Hackaday Prize Entry: A Better Bench Power Supply

Back in February, [The Big One] started building the bench power supply to rule them all. His previous power supply was just an ATX computer power supply. It worked, but that didn’t give him fancy stuff like different channels of individually adjustable voltages. Since then, we’ve spun up the 2015 Hackaday Prize, and [The Big One] has changed his DIY power supply into a Hackaday Prize entry that competes well against $1000 mid-range commercial units.

The single most expensive component in this power supply are a pair of isolated switched power supplies rated for 15V and 7A. This is a change from [The Big One]’s original plan to use a big ‘ol transformer; a switched mode supply is smaller, lighter, costs about the same, and is much better suited to the modular nature of the project.

The final design for this power supply has some interesting features: up to six channels are possible, voltage and current can go all the way down to zero, and everything can be controlled over USB. Those are amazing features that won’t be found in any $100 cheapo bench power supply, and [The Big One]’s amazing documentation for this project makes it a perfect entry for The Hackaday Prize.

The 2015 Hackaday Prize is sponsored by:

20 thoughts on “Hackaday Prize Entry: A Better Bench Power Supply

  1. Not nearly this level of quality or sophistication, but I made one once with two 24V switching laptop power-supplies, and one of the nicest aspects I hadn’t planned originally was that they could have separate ground-outputs (make sure they’re not grounded through your AC’s outlets!) … Connect one’s positive to the other’s negative, and you got up to +48V, or use that connection as “ground” and you’ve got +-24V… (beware, some/most(?) regulators and possibly even switching don’t work with reverse-current; e.g. tying their grounds together, running one at 3V and the other at 5V, then trying to run a 2V load across the two, the current flowing into the 3V regulator will bring its output up to 5V; no current.)
    Very nicely-done TheBigOne. And USB control, to boot.

    1. That’s a good trick to know. I used it once to replace a failed multi-output linear power supply in a piece of scientific equipment. It was a beast, and an OEM replacement would have cost $1,500. A pile of switchers strung together were much smaller and cooler, cost $50 total, and worked perfectly.

    2. Thanks!

      I had considered going for isolated outputs, but that is unfortunately beyond my skills at the moment. Perhaps rev 2.0… :-)

      (My experience with power electronics at the beginning of this project was essentially nil. I have learned a lot over the 8 months I spent on this project, but I know that I have a lot more to learn.)

      Cheers

      1. Mine’s little more than linear regulators controlled by potentiometers and a built-in multimeter… I had to use a wall-wart (in-box) to isolate the meter’s power from the other two systems… I could only imagine how much more complicated it’d be to have controller-circuitry trying to run isolated circuits, trying to measure current, adjust the voltages and more. That stuff’s complicated-enough, as it is. Good-on-yah!

        1. Yeah, it was the controller which was the tricky part.

          If I were to make an isolated supply today, I would start by just using a single isolated switch mode supply per channel, and throw a linear regulator in front. If I put a microcontroller + DAC on each channel, and communicate with the main display over an isolated I2C bus, I could probably do it fairly easily, but that adds a lot of complexity and cost. Oh well, I am very happy with how this one has turned out. If in 5 years I decide that I want to re-make the internals, I can do so then.

  2. The USB connection really should be isolated. This reduce the “ground loops” on the computer side. Not too sure whether one would need that many output channels. I tend to have built-in power supplies for my designs that takes a single input. Opamp these days run single supply and with Point of Load (POL) switchers, I wouldn’t use bench supplies for low voltage logic either.

    I have a pair of 60V/5A bench supplies that serves me well. Most of the time I use one of them. On rare cases, I use the two. A few times, I had them in series for testing AC power supplies – current limits comes in handy for that. (it really should be 170V for rectified DC, but 120V is close enough.)

    1. Although the controller can do up to 6 channels, the one I made has 4 channels – 3 negative and 1 positive. 4 is right on the high edge of actual usability – I have done projects in the past that needed 3 channels (+/- 12v for an op amp with 5v for a microcontroller). It is not too far of a jump to say that you want 5v and 3.3v, and there you are at 4 channels.

      Furthermore, the different channels have some different properties. Channels 1 and 2 can go up to 5A, but they have less accuracy for current limiting at the low end of the scale. Channels 3 and 4 can only go to 1.5A, and have more resolution at the low end.

      Cheers

    2. To be honest, I wouldn’t try to power low voltage circuits (3.3V or lower) with a bench supply that doesn’t have +/- sense inputs. The sense inputs compensate for the I*R drop across the cables/connectors so that the connection at the far side of the cable would get the voltage you set the supply to regardless of the load. I would already have POLs on my designs. 1.8V/2.5V logic isn’t uncommon and it doesn’t take a lot if drops to be -3% or -5% and completely out of specs.

      Bipolar analog supplies are pretty much for stuff from the 1970’s. I don’t recall last time I uses analog parts that requires more than one rail. Mind you I use spice instead of breadboarding and when I do protos, I would have on board supplies for it already.

  3. This is an awesome supply for a lot of tasks but most mid range supplies around $1000, even the simple two hundred dollar ones are linear with the big old transformer. This is to reduce the amount okc noise on your power rails. Some sensors are very sensitive to rail noise and switch now supplies can give you extra unexpected headaches.

    1. I use the switching supplies to convert from AC to 15V DC, and use linear regulators to adjust the 15V to 0 – 12.5V. I.e. the switching supplies replace the rectified and filtered transformer. When pulling multiple amps from a transformer, you will likely see more ripple from the transformer supply than you would from the switch mode supply (you need a huge amount of filtering caps to get rid of ripple when you are pulling, say, 5A current).

      I have done tests on the output, and things look good to me across a wide range of output voltages and currents.

      Cheers

  4. I’d like to point out that 1000 $ commercial units stand out by quality, not features. I have seen no performance tests for this one and comparisons. So just throwin in impressive features gives you nothing is the software glitches or even just a pushbutton is of poor quality (why I trashed a 1 year old 500 Dollar unit last year).

    1. The quality of this supply meets (and exceeds) my needs. It may or may not be suitable for yours.

      I have a video showing it in operation posted on my project page. Feel free to watch if you are interested.

      As far as performance tests, I have verified the accuracy of the voltage + current setpoints vs actual measurements, and current limiting. I have also verified ripple on the output at various loads. The details on these are scattered throughout the documentation on my project page if you are interested. I don’t have the equipment to do much more than that, but again, what I have done is sufficient for my needs.

      Cheers

      1. I had to. It is used at work and we are not allowed to tinker with the devices if they are using line voltage. And the button was so bad, it would – instead of gating the output – sometimes do nothing, sometimes bounce and cause several an/offs in a few hundred ms or sometimes work. Way to dangerous for valuable prototypes. Switched to the new three channel Rhode and Schwartz supply, that’s 1000 bucks and absolutely amazing and a joy to use.

Leave a 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.