How Much Current Does That Thing Draw?

If you ask us how to measure the current draw from something, we’ll break a power lead and put a multimeter in series with the power supply. If that’s not handy, we’ve been known to take the fuse out of the power supply and replace it with the meter. Crude, but effective. But if you have about $8,000 sitting around, you could go grab a Keithley 2460 SourceMeter.

What’s a SourceMeter? Well, as far as we can tell it is a power supply with very accurate built-in current monitoring and a microprocessor that can display lots of interesting statistics and graphs. In all fairness, this looks like a souped up model, but they start at about half the price which is still a lot more than most hacker budgets.

You might wonder why we’re talking about an $8,000 instrument. Well, it is nice to see what will be popping up on the surplus market at some point. But more than that, this seems like this would be a highly doable hacker project. There are lots of ways to measure current from a shunt resistor to a hall effect device. We certainly know how to make cool embedded systems with nice touch screens. Or you could just pipe the data back to a PC and crunch it all there using your favorite language.

We couldn’t think of any “smart power supply” projects that would compete with the SourceMeter. We’ve seen monitoring on loads and converters. We also see plenty of projects for monitoring AC. After browsing the videos for these SourceMeters, though, we are wondering why we don’t always put a digital monitoring interface on any bench supply we are building.

28 thoughts on “How Much Current Does That Thing Draw?

  1. “But more than that, this seems like this would be a highly doable hacker project. There are lots of ways to measure current from a shunt resistor to a hall effect device.”

    Sure, but doing it in a repeatable way that is linear across range and fully calibrated, something where you’d stake your career on the measurements being correct…that’s when you pay $8,000.

    1. Couldn’t agree more. When I was working as a HW developer for a multinational company I had to write the type of the power supply down into the test report. And it would go to the customer which was paying $3M. It would have been shameful to write down the name of a cheapish brand there. That does not mean that I cannot build myself a reliable ammeter power supply which can be calibrated to show precise values. I could even connect a cheap DSO203 with a shunt on the output and have that stored current oscillogram and that would still be under 300$. I think it also depends on how reliable and trusty do you want it to be when we’re talking about industrial design. You cannot afford loosing time and remake all the tests just because you find out that the multimeter or the power supply was giving some erroneous measurement.

      https://hackaday.io/project/27794-build-your-own-panel-meter
      https://www.youtube.com/watch?v=Zpa7TBHG6Ks

  2. It’s actually much more than that. I have a 2450, lower powered version of the 2460. What it actually does is operates in all four quadrants, so can source voltage of either polarity, and also source or sink current at the same time. 0 – 200 volts, 0 – 1A, but only 20watts. Want to monitor a 1.8 Volt CPU drawing 600nA, no problem. Measurement limits are 10nV and 10fA, but can only be done with triaxial leads plugged into the back panel. Pico amps are easy with normal, albeit short leads. It has made very low power design a lot easier.

      1. Trouble with a uCurrent, and I do have a few of them, is that one side of the current shunt is connected to one side of the output. Trying to measure 50nA increasing to 4mA for a few microseconds every six seconds really wasn’t fun.

  3. “In all fairness, this looks like a souped up model, but they start at about half the price which is still a lot more than most hacker budgets.”

    Some hobbies are like that. Cheap only goes so far.

    1. Thanks for the tip with Hantek, never heard of them before. our local reseller seems to be a bit greedy, asking 120$ for the 20kHz/65A clamp, but still beats every Keysight, Rhode&Schwarz or Tektronix offer i’ve seen, they only just start at around 900$ for a 20A 100kHz current probe

  4. Analog Devices have a low powered version. http://www.analog.com/en/analog-dialogue/articles/studentzone-december-2017.html Less than $40 at DK, who’s out of stock right now, plenty at Mouser. If you only need to measure the current then the EEMBC new hardware works well. https://www.st.com/content/st_com/en/products/evaluation-tools/product-evaluation-tools/stm32-nucleo-expansion-boards/x-nucleo-lpm01a.html?icmp=tt5907_gl_pron_oct2017

  5. These are used to replace things like curve tracers and can measure currents down to fA. This is a “SMU” a source measure unit. Sets of SMAs together make a “parameter analyzer” which is used to do quasi-static semiconductor/device characterization. Its not really just a power supply. You can do pulsed measurements and things with them also. Yes people do use 24xx Keithleys as fantasy power supplies with tight complacence limits but that isn’t really their design function.

  6. If you’re on a budget, consider getting a keithley 236/237/238. Those things are beasts even by today’s standards if you hook up a gpib adapter. Watch out for the triaxial connectors though…you need 3 and just one will run you about 50$.

  7. You could DIY the functions in this, but 6 digits of both current and voltage measurement precision is pretty spectacular, especially four-quadrant. That is not a trivial design.

    1. also far from a trivial build even if you have a free design. We’re talking 120dB of dynamic range here, which means construction quality (precise ground layout, shielding/cross-talk, etc) are absolutely critical.

    2. Don’t disagree. But I have a lab-quality bench meter (a Fluke 8800 series) and you know what? For 99% of what I do my cheap no name DVM is just fine. I bet I fire up the Fluke 4 times a year and 2 of those are just because I’m too lazy to fetch the cheap meter from where I left it.

      I wasn’t trying to imply that the instrument wasn’t worth it for what it does. But I’m surprised we don’t see more hobby-level gear like this that maybe doesn’t have the 6 digits and the dynamic range, etc. It seems pretty useful and — in the simple case — pretty easy to do.

  8. A pretty good option that covers most cases for a reasonable price is the Otii from Qoitech. I upgraded from a combination of a modified uCurrent (for greater range) and Salae Logic with analyzer scripts a couple of months ago. It has a nice user interface and can log a couple of more signals including UART.

  9. I used to be annoyed when I saw such expensive kit being showcased in amateur publications.

    Now I just look at it as a shopping list for a few years from now when the same functionality has been duplicated with perhaps an acceptable loss in precision for about 1% of the price by some hacker in China that sells on FleaBay.

  10. For those on a budget that want to use a multimeter to measure current or energy consumption of microcontroller-based products, check out this article: https://joulescope.com/learn/current-measurement-challenges/.

    Although you can make due with a multimeter, having better tools helps significantly, but the price has been too high for most of us. I have been working for over a year on a low-cost product named Joulescope that measures current, voltage, power and energy with sufficient bandwidth and resolution. Hope to ramp up manufacturing by end of year!

          1. That makes me sad too, especially since it uses AWS Lambda@Edge. It should never be unavailable! You can just send an email to matt [at] jetperch [dot] com with your email & name, and I will make sure you get added. Thanks for letting me know, and sorry for the inconvenience!

Leave a Reply to BrianCancel reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.