EE nerds: Use your brain to help Icrontic

ThraxThrax Professional Shill, Watch Slut, Mumble Hivemind DroneAustin, TX
edited February 2010 in The Pub
One of the most remarkable things about Icrontic is that it has an outstanding community filled with subject matter experts: Linux, Windows, Mac OS, smartphones, graphic design, programming and--most relevant to today's post--electrical engineering.

You see, Icrontic is looking to get back into the game of power supply reviews. But in this day and age, it is not enough to throw the unit in a system, hook up a multimeter, and elaborately declare "YEP, IT WORKS."

Today's readers demand a whole lot more. They want to know if the power lives up to its label at full load and in high-temp conditions. They want to know about ripple, jitter, power efficiency, transients, etc. etc. Anything less makes for an embarrassing review that our peers (not to mention knowledgeable readers) will look down on, and that's not something we're interested in.

In order to obtain this data, many sites larger than us have opted for a SunMoon SM-268 unit, which connects directly to a power supply and can increase the voltage on the rails in 0.5v increments up to the PSU's maximum. This unit also costs $3000-5000. We cannot afford that.

It is my understanding, EE nerds of Icrontic, that it is possible to build a similar device using modular banks of wire-round resistors that can be added or removed to simulate a load. It is also my understanding that such a device can be built for a fraction of the price of a SunMoon unit.

So, my friends, I am calling on you to help Icrontic: Can you design, build, or simplify what is required to make a homebrew unit that can simulate loads? It needs to be capable of pushing power supplies like this to the limit.

I also need a crash course in using the damn thing, because I do not have much experience with electronics.

I know this is a tall order, but we've avoided power supplies for a long time because we do not have the gear to give you the reviews that you deserve. Help us get back in that game, and give you good content.

Thank you. :)
«1

Comments

  • jaredjared College Station, TX
    edited February 2010
    Well since Dras created his own GPS and god-knows-what-else from scratch, I expect him to chime in :D
  • _k_k P-Town, Texas
    edited February 2010
    Just hook a bunch of these together and buy a multimeter.

    The only reference to a home built SM is a toaster loader. Gonna need a few adapters so that you can break out all the 12v, 5v, and 3.3v lines into a single cable that runs back to the set-up and have a big solder point where you can use a multimeter easily on it. I assume you want some decent accuracy and logging functions, some thing to start with.
  • PetraPetra Palmdale, CA USA
    edited February 2010
    My wife and I had talked about building exactly what you're talking about a few years back...before my original vision for Petra'sTech was shot to pieces (even ran through the simple math to determine amount of wire required for each wound resistor, solid and stranded, copper and NiCr, worked out a cooling stragety, etc.). Needless to say, the project was scrapped.

    Anyway, putting together a crude load tester like that is pretty simple... you just won't have the fine adjustability of an electronically controlled load tester (which probably also wouldn't be that hard to design). With either loading unit, you'll still need a few things to pull off a PSU review: good multimeters, an oscilloscope, and a basic understanding of circuit theory so that you can comment on the design and component choice of whatever PSU you're reviewing.
  • ThraxThrax Professional Shill, Watch Slut, Mumble Hivemind Drone Austin, TX
    edited February 2010
    Thankfully, the one thing I do understand about electronics is PSU design. If we can get a load tester in line, I know how to work a digital+USB oscope and I have a good DMM already.
  • edited February 2010
    I can't build the electronics portion but I can make a cool enclosure for you to house it in on my cnc machine. Complete with the IC logo, fan mountings ( if needed), and whatever else you will need. I probably have enough scrap aluminum and plexi lying around where I wouldn't even have to buy any materials. Let me know if I can be of help.
  • PetraPetra Palmdale, CA USA
    edited February 2010
    I'll see if I can find any of our notes from the old project...
  • shwaipshwaip bluffin' with my muffin
    edited February 2010
    I think it might be beneficial to get a group together to brainstorm what we want in this, and then movie from there. I'm glad to volunteer, but I'd have trouble specifying all parts of the system - there are many intricacies I wouldn't know how to deal with, for example, I have no idea how we would want to load the -3.3V rail...or how fine of steps we would like to be able to test a rail at (do we want to be able to load the 12V rail in steps of 1A, .5A, .25 A, etc). Many new PSUs have multiple 12V rails - do we want to be able to load them all at the same time with different load values, or always the same.

    I have the EE design and soldering skills to build whatever we want, but probably not the design skills to know exactly what it is we want.
  • edited February 2010
    I could also mill out the traces on circuit boards if someone else could draw up a clean schematic in a dxf format or PDF.
  • primesuspectprimesuspect Beepin n' Boopin Detroit, MI
    edited February 2010
    Waiting on input from Drasnor and J.
  • QCHQCH Ancient Guru Chicago Area - USA
    edited February 2010
    Other option... I could try to barter with one of the EE here at Fermilab to have one of them do it... many of them LOVE these kinds of projects.
  • PetraPetra Palmdale, CA USA
    edited February 2010
    Okay, I can't find our notes... but I can still give you the basic run-down on how to get everything together (as can many others). However, as shwaip pointed out, the details of how the PSUs are to be tested needs to be determined first (e.g. load increments desired, form factor of the test system, how many rails does the tester need to be set up for (one load for all 12V PSU rails vs. independant loads for 12V rails and how many), etc.) because that'll determine how the unit is designed and built.
  • drasnordrasnor Starship Operator Hawthorne, CA
    edited February 2010
    I built one. Do you want me to mail it to you?

    EDIT: it would probably be cheaper for you to roll your own. I made mine using a 2x4 and some nichrome wire. Clamp the wire at two ends and put it under a lot of tension, then clip high-current alligator clips to the wire. Put the system ground in the middle of the wire (or wherever) and slide the other clip along the wire until you get the desired resistance for the load you're planning to test. Use multiple wires for multiple simultaneous loads. A guitar pegbox mechanism would make a good tensioner.

    P = V^2 / R

    Measure voltage with a good multimeter, transient and ripple with an oscilloscope set for AC mode. Objects of interest are going to be in the Hz - kHz range so you don't need to shell out for a really expensive oscilloscope. One of the USB dongles should be fine. Resistance of the wire increases with load, so you may want to get an inductive ammeter to measure current and then slide the clip until you get the desired power if you need more precision:

    P = V * I

    -drasnor :fold:
  • ThraxThrax Professional Shill, Watch Slut, Mumble Hivemind Drone Austin, TX
    edited February 2010
    Can you post a picture of this device, Dras? I remember it for our Tuniq and Corsair reviews.
  • AnnesAnnes Tripped Up by Libidos and Hubris Alexandria, VA
    edited February 2010
    drasnor wrote:
    I built one. Do you want me to mail it to you?

    You are SUCH a badass, sir.
  • drasnordrasnor Starship Operator Hawthorne, CA
    edited February 2010
    Thrax wrote:
    Can you post a picture of this device, Dras? I remember it for our Tuniq and Corsair reviews.
    I'll post a picture tonight. It wasn't as refined as the one I just described to you (no variable load) but I got to see some interesting results at like 250W on those PSU's.
    Annes wrote:
    You are SUCH a badass, sir.
    :rarr:

    -drasnor :fold:
  • jaredjared College Station, TX
    edited February 2010
    As, I predicted earlier :D

    NASA better get in line ;)
  • shwaipshwaip bluffin' with my muffin
    edited February 2010
    Won't that get pretty hot loading it to 25A? Not that resistors wouldn't get hot, but http://www.wiretron.com/nicrdat.html tell me that loading even 8 AWG wire to 25A raises the temperature to > 400F
  • drasnordrasnor Starship Operator Hawthorne, CA
    edited February 2010
    shwaip wrote:
    Won't that get pretty hot loading it to 25A? Not that resistors wouldn't get hot, but http://www.wiretron.com/nicrdat.html tell me that loading even 8 AWG wire to 25A raises the temperature to > 400F
    Oh yeah, it glows red hot during a load test so it's important to have the wire tensioned (it gets longer when it's hot) and away from anything combustible. What did you expect? It's radiating a tremendous amount of energy; it has to go somewhere.

    -drasnor :fold:
  • MAGICMAGIC Doot Doot Furniture City, Michigan
    edited February 2010
    Submerge it in cooking oil and make french fries at the same time.
  • shwaipshwaip bluffin' with my muffin
    edited February 2010
    drasnor wrote:
    Oh yeah, it glows red hot during a load test so it's important to have the wire tensioned (it gets longer when it's hot) and away from anything combustible. What did you expect? It's radiating a tremendous amount of energy; it has to go somewhere.

    -drasnor :fold:

    Like wood? ;)
  • drasnordrasnor Starship Operator Hawthorne, CA
    edited February 2010
    shwaip wrote:
    Like wood? ;)
    ^Hot.

    -drasnor :fold:
  • shwaipshwaip bluffin' with my muffin
    edited February 2010
    Well played, sir.
  • drasnordrasnor Starship Operator Hawthorne, CA
    edited February 2010
    Thrax wrote:
    Can you post a picture of this device, Dras? I remember it for our Tuniq and Corsair reviews.
    See attached. The notches on the side are inch marks and the pegs are spaced to load an attached power supply at 250W on each of 5V and 12V iirc, but I'd check again before using it. Wood makes a good backing material since it's an insulator and won't conduct electricity between multiple loads on the same board. I used a standard ATX PSU "tester" to turn on the attached power supply.

    This one was dirt cheap to make since I didn't go any further than my parts box, but if I were going to do it again here is what I'd do differently:

    * Use a bigger board so I can have more loads for more rails.
    * Get a real tensioning mechanism so I don't need three hands, a wrench, and a ratchet to tension the wire.
    * Use longer pegs to get some more clearance between the wire and backing.
    * Use battery clamps to attach to the wire so I can vary the load for a given rail.
    * Use thinner wire so the load can be a bit shorter. This was built with 16 ga nichrome, use 18 ga.
    * Use a different PSU connector.
    * Use a bigger, higher current switch or relay.

    Bear in mind this is just a load; you still need actual instrumentation to measure things.

    -drasnor :fold:
  • ThraxThrax Professional Shill, Watch Slut, Mumble Hivemind Drone Austin, TX
    edited February 2010
    Here's another example of the type of equipment I am talking about:

    http://www.hardwareheaven.com/reviews.php?reviewid=907&pageid=4
  • edited February 2010
    Is it me, or is the aspect of the potential home brew electronics project actually as interesting than the potential power supply reviews that could come from it?

    I'm as interested in seeing how to build this thing than seeing any PSU results.
  • PetraPetra Palmdale, CA USA
    edited February 2010
    Yeah, that's pretty much what we were looking at building... with hand-wound resistors, though (granted, the time savings from buying commercial resistors like those would probably be worth it).

    To build it, you really don't need much more than Ohm's Law and an understanding of series/parallel resistance... this site could probably be of some help: http://www.ifigure.com/engineer/electric/electric.htm
  • _k_k P-Town, Texas
    edited February 2010
    LIGHT BULBS a lot of light bulbs
  • PetraPetra Palmdale, CA USA
    edited February 2010
    _k_ wrote:
    LIGHT BULBS a lot of light bulbs

    You ever tried to find a burned out bulb on an old string of christmas lights? ;D

    It would work, but probably be a bit of a PITA to deal with (not to mention that you're stuck with whatever resistance value the bulbs happen to have, rather than choosing from a wide variety of commercial resistors or winding your own).
  • shwaipshwaip bluffin' with my muffin
    edited February 2010
    Here's a simple "schematic" of my idea. You'd need one of these for each 12V rail, and a similar one for each 5V rail

    attachment.php?attachmentid=28217&stc=1&d=1266872111

    The resistor values are set up so that the number of amps that flow through each parallel branch increase by powers of 2. This allows you to use binary to count up the total number of amps in increments of 1A. Want 6A? Use the 4+2 branches. Want 15A? Use the 8+4+2+1 branches. The downside is this can get kind of annoying to do manually, so I designed the resistor branches not to exceed 4A, so 5A relays could be switched with a microcontroller.

    Each individual branch will be made up of multiple 12 Ohm resistors, allowing lower power resistors to be used (cheaper, no need for a heatsink...a fan blowing over them would be a good idea, though. To make the 2A branch, 2 12 Ohm resistors will be in parallel. To make the 3A branch, 3 12 Ohm resistors will be in parallel. If someone is willing to make or point me to heatsinks that would work with the higher power/single resistors that would work as well.

    The downside is that this may take up a lot of space once you need to test 6 12V rails and several 5V rails.
  • drasnordrasnor Starship Operator Hawthorne, CA
    edited February 2010
    Attached is my first pass at an analog high-current variable resistor stack. The idea is that each wire can test a rail by varying the spacing between clips to achieve the desired resistance. I ran a couple of test cases, but if you were actually going to spec this thing out you'd need to decide what power ranges you want to test over and how many simultaneous rails you want to test. There's a lot of good information on sizing NiCr heater systems here. The advantage of a system like this over a resistor ladder like shwaip described and HH implemented is that the parts count is low and the parts you need are really cheap. The disadvantage is that it needs you to do some prep work to set it up. Oh, and don't touch it while it's running because it will be glowing.

    shwaip: your higher-order resistors are going to be rated for some obscene power. Heat sinks and active cooling are required because even though you may only be radiating 12-15W per resistor, your array is still radiating the full rating as heat and that heat needs to go somewhere.

    -drasnor :fold:
Sign In or Register to comment.