• EEVblog #259 – PSU Rev C Schematic – Part 12

    Previous Video HERE: http://www.youtube.com/watch?v=xa9Lyb45oJM
    Rev C Schematic HERE: http://www.eevblog.com/projects/usupply/

    Part 12 in the PSU design series.
    A thorough look at the Rev C schematic and all the changes.
    18650 Lithium Ion battery charger, software controlled switching pre-regulator, replacement of the uCurrent capability, return of the PWM DAC, accommodating the Wiznet Ethernet module, and a whole lot more.

    Be Sociable, Share!

      About EEVblog

      Check Also

      EEVBlog #819 – Kirchhoff’s Laws Tutorial

      EEVBlog #819 – Kirchhoff’s Laws Tutorial

      Dave explains the fundamentals of Kirchhoff’s Current Law (KCL) and Kirchhoffs Voltage Law (KVL) leading ...

      • Raphael (germany)

        Hi Dave!
        I’m not sure that the new differential-amp with the LM358 is gonna work.
        First, this opamp is not a Rail-to-Rail type on its outputs and you want to buffer a voltage equivalent of V+ (when there is no current flowing through shunt) with U12A. The opamp is supplied with V+ ! According to datasheet, it’s also not possible to sense an input voltage greater than (V+)-1.5V . And finally: The output of U12B will never go near 0V when the opamp is powered uni-directional with 0V and V+, especially when there is current flowing IN the output pin(through R17, R18).
        I had some bad experience once with these self-made-diff.amps, when you expect them to have a decent accuracy….
        But apart from that: Great new design!

        • http://www.eevblog.com EEVblog

          Yep, I thought I was still using a rail-rail device and forgot about that. It’s not going to work, time for a change there.

      • Worf

        Just FYI – the battery connector was redone in the PDF from the one in the video, in case you were wondering how the circuit in the video could possibly work.

      • Manu


        Great design! I love the idea of a battery powered power supply. I don’t have a dedicated bench for my electronic projects at home and I set my electronic boards where I can. So, this is a very good idea to have a portable power supply that I can put in my cardbox before stocking all that in my cupboard.

        If I could make a wish, I would replace the wiznet module by something more configurable as a wifly module like this one:
        I like this xbee form factor that allows you to use different wireless protocols:
        – 802.15.4 (wireless UART): http://www.sparkfun.com/products/8664
        – Zigbee version: http://www.sparkfun.com/products/10415
        – bluetooth: http://www.seeedstudio.com/depot/bluetooth-bee-p-598.html

        That could make it totally wireless.

      • Old Dude

        Hi, Dave
        Looking Back At Part 1(#221 i Believe) It
        Seemed To Be A DIY Thru-Hole Design.
        It Appears to Have Morphed Into A Buy Only
        Design I’m Sad To Say.
        I Enjoy The EE Aspect Of The Design Though.
        How About A Software Session?
        Best Reguards

      • http://joost.damad.be/ Joost Yervante Damad

        You might want to change the FTDI connection from a 5 pin to a 6 pin and make it compatible with the FTDI Basic from Sparkfun.


      • Galileo

        3x3mm MLF is a pain to solder by hand and you are going to have these boards machine made, so what’s the point with the thru-hole design?

        Switching to smt design would also cut down the cost of the whole board.

      • Mav

        Dave when you say you lost serial (which you wanted), why can you not use the FTDI USB>UART bridge as serial coms when the port isn’t being used for Aurino’s bootloader ??

      • http://www.mi-engr.com Robert

        I was wondering if you were going to package the various revisions into a zip package? (schmatics, boards, software)

        Then people can select and build the various rev’s depending on whether they are through-hole or surface mount types.

        • http://www.mi-engr.com Robert

          Ignore this, I found the download area.

      • John

        This is probably a stupid question (Sorry) but with all of this I2C stuff going on, and the updating of the display in the front and the inspection and setting of various e-pots here and there, um, is the processor going to be fast enough to catch a serious current spike at the output and trip fast enough to save the (external load) circuitry? I seem to recall that I2C has never been legendary for speed with multiple devices on a bus.

        Apologies if I’m completely wrong and this is really an utterly silly question.

        • http://www.eevblog.com EEVblog

          The current limit is not handled by the software, it’s direct analog, so responds instantly. The I2C pre-regulator is just that, a pre-regulator, and nothing to do with current surges.

          • dougg

            In my experience, processors (e.g. ATmega168) are a lot more reliable than I2C devices. So seen from the firmware, you can detect the I2C bus is locked up (e.g. an I2C device holding down SDA or SCL) but often you can do nothing about it. Two of the I2C devices in this design have a reset pin; can their chains be pulled by the MCU? The only way to reset the other three I2C devices would be by controlling their 3V3 supply (e.g. via a small power mosfet controlling their supply gated by reset).

      • http://www.bastli.ethz.ch Mario

        Hi Dave,

        Isn’t it possible to use the AVR to directly control your boost DC/DC converter? A simple PI-controller is relatively easy to implement and may not use too much ressources. That would save you the Epot and the designated boost-controller-IC at the cost of a timer interrupt and an ADC channel.


      • Jan

        Hi Dave,

        sorry for saying this, but the µSupply is suffering from featuritis :), i.e. you’re cramming every possible feature into the power supply design.

        IMHO, it’s rather stupid to have a battery powered PSU with a wired network connection.

        A basic tool like a small lab power supply should be buildable with average skill and below-average tooling. Furthermore it’s an OSHW design. But relying on machine-populated boards with DFN packages will make it hard to build the PSU from scratch.


        • http://www.eevblog.com EEVblog

          No need to use the ethernet thing with the batteries, it’s optional. It can be mains powered too. Most people won’t even bother buying the Ethernet module I’m sure.
          The Ethernet option is FREE, it costs nothing to add the capability.
          As for the building part. Ok, show me how it’s possible to get this thing at a reasonable cost with ALL through hole components. OSHW isn’t just about being to build it as-is, it’s also about the circuit techniques that can be modified and re-used in other designs. Nobody build the Arduino from scratch as as kit for example, but it gets modified and put into other designs in many different was, that’s how OSHW works.

      • MikrySoft

        Isn’t using linear 3V3 regulator with possible ~150mA current draw (with ethernet) a waste of power? You’re dissipating about 750mW, that can’t be good with battery power.

        150mA from 2500mAh batteries is 16h using linear. With buck regulator you would double that time (assuming 80% eff.).
        Adding battery gauge would be nice too.

      • Zizzle

        Wired ethernet on a battery powered device seems a bit strange.

        Did you consider one of those $8 Bluetooth-to-serial modules that you see on ebay (or similar)?

        Serach ebay for “arduino serial bluetooth”.

      • Rasto

        Do You aim on certain sell price for this kit?
        If so, how much?

        • http://www.eevblog.com EEVblog

          The aim is just to keep it as low as I can whilst keeping the features I want. No real target price, it will be what it will be. Don’t know the final value yet. I would it to be under $100, but I don’t think that’s realistic.

      • Ice Skate

        Hi Dave, very instructing video done by a great electronic engineer! Thank you very much to share your ideas. 😉 While I love your last improvement, especially battery powered uSupply, all the idea you put in your usefull open project provide me even more refinement. And the shunt part seems waste too much power and precious space if you switch for CMS components, this attract me to try using something like a precision resistor, for example OARS Series Open Air Surface Mount Resistors from irctt.com (http://irctt.com/pdf/OARS.pdf) coupled with a change in U12A/U12B AOP gain to multiply the voltage drop. Yes, it is more expensive, and the upper value is only 50mOhm. I keep in mind the potential instabilities using such low shunt value, but It is only 50mW dissapted at 1A (against 1W with the actuel 10 // 10Ohm), very attractive to save even more juice and to increase the output voltage range!!!

      • rob

        TPS61170 is a neat device with a programmable bandgap reference so you wouldn’t need the epot.

        what do you think of building a PSU around this device rather than the lt3080?

      • http://www.torbalscales.com/home.php?cat=254 Piotr

        You might want to change the FTDI connection from a 5 pin to a 6 pin and make it compatible with the FTDI Basic from Sparkfun.

      • BBQdChips

        You may have mentioned it and I didn’t see, but, may I ask what you pay for the boards you are using on this project? I’m in the states, and for me to get a proto board, usually min qty of 3, it’s expensive. If I want those with a solder mask and silkscreen, it’s out of the question.

        As an example, a board about half the size of yours with no solder mask or silkscreen, two layer, I’d be looking at $60 US incl shipping. 3 day turnaround. For those with soldermask and silkscreen, the board cost isn’t bad, but there’s a $250 base fee before boards. So, say $265 minimum. Well, I’m not gonna get to Rev C at that price 😀 If I can stay within a standardized size of 3.8 x 2.5 inches, I can get 3 boards with solder mask and silkscreen for $75. But only in that size, and no routing.

      • MSP

        Hi Dave
        Excellent series of blogs. I’m willing to build the PSU, ie my version of it… (0..30V/1A, 0..15V/2A).
        I like LT3080/83 a lot for their ability to provide output voltages down to 0.0V, which no other LDO can do as far as I’m aware. However the fact that the output has to be loaded with ~1mA to bias the part is ugly. Your solution with the LM334 current source on the LDO output unfortunately can only work down to VOUT ~ 1V, which is the min operating voltage for the LM334. So below that nothing goes. Do you have a solution for this?
        I could imagine placing the LM334 on the LDO input side and to mirror the current with a NPN current mirror to the output side.

      • Enzo

        Has enough testing been done on the current limiting circuit piece to see what really happens when the output is accidently shorted?

      • http://www.torpeymalpensado.com.ar/ Guille

        Mmmm… the MCP73213T is very nice, with all this intelligent charging logic and all, but, wouldn’t a charge (the rest of the board) in parallel with the battery throw all that logic into the drain? Don’t these devices sense current flow on the battery to regulate the charging? Just wondering, because I need to make a similar project but with lead batteries and I fear I might face that problem too.

      • Hyuckjin

        Dave. Psu with the rotary encoder to what used to let me know please.

        Thank you.

      • Lost

        Hi Dave,

        I jus watched the whole PSU series and i love it.
        But i can not find anything newer than part 12.

        Is there any more?,
        would be a real shame to let this concept die.

        Kind regards,

      • Thierry


        I’ve been watching the series on the PSU and already learned ALOT off it. I am now even starting to collect components for my own PSU. But is there any chanse you could upload the arduino code and library’s you wrote for part 8? I haven’t done alot of arduino programming at all (just some serial communicating and generating pwm signals). So it would be very handy for me to have some kind off example code to write a program for the REV C.

      • Voler


        I think you did not consider the req’d power dissipation if the output is nearly zero volts with almost full current (1 amp) (e.g. the output shorted). You cannot have the input of the LDO driven to +2 volts because the V+ terminal is at least the battery voltage (which can be max. 8.4 volts).
        This would result in a max. power dissipation of 1 watt for the shunt resistor and 7.4 watts (!) for the LDO.
        Or did I miss something?


      • Diego

        Are the altium files for the project or the software available for us to download?

        I would like to make some mods of my own and it would be great if I dont have to start from scratch.

        PS: Great videos!

        Best Regards,

      • Pingback: ccn2785xdnwdc5bwedsj4wsndb()