[TriEmbed] TriEmbed Digest, Vol 16, Issue 18 - Power regulator questions

Tue Jan 27 12:57:30 CST 2015

Chip,

Here is the link for using a high-side P-FET to connect your battery
voltage divider long enough to measure the voltage. This would let you use
lower value voltage divider resistors for a more accurate measurement.
http://jeelabs.org/2013/05/18/zero-power-measurement-part-2/

I did a blog post on options for reducing voltage divider power consumption.
http://fettricks.blogspot.com/2014/01/reducing-voltage-divider-load-to-extend.html

You would still need to measure your internal 1.1V reference with your ADC
so you can convert your ADC counts into millivolts. There may be a newer
article on this but I do not have it bookmarked if there is.
http://provideyourown.com/2012/secret-arduino-voltmeter-measure-battery-voltage/

Shane

On Tue, Jan 27, 2015 at 1:43 PM, Charles McClelland <chip at mcclellands.org>
wrote:

> Shane and all,
>
> I am coming back to this after some time.  I have re-red your message
> below a few times and read the linked article and I want to make sure I am
> understanding this correctly.
>
> Here is the problem I am looking to solve:  I am building a battery
> powered sensor that depending on the use case, could have battery packs
> supplying from 1.8 to 5.5V.  I then use a TI TPS63031 (which with Pete’s
> help we can now solder reliably despite the diabolical SON10 PowerPad
> packaging) a buck-boost regulator to provide up to 800mA at 3.3V.  This
> device will be one of those Internet of Things things so I want it to
> report its batter level so I know when it is time to go retrieve it and
> replace the batteries.
>
> I looked at the link below and decided that it may not relevant as it is
> solving for a problem I don’t think I have - that Vcc is not stable.  Not
> sure, so that is why I am reopening this thread.
>
> I am making one (large?) assumption, which I think makes my problem easier
> to solve:  That the TI buck-boost chip will do its best to do its duty to
> keep the voltage at 3.3V as long as the following are true: 1.8 <= Vin <=
> 5.5V, Imax <= 800mA, adequate filtering for noise and spikes in current.
>
> If this assumption is correct, I think I could simply use one of the
> Analog In pins to do the measurement.  Right?
>
> Also, I recall there was a discussion in one of our meetings about a
> circuit that would allow for dividing voltage without suffering constant
> current loss.  If you know of the link for that circuit, I would be
> interested.  I remember reading that as you increase the total resistance,
> you reduce that current loss but introduce potential error and the prospect
> of noise.  If the input impedance of the Arduino analog ports is 100M ohms,
> I am thinking that I can use the following:
> - Put a 0.01uF capacitor between the Analog pin and ground to reduce noise
> (battery levels change slowly after all)
> - Use the following resistors: Vbat to Analog 47k and Analog to ground
> 56k.  So when Vbat is 5.5V, Analog In should be 2.99V.
> - 100k total resistance is 0.033mA or 5.5mAH over a week which is only
> 0.3% of my LiPoly battery’s capacity
> - Vbat will then be = analogRead(Analog) * 6 (in mV) give or take percent
> or two
>
> Am I thinking about this right?
>
> Thanks,
>
> Chip
>
>
>
> Charles McClelland
> chip at mcclellands.org
> 919-624-5562
>
>
>
>
>
> On Sep 19, 2014, at 10:10 AM, Shane Trent <shanedtrent at gmail.com> wrote:
>
> Cool work Chip. I look forward to reading more about the project.
>
> I wanted to pass along a trick that allows you to measure your battery
> voltage in a standard Arduino setup without requiring a voltage divider. In
> short, they use Vcc as the ADC reference (as usual) but measure the 1.1V
> internal reference to calibrate the ADC and then convert ADC counts to
> millivolts. Just using the 1.1V reference on the ADC would require a
> voltage divider to drop Vcc to less than the 1.1V ADC max limited by the
> 1.1V reference. I am using it with a Li-poly battery powered project and
> seemed pretty accurate. There is extra code to get the correct registers
> for several chips but it is pretty straight forward.
>
>
> http://provideyourown.com/2012/secret-arduino-voltmeter-measure-battery-voltage/
>
> Shane
>
> On Fri, Sep 19, 2014 at 8:48 AM, Charles McClelland <chip at mcclellands.org>
> wrote:
>
>> Glen and Rodney, yes the tradeoffs part is typically how things work -
>> but not today.
>>
>> I am not sure exactly what RTFM stands for but I have received this
>> friendly advice in the past.  Well, when I looked at my FONA board’s
>> instruction set, it turns out there is a function in Adafruit’s library to
>> get the battery voltage.  So, looks like I am getting this functionality
>> for “free”.
>>
>> Once I saw this, I implemented the situation I described in yesterday’s
>> post.  My trail counter now sends the current battery voltage when it
>> reports the trail counts.  Event better, Ubidots (will do a writeup on them
>> for the blog) will send me an email or text message when the battery
>> voltage drops below a certain level.  While I am in the learning phase, I
>> think this will be very helpful.
>>
>> Here is what it looks like: <Screen Shot 2014-09-19 at 8.39.27 AM.png>
>>
>> <Screen Shot 2014-09-19 at 8.46.52 AM.png>
>>
>>
>> My next step is to add one more “widget” to send the signal strength
>> (more discovery in the Adafruit library) at the beginning like I do with
>> the GPS data.
>>
>> Of course, the FONA board is not cheap - $39 - so I know I am paying for
>> this functionality.
>>
>> Chip
>>
>>
>>
>> On Sep 18, 2014, at 3:13 PM, Glen Smith <mrglenasmith at gmail.com> wrote:
>>
>> Charles,
>>
>> As you said in a previous email, "As with all things electronic, there is
>> a trade off."
>>
>> How much is it going to add to your BOM cost is the bottom line. Once the
>> part is on board, measuring and transmitting the data is virtually free,
>> and for your prototypes, can provide useful information. Bench testing
>> batteries and extrapolating their lifetimes gets you close to an answer,
>> but ambient conditions on your bench are a far cry from real world. Getting
>> the metadata and watching it graph would be interesting to ME. But now we
>> start down a slippery slope again - how does temperature affect that
>> battery life? Let's add a temp sensor to your system... (Do I remember that
>> there is a temperature sensor in the 328? Google says yes:
>> http://playground.arduino.cc/Main/InternalTemperatureSensor ). So
>> assuming that the 328 is sleeping most of the time, the die should be
>> pretty close to the ambient temp inside your enclosure. Temperature might
>> very well affect trail usage - so reporting that regularly might be useful
>> for the primary function as well.
>>
>> Collecting a graph of battery voltage and temp vs time may be of value
>> for other projects that you come up with in the future. It may not be of
>> much interest to Parks & Rec - except to know when a sensor may need
>> maintenance. It might be nice if a sensor went unexpectedly off line to be
>> able to look at the last reported voltage. If it is close to nominal,
>> chances are something else happened. Close to the built in cut off voltage
>> of the battery would lead me to suspect things are still intact, but have
>> no juice.
>>
>> Personally, I think I would add a spot for the component on the board,
>> with a jumper setting or code support to know if it is populated. Early
>> deployed units would be populated and the data watched. If I discovered
>> that the data was boring, future units would lack the sensor.
>>
>> Glen
>>
>> On Thu, Sep 18, 2014 at 2:27 PM, Charles McClelland <chip at mcclellands.org
>> > wrote:
>>
>>> Rodney,
>>>
>>> Agreed, it is overkill for my current non-connected sensor for exactly
>>> the reasons you state below.  I should have mentioned that this is for the
>>> GPRS based connected sensor which has much higher and more variable power
>>> demands.
>>>
>>> I was reading this article
>>> <http://e2e.ti.com/blogs_/b/fullycharged/archive/2014/08/12/where-39-s-the-gauge-part-1.aspx?DCMP=fullyc&HQS=pwr-bms-gmap-wheresmygaugept1-fullyc-20140812-myti-en> from
>>> the TI power folks who - unsurprisingly - thought that fuel gauges should
>>> be included in all battery designs.  I started thinking about how emotional
>>> this engineer was getting with words like “baffled”, “excited” and
>>> “flustered”.  I figured there must be something here.  Full disclosure - TI
>>> also sent me a sample of their BQ27410-G1 Gas Gauge chip
>>> <http://www.ti.com/product/bq27410-g1>
>>>
>>> I am planning on putting a connected sensor in the local Ulmstead park
>>> to measure vehicle traffic.  I was thinking about the “metadata” I could
>>> send with each transmission and I realized that including the battery
>>> voltage might be a good thing.  I could watch the battery levels and know
>>> when I needed to swing by the park to change the battery.  I am using the
>>> Sparkfun one you pointed out below so, it is not about under voltage but
>>> about remote monitoring.  Over time, I will probably get a sense of how
>>> long a battery lasts so perhaps this is just for the learning phase.
>>> According to the RC blogs, cold weather can affect LiPO performance so, the
>>> battery life could change significantly as we go from Fall to Winter.
>>>
>>> Does this make sense or am I making this more complicated than it needs
>>> to be?
>>>
>>> Thanks,
>>>
>>> Chip
>>>
>>> On Sep 18, 2014, at 2:09 PM, Rodney Radford <ncgadgetry at gmail.com>
>>> wrote:
>>>
>>> The subject of Lipo batteries and battery fuel guage were discussed at a
>>> previous TriEmbed meeting.  For your application, I think it is overkill as
>>> what you are doing is constant - wake up, take a measurement, log it, and
>>> go back to sleep - so you should be able to determine the length of time a
>>> charge will last without it.
>>>
>>> The issue with Lipos is that you should not overuse them so they go
>>> undervoltage, but why not go with one with builtin overcharge, overcurrent
>>> and undervoltage circuitry built in, like this one from Sparfun:
>>>
>>> https://www.sparkfun.com/products/341
>>>
>>> They have a variety of sizes from from 40mah to several thousand mah, so
>>> you could first test your circuit with the smaller unit, determine how long
>>> it will last, and be able to better extrapolate out how big of a battery
>>> you would need to last for the amount of time you would need between
>>> charges.
>>>
>>>
>>> On Thu, Sep 18, 2014 at 1:11 PM, Charles McClelland <
>>> chip at mcclellands.org> wrote:
>>>
>>>> Michael,
>>>>
>>>> Thanks for taking a look at the power supply thread,  as my projects
>>>> run on batteries, I have put some time into this.
>>>>
>>>> Here are a few thoughts:
>>>> - Like you, my circuits draw a relatively small amount of current.  I
>>>> developed these tests at a higher current so they would not take a month to
>>>> run
>>>> - your current regulator (MCP1700-3302E
>>>> <http://ww1.microchip.com/downloads/en/DeviceDoc/20001826C.pdf>) is
>>>> Linear so it will dissipate - read waste - power equal to the product of
>>>> the difference between Vin and Vout time current (Eqn 6-1 in data sheet)
>>>> - This is where I started but I decided to move to a switched power
>>>> supply for three reasons:
>>>> - They are battery omnivores - since the TPS63031
>>>> <http://www.ti.com/lit/ds/symlink/tps63030.pdf> is both boost and buck
>>>> and is over 90% efficient over a large range (1.8-5.5Vin) - I can throw all
>>>> sorts of batteries at it
>>>> - As they are more efficient, they should throw off less heat -
>>>> potentially a big deal since I put them in a box and bury them
>>>> - This supply can handle from very small 25uA to 800mA currents
>>>>
>>>> As with all things in Engineering, there are some trade-offs I had to
>>>> accept:
>>>> - more expensive
>>>> - QFN packaging - ugh
>>>> - weird and exotic new component required - an inductor - no really!
>>>> There are a bewildering number of choices here!
>>>>
>>>> Some suggestions:
>>>> - I would be happy to loan you my test rig which you could use to test
>>>> your current power supply and compare it to the three I made.
>>>> - I added the Excel spreadsheet with the full set of tests and results
>>>> to the posting on the Triembed.org
>>>> <http://triembed.org/blog/?page_id=234> site
>>>> - I could give you a TPS63031 breakout board so you could test the
>>>> supply in your circuit without the pain of soldering - I have some extras
>>>> - You could look at commercial options such as the little breakout
>>>> boards from Pololu - I have one of these you could borrow
>>>> <http://www.pololu.com/product/2122> if you like
>>>>
>>>> Hope this helps.
>>>>
>>>> BTW, my next step in my power supply obsession is to add a battery fuel
>>>> gauge to the circuit so my Arduino can check it and so I could use LiPO
>>>> batteries with more confidence.
>>>>
>>>>
>>>> Chip
>>>>
>>>>
>>>>
>>>> On Sep 16, 2014, at 5:00 PM, triembed-request at triembed.org wrote:
>>>>
>>>> MCP1700-3302E
>>>>
>>>>
>>>>
>>>> _______________________________________________
>>>> Triangle, NC Embedded Computing mailing list
>>>> TriEmbed at triembed.org
>>>> http://mail.triembed.org/mailman/listinfo/triembed_triembed.org
>>>> TriEmbed web site: http://TriEmbed.org <http://triembed.org/>
>>>>
>>>>
>>>
>>>
>>> _______________________________________________
>>> Triangle, NC Embedded Computing mailing list
>>> TriEmbed at triembed.org
>>> http://mail.triembed.org/mailman/listinfo/triembed_triembed.org
>>> TriEmbed web site: http://TriEmbed.org <http://triembed.org/>
>>>
>>>
>>
>>
>> _______________________________________________
>> Triangle, NC Embedded Computing mailing list
>> TriEmbed at triembed.org
>> http://mail.triembed.org/mailman/listinfo/triembed_triembed.org
>> TriEmbed web site: http://TriEmbed.org <http://triembed.org/>
>>
>>
>
>

-- 
A blog about some of my projects.  http://fettricks.blogspot.com/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.triembed.org/pipermail/triembed_triembed.org/attachments/20150127/76eadff6/attachment.htm>