the datasheet shows that AVDD and DVDD can’t have more than +- 0.3 V difference between them.

  37. Luke says:

    I can’t find any mention of the voltage rating of R2 at start up this could have the full peak AC voltage across it, and as mentioned could be even be a spike.
    Perhaps use through hole parts to give better creepage and clearance or multiple resistors in series. A MOV/transorb with PTC fuse wouldn’t go a miss either.

    There are situations where the electronics need to be “hot” electrical installation testers, PAT testers and even multimeters for example. These devices tend to be battery powered with the charger and communication interfaces being isolated. One option would be to develop with battery power and RF link to the PC.

    You could use the incoming mains to power the circuit, Capacitor step down might be an good option for a low current always on circuit like this.I found this tear-down useful when i was looking into power monitoring – still in virtual project stages http://jeelabs.org/2012/03/13/td-kaku-remote-switch/ and there are a number of people on the open energy monitor forum developing similar devices http://openenergymonitor.org/emon/forum

  38. John Jackson says:

    Grounded = BAD as others have said.

  39. drew says:

    So as everyone has said, quite possibly a ground issue. However let me suggest this: what is the voltage rating of the divider resistors? If they are only rated for 50V (or less) as many SMD parts are, then you have functionally connected mains directly to a part with a 1V input. It will quickly burn through everything after that.

    Also, isolation is nice but isn’t necessary. I mean it is necessary and there needs to be physical separation between the 110 and 5 on the board… BUT the circuit should work without it.

  40. Cliff Miller says:

    A small transformer for the voltage input to the 7753 isn’t necessary, but it would be helpful – you could ground the secondary properly to match the circuit board ground, eliminating that variable and you could re-calculate the voltage divider to go with the secondary voltage of the transformer if the ratio wasn’t 1:1.

  41. tomh says:

    The toasted PCB in the picture is not the artwork on the github site.What else is different besides the positions of the Ct and the mains connection being swapped?
    When working with the mains the keyword is isolation, isolation, isolation. As has been observed a transformer between the live wires and connections on the board or let the board float on the line so to speak by making sure the boards power supply is isolated and use opto’s to couple the digital signals out. By all means check that the wiring you were hooking up had the correct sense. Line was the line, neutral was neutral, earth was earth. I’ve also seen buildings where some plugs were wired correctly and some weren’t. Get across the wrong ones you might find one neutral may be anothers line. Perhaps you could post more detailed description of what exactly burned up. If the divider resistors are Ok then it was probably a ground/neutral issue. Ground/earth and neutral should be at the same potential, but if it were me I’d leave the earth connection off the board. A quick check with a meter for a potential difference between the line and the board and the neutral and the board before hooking them up would probably have saved the day.

  42. Tech in Oz says:

    I read through the first three documents on this page –

    http://www.analog.com/en/analog-to-digital-converters/energy-measurement/ade7753/products/product.html

    In AN-564 page 12 you will see an example circuit.

    Note L1, L2 of the meter circuit and L1, L2 of the PSU.

    Note TP6 of the PSU and a note in the meter circuit that says –
    “Connected to TP6 of the power supply board”.

    All of these inductors are necessary to decouple the the Analog ground from the digital ground.

    Your circuit just has one ground with no decoupling. This will case very high currents back to the PC where the digital ground is connected to earth.

    It seems that the IC can be calibrated to correct for phase error between the voltage and current phases in the load.

    I have seen many many cases where simple ground decoupling is just not going to work.

    If this chip can in fact correct for phase error between measured current and voltage phases then I strongly suggest that you use a simple transformer for the voltage sense input. I will save you so much trouble in the long run. Then you can forget about all the decoupling issues and the problems they cause and just use one common digital ground.

    In these documents there is a section on calculating a value for burden resistor. You will need to read this for the burden resistor on the secondary of the voltage sense transformer.

  43. Monomo says:

    Hi

    Maybe it helps.

    I have noticed you have used the “Test Circuit” from the Datasheet.
    It look like you have no Ground on the CT2 – R1 node. So I thing your Resistor R3 and R4 are not abel to limit the Voltage because the Circuit is not closed. So the Voltage takes the Way over the IC and fries everything with the Ground Reverence with the Ground of C3 and C4. (Adruino, PCB). Your PC survived because of his own Mains Ground Referense and the fast death of the PCB. The Resitors R3 and R4 only worked as Current-Limiting Resistors, but not many milliamps were needed to frie your IC with high AC Voltage.

    The Voltage over CT1 and CT2 is isolatet to the Mains Voltage.

    I hope I could help.

    Monomo

  44. AMS says:

    Just another thing: usually 0805 or smaller is only good for about 200V creepage. If you’ve got 230V mains, you’re seeing almost 400V peak relatively often. Replace that 560k with two or more physical resistors in series.

  45. Malcolm says:

    Suggest dropping 110Vrms across a single SMT resistor may result in flashover. Try multiple resistor string to limit voltage drop across each resistor, or preferably an isolation transformer if you want to live…

  46. The relevant safety standard is 60950.1 Creepage, clearance, voltage ratings from a primary circuit to a SELV circuit should be implemented. Re-inforced insulation to SELV, and Basic to earth is required. The standard is now harmonised across US, ASNZ, and EU.

    Read the standard. Get a bootleg copy if you have to. Understanding standards its the main game if you get into electronics professionally.

    From a topology point of view, an ADC on the primary circuit with a digital isolator would be the go IMO. You would also need to provide isolated power, or build a mains supply. The isolated supply is the go for this type of experiment. Recom, Mornsun, meanwell all make a variety of products that are suitable.

    Avago make some other nice analog isolators that you could maybe use instead of implementing the digital isolator and ADC separately.

    It is also possible to accomplish acceptable levels of isolation that are not galvanic using large resistors, and more than 1, i.e. 2x 10M (go big, like 2152) in series, then divide it down as was attempted… not 100% sure on the details but it can be done.

    Protect yourself with earthing if you can. MOV’s, fuses, etc, are there for a reason. Fuses will save your design in fault conditions. 1 on each of the neutral and active lines is the go. MOV’s to earth and between the lines will help save your device if you have surges, or other abnormal over voltage conditions present on the mains. Put MOV’s after the fuses so they blow if there is something bad on the line.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Follow

Get every new post delivered to your Inbox.

Join 96,311 other followers