Press "Enter" to skip to content

The Energy Detective TED 5000-G Teardown

Last updated on February 7, 2011

Back when I owned a TED 5000-G, before realizing that it has a critical software design flaw which makes it unusable on my network, I decided to open up the enclosures and see what’s inside. Since it’s a power monitoring system which uses PLC (power line communication, one type of which is the familiar X10) to communicate to a data logger and an embedded webserver I figured it would be fairly interesting.

Here’s the teardown photos, with the more interesting chips called out. Links to full-res photos are in the top right corner of each page:

    · Unopened
    · Enclosure Opened
    · PCB Top
    · PCB Bottom
        – Microchip PIC24FJ64GA004: Microcontroller.
        – NXP TD5051AT: Home automation modem. Used for PLC.
        – Cirrus Logic CS5461: Power measurement device; does the monitoring itself. EOL’d on 15-Dec-2006, Data Sheet Mirror (PDF).

    · Unopened
    · Enclosure Opened: ZigBee Daughter Board plugged into mainboard.
    · Mains Connection: Ground / earth pin is not connected.
    ZigBee Daughter Board:
        · PCB Top
        · PCB Bottom: Note printed antenna near bottom of photo.
            – Microchip PIC24FHJ64GA106: Microcontroller.
            – Microchip MRF24J40: IEEE 802.15.4 radio, used for ZigBee communications.
        · PCB Top
            – Microchip ENC28J60: Ethernet Controller.
            – Microchip PIC24FHJ256xxnnn: Microcontroller. Can’t read entire part number.
            – Macronix MS25L3205D (PDF): 32Mbit Serial Flash
        · PCB Bottom
            – Vossel VS1307Z (PDF): Real-time clock, keeps time with main power off by automatically switching to backup source (CR2032).
            – NXP TD5051AT: Home automation modem used for PLC; same as in the MTU.
    · CTs: Sealed, not easy to disassemble. Likely nothing interesting inside.

I was really looking forward to getting a TED, and save for the PLC quirks and the webserver bug that made it unusable, I really liked the device. The PCBs themselves also appear to be very well made, and I particularly appreciated that they were closed with simple Philips and Torx fasteners and thus easy to look inside of. If Energy, Inc. releases an updated version with the firmware fixed I’ll consider buying another one, particularly if they stop using PLC and move to all-ZigBee (or some other PAN) for communications from the MTU(s) to the Gateway.


  1. Duane
    Duane February 7, 2011

    This thing screams for an open source implementation. Either a dedicated Shiva plug or just run a syslog daemon on something and blast periodic updates. Write a small script to produce graphs or web pages and run a simple web server have a simple PC application that retrieves the file and make pretty Excel spreadsheet and graphs.

    The remote side could use either wireless or even data over power line since the circuits are readily present. Really, a cheap current tap and quick little program to clean up and linearize the current data would be good. All implementations I see are pretty expensive. This really doesn’t look to be that difficult obtain reasonable quality data for voltage, current, and power factor…. but I may be wrong.

  2. TedDev
    TedDev February 7, 2011

    Duane, I’m one of the TED software engineers. We currently publish a couple of different API’s (both poll and post) to allow users to get the information from the device directly. These are both free and available via our website. We’ve had users write iPhone and Android apps as well as a variety of widgets using our polling API. Our post api is newer and allows dta to be posted from the TED to a server). We are currently getting ready to kick off an opensource reference implementation of the serverside component to this, but we already have 3rd parties implementing their own monitoring solutions using this API.

    Most of the time, the PLC issues can be overcome w/ cheap X10 filters that you can buy off the web as well as changing which breaker the MTU is wired to in the panel (decreasing the path to the Gateway). We’ve also had some users pair up the gateway w/ a wifi adapter to keep it closer to the panel.

  3. c0nsumer
    c0nsumer February 7, 2011

    TedDev: I was going to hunt down the PLC issues, but I ended up shipping it back before then. I’d found that for testing it worked off of a particular outlet not far from both my switch, so an extension cord and some network cable did the trick.

  4. Scott Finegan
    Scott Finegan March 31, 2011

    Based on what I can find…

    Cirrus Logic CS5461 is Eol’d

    Cirrus Logic CS5461Axxx is the replacement. The chip in your photo is an A version, and therefore not obsolete.

  5. Ben
    Ben June 1, 2011

    This product is an amazing implementation of cool devices put up together, the pic24 the TD5051A, MRF24J40, ENC28J60 and CS5461!
    From an EE perspective all of those IC’s are cool devices on it’s own arena, pretty much industry standards, what is amazing and honestly inspiring is the way they put it all together it makes you think that this cool device was produced by a small group of engineers not a large corporation that would have user other proprietary solutions.

  6. Mark
    Mark January 22, 2012

    Does anyone know how the TED Current Transfomers work? Do they each have built in Burden Resistors? Does anyone know the number of Turn’s on the coils? The labellings suggests they output to voltage meaning the resistor is internal to the clamps, but I can’t seem to get a voltage output reading.


  7. gar
    gar July 3, 2012

    120703-2159 EDT


    Each current transformer has a built-in burden resistor, and adjustment resistor. I believe the current transformers are part of TED’s problem. Power measurement has a substantial error on reactive loads. While it seems to be fairly good on resistive loads. I have not yet studied the characteristics of the TED current transformers to prove this is the error source on reactive loads.

    Using a TED 1000 I was not able to prove that a consumer power factor correction capacitor at the main panel did not save energy. TED showed about a 5% improvement, while the power company spinning disk kWh meter proved no saving, and theory says there is no saving.

    This also means that power measurements on things like whole house air-conditioners (no external power factor correction provided) are probably going to read high, assuming calibration was done on a resistive load. However, single phase capacitor run motors can have fairly good full load power factor.

    A Kill-A-Watt EZ has much better characteristics on reactive loads. It uses a resistive shunt for current measurement. However, the Kill-A-Watt can not work with loads that TED can handle.

    I don’t like PLC, but it can be made to work on an isolated circuit with zero errors.


Leave a Reply