• AVR Freaks

Low Power Voltage Divider

Author
invertedflight
Senior Member
  • Total Posts : 139
  • Reward points : 0
  • Joined: 2004/11/14 19:01:31
  • Status: offline
2006/04/02 21:30:10 (permalink)
0

Low Power Voltage Divider

I'm working on a power-sensitive project that does some very simple battery voltage monitoring.  It's a PIC18F6527 and it'll probably sample the battery voltage every few seconds.  Right now I just have a 30K + 10K&.1uF voltage divider on my 12V battery.  This eats up 3.6mW of power all the time and already has some error due to the high resistor values. 
 
I saw a previous post mentioning use of MCP6041 to buffer the signal.  I've never used op-amps before so I want to make sure I have this clear.  The op-amp has an input bias of 1pA so I can easily get away with 100K+300K resistors, right?  And then I just build a simple voltage follower?  How high can I make my resistances, practically?  1M and 3M? The circuit will sit outside in a waterproof case, but could still get slight moisture exposure.
 
Thanks,
Chris
 
#1

18 Replies Related Threads

    chrislev
    Super Member
    • Total Posts : 318
    • Reward points : 0
    • Joined: 2005/02/18 11:04:37
    • Location: Kaysville, Utah, USA
    • Status: offline
    RE: Low Power Voltage Divider 2006/04/03 09:28:42 (permalink)
    0
    Hi,

    Why not consider using an N-Channel MOSFET to switch your divider on and off at the ground side? Then you can use lower valued resistors to improve noise rejection. You can also make a divider who's output matches the pics A/D input impedance (further improving accuracy).

    When you are not measuring battery voltage, switch off  the divider and your current consumption goes to zero.

    You can calculate out the voltage drop across the MOSFET OR I *think* you could simply connect Vref- to the high side of the MOSFET (assuming you are switching the ground on/off) and ignore the voltage drop. I have not thought the VRef- idea out completely so I'm not sure if there would be any issue doing this. It seems right at first thought though.

    Hope this helps.

    Chris

    #2
    invertedflight
    Senior Member
    • Total Posts : 139
    • Reward points : 0
    • Joined: 2004/11/14 19:01:31
    • Status: offline
    RE: Low Power Voltage Divider 2006/04/05 00:48:31 (permalink)
    0
    That's an interesting idea - has anyone done this?
     
    I'm still trying to get an idea for what resistor values would be reasonable if I want to go high impedance.
     
    Is there a general purpose sot-23 fet that you can reccomend?  A 2222A of the FET world?
     
    Thanks,
    Chris
     
    #3
    DSchabel
    Super Member
    • Total Posts : 1714
    • Reward points : 0
    • Joined: 2005/05/24 14:00:34
    • Location: Western NY State
    • Status: offline
    RE: Low Power Voltage Divider 2006/04/05 09:42:07 (permalink)
    0
    ORIGINAL: chrislev

    Hi,

    Why not consider using an N-Channel MOSFET to switch your divider on and off at the ground side? Then you can use lower valued resistors to improve noise rejection. You can also make a divider who's output matches the pics A/D input impedance (further improving accuracy).

     
    That's a great idea, but there are some practical problems. 
    If you switch the GND off, then you have a resistor from +12 to your PIC input pin.  You can make the resistance high enough so that the protection diodes in the PIC can shunt the current away, or, (preferably) add a protection diode externally.  But that means that you're STILL drawing current from the +12V supply into the PIC. 
     
    The op-amp idea will work fine, but the maximum resistances depend on a couple of things, mostly the op-amp and the amount of accuracy you require.  Also, the possibe susceptibility to noise, but that can be mitigated by adding a good bypass capacitor between the junction of your voltage divider and GND.  I'd say it's likely that you can "get away with" 1M pull-downs and 3.3M to +12, with a 0.01uF cap. 
     
    Now, if you wanted to switch the high-side, you could do that without the above problem.  This might be the best way to go. 
     
     

     
    Note that R2 & R3 in parallel should be < 10k to comply with the source resistance requirements on the PIC's A to D.  The capacitor will shunt noise, but remember, it will also increase your settling time slightly.  A 0.001uF cap in this circuit will take about 30usec to settle (4 time constants). 

    When you are not measuring battery voltage, switch off  the divider and your current consumption goes to zero.

    Yep, that's a great feature.  Depends on how often you need to measure, and how long you need to keep the circuit on each time (the "duty cycle"). 
     

    Attached Image(s)

    #4
    Guest
    Super Member
    • Total Posts : 80500
    • Reward points : 0
    • Joined: 2003/01/01 00:00:00
    • Location: 0
    • Status: online
    RE: Low Power Voltage Divider 2006/04/05 09:56:24 (permalink)
    0
    DSchabel circuit is exactly what I would advise. I have done this a few times.
     
    My choice of transistors are the IRLML2402 for the n-channel, and the IRLML6402 for the p-channel FET. The 6402 has a RDS(ON) of 40mohms, being completely irrelevant to the divider. The MOSFETs leakage current is a little high, though, typically 1µA at ambient temp and rising to over 10µA for 50°C. Normally this is not a concern, especially for a 12V battery.
     
     
    #5
    DSchabel
    Super Member
    • Total Posts : 1714
    • Reward points : 0
    • Joined: 2005/05/24 14:00:34
    • Location: Western NY State
    • Status: offline
    RE: Low Power Voltage Divider 2006/04/05 10:24:46 (permalink)
    0
    ORIGINAL: j_doin
    The MOSFETs leakage current is a little high, though, typically 1µA at ambient temp and rising to over 10µA for 50°C. Normally this is not a concern, especially for a 12V battery.

    Hmmm ... I hadn't realized this.  Other than looking for a "better" P-channel FET, I don't know what else could be done here. 
     
    For the N-channel driver, though, you could replace it with an npn transistor.  A common MMBT3904, for instance, has an off-leakage of 50nA. 
     
     
    #6
    Guest
    Super Member
    • Total Posts : 80500
    • Reward points : 0
    • Joined: 2003/01/01 00:00:00
    • Location: 0
    • Status: online
    RE: Low Power Voltage Divider 2006/04/05 12:07:11 (permalink)
    0
    ORIGINAL: DSchabel

    [...]Other than looking for a "better" P-channel FET, I don't know what else could be done here. 

    For the N-channel driver, though, you could replace it with an npn transistor.  A common MMBT3904, for instance, has an off-leakage of 50nA. 

    That's right, a BJT transistor can be a much better OFF switch, especially in higher temperatures. But the BJT VCE(sat) is not a very controlled parameter, and you often have very high VCE(sat) for low-leakage transistors. A MMBT2222 has a very low leakage of 1nA at 30°C, but has a non-controlled VCE(sat) of ~300mV. The PNP 2N2907 has even higher VCE(sat) of -400mV. Those values will vary from part to part, so unless you do calibration, you will have a error voltage that can be significant.
     
    ZETEX specify -1µA MAX leakage for the p-channel ZXM61P02F series, but offer no temperature derating info, so I am suspicious about this. I bet they are as leaky as any other.
     
    To get the utmost lower current, a JFET woud be necessary, with cutoff currents in the picoamp range. The caveats of a JFET are: it is a depletion FET, so you have to maintain a VGS more negative than any of its terminals to have it turned OFF, and its RDS(on) is a bit high, at the hundreds of ohms.
     
    If leakage currents under 1µA and simple circuitry are required, I would use an opto-FET. There are opto-FETs with remarkably low leakage current, like 1nA at 30°C, and rated maximum 5nA at ambient temp. The NEC PS7205B-1A is such a opto-FET. You can place the FET switch as a high-side switch. The caveat is that you have to drive the LED with 2mA to get a sample, but this can be done at a small duty cycle and sampling frequency, to keep overall current under the required budget. If you only sample at 0.5Hz to 0.1Hz, this can be met.
     
     
    #7
    invertedflight
    Senior Member
    • Total Posts : 139
    • Reward points : 0
    • Joined: 2004/11/14 19:01:31
    • Status: offline
    RE: Low Power Voltage Divider 2006/04/05 23:00:42 (permalink)
    0
    I had the IRLML2402 circled on my printout!  Although the IRLML2803 may win due to lower cost.  Do you know what the practical gate capacitance limit will be if I want to connect directly to the PIC's IO pin (no gate resistor)?  I want to use something that won't cause read-modify-write problems with the PIC16's.  I mostly just want to find something that I can buy a thousand of, use for everything, and not thing about for a long time.
     
    The powered-off divider is a clever idea, but I'm still leaning towards micropower op-amps.  I can probably build it to run at 10uA.  Maybe I'll build both setups into my pcb and try each.
     
     
    #8
    Guest
    Super Member
    • Total Posts : 80500
    • Reward points : 0
    • Joined: 2003/01/01 00:00:00
    • Location: 0
    • Status: online
    RE: Low Power Voltage Divider 2006/04/06 09:49:02 (permalink)
    0
    I have done that with the IRLML2402, got reels with 1000. Actually I used it in several low-volume industrial control modules and sensors that we manufatured a while ago, and I have got a reel left from production. Its main advantage is the very low VGS(sat), the lowest on the family with 0.7V. Its total gate charge is also one of the lowest of all power HEXFETs, and it has very good RDS(on) for the gate size. For the same reason, it has one of the fastest turn ON/OFF times, and lower Miller capacitance.
     
    The IRLML2502 is my next choice for switching loads up to 3.5A, like DC motors. Its gate is much larger than the IRLML2402, with a 12nC total gate charge, but has excellent RD(ON) and withstands all inductive abuses up to 3.5A.
     
    Either one can be driven directly from a PIC pin. The PIC output high-side channel resistance is roughly equivalent to a 100ohms resistor. A directly connected PIC output will switch a IRLML2402 in less than 220ns, and a IRLML2502 in less than 600ns.
     
    Today I still have about 500pcs of the IRLML2402, but have got 200pcs of the IRLML2502 and 200pcs of the IRLML6402, the p-channel equivalent of the IRLML2502.
     
    The micropower OpAmp approach is also good. I have used the LMC6061, a precision OpAmp with 10µA of quiescent current and 350µV of offset voltage, with a 10femtoamp input bias current. It is not rail-to-rail input, though.
     
    Microchip has the excellent RRIO MCP6041, a 600nA submicroamp current OpAmp. It has adequate open-loop gain and frequency response, and is conveniently sized in SOT-23. It has a high offset voltage of ±10mV, though.
     
    Either way, if you are using very high valued resistive dividers you have to be extra careful about stray capacitances, noise and bulk board leakage. You must use a guard ring at the divider high-impedance node, and drive the guard ring with the voltage follower output. Also, use no-clean flux solder paste, and a clean board.
     
     
     
     
    #9
    invertedflight
    Senior Member
    • Total Posts : 139
    • Reward points : 0
    • Joined: 2004/11/14 19:01:31
    • Status: offline
    RE: Low Power Voltage Divider 2006/04/07 17:28:23 (permalink)
    0
    Can you explain this guard ring?  I usually use water soluble paste...
     
    Also, what is offset voltage?
    #10
    Guest
    Super Member
    • Total Posts : 80500
    • Reward points : 0
    • Joined: 2003/01/01 00:00:00
    • Location: 0
    • Status: online
    RE: Low Power Voltage Divider 2006/04/07 18:12:00 (permalink)
    0
    Guard rings have a very suggestive name for their function. They are exactly that: a ring around the whole perimeter of a high-impedance node that is held at a known low-impedance voltage level, that copy the voltage of the high-impedance node.

    Look the following figure:


    The node connected to the noninverting input of the OpAmp (like your high-resistance resistive divider) is a very high impedance node. This means that any stray currents through the board can upset the voltage at that node. For example, a simple VDD track that is adjacent to the node can create a leakage current that can raise the voltage on the node. If you use a total resistance of 4megaohms, the current through your divider will be 3µA. A adjacent VDD track can easily leak 10nA to your measuring node, and that is enough to generate an error of 4LSb in your reading.

    In the figure above, the guard ring is the dotted line, and it must be laid-out to completely encircle the protected node.

    Likewise, a digital trace that is routed adjacent to the high-impedance node will couple the digital square wave to our sensitive node due to stray capacitive coupling. This can be around 15pf/cm of perimeter linear coupling, and is enough to ruin the measured analog voltage.

    The guard ring kills both error sources by placing a trace around the high-impedance node, on the complete perimeter, and making the voltage follower to drive this ring with the voltage read from the guarded node. Thus, very little leakage current will be present (much less than 1nA), because the 2 traces are at the same voltage, and also very little capacitive coupling will reach the sensitive node, because the OpAmp output will shunt any current before it reaches the protected node.

    I always place a small polygon on the underside of the guarded node, like a ground plane, but connected at the OpAmp output, to shunt all PCB bulk leakage currents as well.


    About offset voltage, I recommend you to read a good textbook on OpAmps. The Art of Electronics is one of the BEST references to learn about instrumentation and analog OpAmp circuit design.

    In a few words: every OpAmp with a negative feedback loop will try to keep a null voltage differential between its 2 inputs. So, if the voltage on the [+] input rises 0.5V, our voltage follower will raise the output whatever it takes to null the voltage differential, bringing the [-] input up by 0.5V also, thus copying the input. However, internal errors preclude the OpAmps to reach absolute zero voltage on the inputs, and a 'offset voltage' is always present. Precision OpAmps will have a offset voltage below 1mV, and high-precision OpAmps will have a offset voltage below 50µV. General-purpose OpAmps will have a 3mV to 10mV offset. If your circuit will be calibrated, the offset voltages are cancelled off in the calibration, but if you want uncalibrated accuracy, you need to use precision OpAmps. For a 10bit system, 5V VREF, 1LSb error is equivalent to 4.88mV. This means that you only have 4.88mV to distribute among all offset error sources and still have only 1LSb of offset error.


    #11
    invertedflight
    Senior Member
    • Total Posts : 139
    • Reward points : 0
    • Joined: 2004/11/14 19:01:31
    • Status: offline
    RE: Low Power Voltage Divider 2006/04/09 00:26:21 (permalink)
    0
    All of that makes perfect sense - thanks for the great explanations!  Seems making the guard ring will require me to route traces under my resistors, right?  I may have to use something bigger than 0805's then.
     
    BTW, I can't for the life of me find the pinouts for the IRLML FET's we've been discussing - can you help me?
     
    Thanks,
    Chris
    post edited by invertedflight - 2006/04/09 00:33:32
    #12
    Guest
    Super Member
    • Total Posts : 80500
    • Reward points : 0
    • Joined: 2003/01/01 00:00:00
    • Location: 0
    • Status: online
    RE: Low Power Voltage Divider 2006/04/09 12:41:53 (permalink)
    0
    ORIGINAL: invertedflight

    Seems making the guard ring will require me to route traces under my resistors, right?  I may have to use something bigger than 0805's then.
    A 0805 footprint has plenty of space to pass a PCB track below the resistor. I routinely use this to route boards without recurring to vias. If you can use 6mils space and 10mils traces, you can even route 2 tracks under a 0805 footprint.
     
    The trace running under C15, R11 and C14 on the following image is 20mils wide and has 10mils clearance:
     
     

    #13
    Guest
    Super Member
    • Total Posts : 80500
    • Reward points : 0
    • Joined: 2003/01/01 00:00:00
    • Location: 0
    • Status: online
    RE: Low Power Voltage Divider 2006/04/09 12:47:57 (permalink)
    0
    ORIGINAL: invertedflight

    BTW, I can't for the life of me find the pinouts for the IRLML FET's we've been discussing - can you help me?

    It's in the IRLML datasheet, first page: (component top view)
    [image]local://31049/019C2E7334B247A2830AC0DE54C5A4AF.GIF[/image]
    That is the exact pinout of a NPN bipolar transistor, such as the MMBT2222 and BC817, only that the Base <=> Gate, Collector <=> Drain, Emitter <=> Source. That makes easy to substitute NPNs for nFETs in switching applications.
     
    post edited by j_doin - 2006/04/09 12:50:45
    #14
    Guest
    Super Member
    • Total Posts : 80500
    • Reward points : 0
    • Joined: 2003/01/01 00:00:00
    • Location: 0
    • Status: online
    RE: Low Power Voltage Divider 2006/04/09 12:56:13 (permalink)
    0
    For IRF p-channel FETs, like the IRLML6402, and for PNP transistors alike, the pinout is the same (component top view):
    #15
    invertedflight
    Senior Member
    • Total Posts : 139
    • Reward points : 0
    • Joined: 2004/11/14 19:01:31
    • Status: offline
    RE: Low Power Voltage Divider 2006/04/09 22:20:50 (permalink)
    0
    Hmm, I was looking in the pb free data sheet, where they apprently left that out: http://www.irf.com/product-info/datasheets/data/irlml2803pbf.pdf
     
    I'm not sure what a standard 0805 pad is, but I'm using Cadsoft Eagle and the provided packages don't give nearly enough space to route between the pads.  I guess I'll have to modify them...
     
    Thanks,
    Chris
    #16
    Guest
    Super Member
    • Total Posts : 80500
    • Reward points : 0
    • Joined: 2003/01/01 00:00:00
    • Location: 0
    • Status: online
    RE: Low Power Voltage Divider 2006/04/10 13:12:48 (permalink)
    0
    Yes, you probably will need to modify the PCB footprints. I use a 40mils separation on my 0805 pads. That separation is good even when the production guys need to use a 0603 resistor to substitute a 0805 part. Actually, I always use mixed 0805/0603 resistors in my prototypes, since many 1% and 0.5% resistor values in my lab are 0603. I have used hot-air, IR reflow and standard oven reflow and 40mils look good for solder paste.
     
     
     
     
    #17
    chrislev
    Super Member
    • Total Posts : 318
    • Reward points : 0
    • Joined: 2005/02/18 11:04:37
    • Location: Kaysville, Utah, USA
    • Status: offline
    RE: Low Power Voltage Divider 2006/04/12 21:05:17 (permalink)
    0
    ORIGINAL: DSchabel

    If you switch the GND off, then you have a resistor from +12 to your PIC input pin.  You can make the resistance high enough so that the protection diodes in the PIC can shunt the current away, or, (preferably) add a protection diode externally.  But that means that you're STILL drawing current from the +12V supply into the PIC. 



    Ouch, ouch, ouch! I can't believe I did not catch that. Glad you guys spotted it. Especially since it took me a week to check back here. A little more thought and I would probably (hopefully) suggested switching off the divider at the battery instead of ground. My only excuse is that I've been doing a lot of ground switching lately so I have "ground switch on the brain".

    Double check everything you get from the Internet! :)



    -Chris
    post edited by chrislev - 2006/04/12 21:10:21
    #18
    Abraxas
    Super Member
    • Total Posts : 359
    • Reward points : 0
    • Joined: 2009/08/27 17:21:14
    • Location: 0
    • Status: offline
    RE: Low Power Voltage Divider 2010/04/11 21:01:28 (permalink)
    0
    It's all Geek to me. I am of humble stock. A petty 2 year electronics grad some 30 years ago. I'd like to know if you guys ever settled on a 12 V monitor. I'm building a 12 V battery powered project now and would like the 24Fj128ga060 to take some sort of action once my battery runs down to a certain point. I care nothing of the cost of the components as this is a one-off project. Might I humbly beg a final schematic from you ? Thanks.
    #19
    Jump to:
    © 2019 APG vNext Commercial Version 4.5