• State TI Thinks Resolved
  • Locked Locked
  • Replies 9 replies
  • Answers 1 answer
  • Subscribers 64 subscribers
  • Views 476 views
  • Users 0 members are here
Support feedback
Options
Options

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

power supply selection help for low power application

Robbie Valentine
Expert 2975 points
Other Parts Discussed in Thread: BQ35100, TPS62745, TPS62840, TPS60400

Hello,

I have a low power application as shown in the attached image.  

Block 1: analog logic that gates battery pack off when cells hit low voltage limit

Block 2: fuel gauge that outputs a digital toggle to indicate charge count accumulation

Block 3: charge controller

I'm trying to design in the fuel gauge.  Because this fuel gauge outputs a digital toggle/reset pulse/interface I will be selecting a low power MCU (most likely FRAM part) for this and some other power system diagnostics and interaction.  My question regarding power is what is the best "low power" approach for powering an MCU like (just as an example) an MSP430 of some sort.  Do I just look for a low quiescent current linear power supply?  My battery pack outputs 5.5V-6.5V when operating so I need to drop that down for a logic device.  

So I guess my power questions are as follows:

1. What parameters are most important for a low power solution when selecting a power supply?

2. What power supply technology makes the most sense for a very light load (shooting for average of < 1uA).  I"m assuming linear, but what specifically?

3. With such a low power draw is it more efficient to drop down to say 1.9V for my logic power rather than 3.3V?  The flip side is to only drop down to 3.3V and therefore you have a lower voltage drop across the linear with the same current draw - does this equal lower power solution?

  • 0
    TI__Guru*** 148570 points
    Hi Robbie,

    5.5V - 6.5V is a rather strange voltage range for a battery pack. Are these primary cells? TI made a gauge for primary cells a few years ago with bq35100. Maybe this does what you need to do in your system without all the design work required.

    On the power supply portion, this older app note shows that there can be increased efficiency by running some processors at lower voltages: www.ti.com/.../slyt356.pdf You would need to check if your MSP430 performs the same way.

    In any case, for low currents in the 1uA region, you need an ultra-low power supply. TPS62745 and TPS62840 (soon to be on the web) would be good choices.
  • 0 in reply to Chris Glaser
    Expert 2975 points

    Hi Chris,

    Thank you for the reply!  I'm using 1.2 volt NiMH cells in a series arrangement - so I believe they are secondary cells (not totally sure). So when depleted (in theory from the data sheet) they are around 1.05V and when charging to full charge the voltage on the cathode can reach actually 7+V but falls to around 6.2-6.5V depending on charge cycle and circuitry state.  I just said 6.5V because when looking at the voltage after a charge it sometimes takes a little bit of time to come back down to exactly 6.  

    Anyways, I couldn't find a solution from TI for the charger or fuel gauge.  The closest thing was an IC (fuel gauge) for UPS type stuff (power supply battery backup systems) that was intended to be used a very few number of times not for repeated charging cycles.  I do have a solution for charging that is up and running for charging.  I also have a potential solution for a fuel gauge.  

    The issue I have is that I need a very low power solution to count digital pulses from the fuel gauge.  Because of this I was thinking a very low power supply and MCU.  The FRAM family looks like a good solution, but I was unsure about how to approach the power supply situation and what may be my best bet.  I assumed an LDO with the lowest quiescent current.  

  • 0 in reply to Robbie Valentine
    TI__Guru*** 148570 points
    Thanks for explaining. I let our battery team know, in case they have any devices that can work with NiMH batteries. Yes, these are rechargeable.

    From what I have seen at TI and in the market, the best buck converters have lower Iq than the best LDOs. As well, especially with your higher Vin, there are large efficiency differences between a (switching) buck and a (linear) LDO due to the Vin to Vout difference. TPS62745 is the best buck to use for your Vin and Vout.
  • 0 in reply to Chris Glaser
    Expert 2975 points

    So I had assumed at the very light load a buck would have poor efficiency.   Isnt the case that bucks are generally designed for a target load so the efficiency is peaking/ highest at that load.  With my load of a few 100 uAs wouldn't the efficiency ne very poor?

  • 0 in reply to Robbie Valentine
    TI__Guru*** 148570 points
    Hi Robbie,

    Not many bucks are designed this way. Especially for low power, battery operated systems, the efficiency curve should be basically flat to give a good efficiency across a range of load currents. Just open the data sheet for the TPS62745 I sent earlier.

    For comparison, an LDO running from 6.5Vin to 1.9Vout will have a maximum efficiency of 29%. And this value does not include any estimate of losses from the Iq of the LDO. The TPS62745 has greater than this efficiency at 1uA load.
  • 0 in reply to Chris Glaser
    Expert 2975 points
    Chris,

    That part looks great! I had no idea that was in TI's portfolio. I will most likely be moving forward with that switcher for my MCU fuel gauge monitoring block. This part has brought up another question. I'm considering a negative power rail in the product for the same low power battery circuitry. Would I be able to use this part to generate a negative voltage rail as well by connecting the output to ground and using the ground as my negative rail?

    This stems from a need for me to amplify the voltage drop across the two current sensing resistors shown in my diagram. One is for sensing charge current and one is for fuel gauge sensing. Because my current is low the impedances are fairly high ~0.5 Ohms. I was thinking about trying to buffer the voltage drops with amps so the resistance can be lower, but the charge and fuel gauge IC still get their needed voltage waveform for sensing. With the power supply you linked above, I'm wondering if I could create a negative power rail for amps to use in this way.

    Does that sound like a reasonable configuration (negative supply concept) for this power supply?
  • 0 in reply to Robbie Valentine
    TI__Guru*** 148570 points
    I'm glad you like the TPS62745.

    In theory, yes it can be used to make a negative voltage. A charge pump, like TPS60400, would be much simpler but draw more current.
  • 0 in reply to Chris Glaser
    Expert 2975 points

    Chris,

    If I use this part as a negative PS I'm assuming I connect the output to ground and use the ground pins (and EPAD) as my negative output.  This brings up an interesting questions though.

    For the 1.8V TPS62745 in 1.8V operation (section 9.2 of datasheet) where should the VSEL pins be connected.  In normal operation they are connected to ground.  In my negative operation should they be connected to the "negative" ground nets?  I hope that makes sense.

    Example schematic:

  • 0 in reply to Robbie Valentine
    TI__Guru*** 148570 points

    Hi Robbie,

    Yes, you're correct.  All pins need to be level shifted to connect to the correct logic levels, if driven by an MCU.  If tying them low, yes connect them to the IC's GND (which is -Vout).  This app note talks more about this: http://www.ti.com/lit/an/slva514b/slva514b.pdf 

    As well, leave the pads for C39 but don't put anything there.  The input cap needs to be added and goes from Vin to system GND (which is IC Vout).