Searching \ for '[EE] Measurement of cell voltages within a battery' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: www.piclist.com/techref/power/batterys.htm?key=battery
Search entire site for: 'Measurement of cell voltages within a battery'.

Exact match. Not showing close matches.
PICList Thread
'[EE] Measurement of cell voltages within a battery'
2005\08\08@070237 by Michael Rigby-Jones

picon face
I'm trying to design a battery monitor circuit that needs to know individual cell voltage to a good level of precision (+- 5mv would be great, 10mv acceptable) over a temperature range of 10-90 degrees C.  A PIC will be digitizing the voltages with it's 10bit ADC.  Cells are nominaly 3-4 volts, and there could be up to 6 of them in series in a battery.  Now the killer, it must to be as cheap as possible.

I have racked my brains over this and come up with the following
1) Potential divider from each cell tap to ground.  This works fine for the first cell, but resolution rapidly drops off as you move up the battery making this unacceptable.  Obviously the PIC has a bit of math to do to get individual cell voltages.

2) Differential amplifiers wired across each cell.  Works well, but cost becomes prohibitive with decent op-amps, and the voltage levels in the upper cells gets problematic for many cheap rail-rail op-amps e.g. Microchips offerings.

3) Simple transistor level shifter.  This would be perfect apart from the awfull tempco. Can this be compensated for somehow? (at the analog level rather than PIC level, there can be significant temperature gradiants across the battery).

4) Voltage controlled current source with an op-amp and a transistor per cell, essentialy a better version of the above.  Favorite at the moment, but would dearly like to find something simpler/cheaper.

5) Individual ADC's and references for each cell, as good as your going to get for resolution but blows the budget in a big way.

Any ideas gratefully received!


Regards

Mike Rigby-Jones

=======================================================================
This e-mail is intended for the person it is addressed to only. The
information contained in it may be confidential and/or protected by
law. If you are not the intended recipient of this message, you must
not make any use of this information, or copy or show it to any
person. Please contact us immediately to tell us that you have
received this e-mail, and return the original to us. Any use,
forwarding, printing or copying of this message is strictly prohibited.
No part of this message can be considered a request for goods or
services.
=======================================================================

2005\08\08@072202 by Vasile Surducan

face picon face
Hi,
We've talk about a similar design on jallist a while ago. See my opinions below.

On 8/8/05, Michael Rigby-Jones <spam_OUTMichael.Rigby-JonesTakeThisOuTspambookham.com> wrote:
> I'm trying to design a battery monitor circuit that needs to know individual cell voltage to a good level of precision (+- 5mv would be great, 10mv acceptable) over a temperature range of 10-90 degrees C.  A PIC will be digitizing the voltages with it's 10bit ADC.  Cells are nominaly 3-4 volts, and there could be up to 6 of them in series in a battery.  Now the killer, it must to be as cheap as possible.

6x4 = 24V maximum voltage on sistem

>
> I have racked my brains over this and come up with the following
> 1) Potential divider from each cell tap to ground.  This works fine for the first cell, but resolution rapidly drops off as you move up the battery making this unacceptable.  Obviously the PIC has a bit of math to do to get individual cell voltages.
>
> 2) Differential amplifiers wired across each cell.  Works well, but cost becomes prohibitive with decent op-amps, and the voltage levels in the upper cells gets problematic for many cheap rail-rail op-amps e.g. Microchips offerings.

you'll need 24V common mode for the last OA , consider maybe optoisolated OA
(differential topology is not necessary anymore)

>
> 3) Simple transistor level shifter.  This would be perfect apart from the awfull tempco. Can this be compensated for somehow? (at the analog level rather than PIC level, there can be significant temperature gradiants across the battery).

you'll not measure the cell but an assembly of factors, some of them
temperature dependant

>
> 4) Voltage controlled current source with an op-amp and a transistor per cell, essentialy a better version of the above.  Favorite at the moment, but would dearly like to find something simpler/cheaper.

floated voltage to current followed by current to voltage at the PIC
input, quite complicated but nice
>
> 5) Individual ADC's and references for each cell, as good as your going to get for resolution but blows the budget in a big way.

if 10 bits are enough, 12F675 for every cell, followed by an
optoisolated communication

>
> Any ideas gratefully received!

6 DPDT small relays working as a an analogic multiplexer ?

best wishes,
Vasile

2005\08\08@073659 by Russell McMahon

face
flavicon
face
> I'm trying to design a battery monitor circuit that needs to know
> individual cell voltage to a good level of precision (+- 5mv would
> be great, 10mv acceptable) over a temperature range of 10-90 degrees
> C.  A PIC will be digitizing the voltages with it's 10bit ADC.
> Cells are nominaly 3-4 volts, and there could be up to 6 of them in
> series in a battery.  Now the killer, it must to be as cheap as
> possible.

Trivially easy :-).
Well, almost.

Sigma Delta ADCs. Precision is as high as you'd reasonably wish. Speed
is low but should be no problem here.
Using PIC pins alone as the "comparators" Scott Datallo got 12 bits
AFAIR. By using something like LM339 quad comparators you can get far
greater precision. You could eg use a quad of these for the upper 4
cells and straight PIC pins for the lowest 2 cells. The LM339 is about
as cheap an IC as you can buy (around $US0.10 from a realistic source
AFAIR).

I've had bad luck with SD but blame Zilog :-)

Freewheeling brain mode - sensibility not guaranteed:

You may have to define acceptably low price for other solutions to be
checked.

You may be able to generate a precision voltage with a PWM DAC and use
it to stand a lower resolution ADC on. Some ICs allow the top and
bottom of the reference to be floated so you can position the ADCs
sample space anywhere in the supply range.

You may be able to multiplex a single high accuracy ADC (eg Sigma
Delta) with a suitably low cost MUX. This makes sense if the mux cost
can be less than multiple sigma delta converters. A sensibly
realisable SD converter can probably be 16 to 20 bits. Insanely
unrealisable ones that are nevertheless available run to 24+ bits.


       RM

2005\08\08@082211 by Alan B. Pearce

face picon face
>> 2) Differential amplifiers wired across each cell.  Works well,
>but cost becomes prohibitive with decent op-amps, and the voltage
>levels in the upper cells gets problematic for many cheap
>rail-rail op-amps e.g. Microchips offerings.
>
>you'll need 24V common mode for the last OA , consider maybe
>optoisolated OA (differential topology is not necessary anymore)

I don't think you need 24V common mode, as using suitable resistors on a
standard differential input circuit would bring it down lower than 24V at
the op-amp. I would have thought that measuring the bottom cell direct, then
a quad op-amp for the next four cells, and maybe a resistor divider from the
top one, with the actual voltage obtained by adding the other voltages and
taking the difference,

...
>> 5) Individual ADC's and references for each cell, as good as your
>> going to get for resolution but blows the budget in a big way.
>
>if 10 bits are enough, 12F675 for every cell, followed by an
>optoisolated communication

Or even a 10F device per cell, using Russell's suggestion of it doing sigma
delta conversion. Would be worth hitting Microchip up for a bunch of samples
to try it.

2005\08\08@082303 by Gerhard Fiedler

picon face
Michael Rigby-Jones wrote:

> I'm trying to design a battery monitor circuit that needs to know
> individual cell voltage to a good level of precision (+- 5mv would be
> great, 10mv acceptable) over a temperature range of 10-90 degrees C.  A
> PIC will be digitizing the voltages with it's 10bit ADC.  Cells are
> nominaly 3-4 volts, and there could be up to 6 of them in series in a
> battery.  Now the killer, it must to be as cheap as possible.

How about one PIC per cell? If the cells have >3V, that should be good
enough. And could be cheap PICs... the only thing you need is a 10-bit ADC,
a GPIO (and an internal oscillator).

Communication (one-way) could be done with optos or with some kind of
digital level shifter, which probably isn't too difficult to do with a
couple resistors and a schmitt-trigger gate or so.

The PIC that's connected to the grounded cell measures that cell and
receives the comms from the others and presents the result to the outside
world.

Gerhard

2005\08\08@082553 by Michael Rigby-Jones

picon face


{Quote hidden}

Oops, forgot to mention that one, been a long way down this path.  Main problem is simply the length of time it takes to take a sample.  With the cheapest 18pin nanowatt PIC I was using (16F627A) internal clock speed is just 4MHz, and the best I could get each ADC phase down to was 8 clock cycles.  For a 12 bit converter this is ~33ms per conversion.  This normaly be ok, but the goal was to have the PIC regularly wake up, sample everything and go back to sleep whenever the current draw from the battery is ~zero, in order to put minimal current draw on the battery.  With 6 cells, temperature monitor and current monitor the PIC is going to be spending a lot of time awake which means higher discharge when the battery is idle.


{Quote hidden}

That's an interesting idea,  I'll have a look into this.

>You may be able to multiplex a single high accuracy ADC (eg Sigma
>Delta) with a suitably low cost MUX. This makes sense if the mux cost
>can be less than multiple sigma delta converters. A sensibly
>realisable SD converter can probably be 16 to 20 bits. Insanely
>unrealisable ones that are nevertheless available run to 24+ bits.

Yep, intial design use a 4051 + software ADC.  It's the cheapeast solution (by far) I've found with the above limitation.  Another thing I didn't mention is that the cell monitors must draw no (or at least very little << 1uA) current when not being used, so far I'm using a bunch of P-Channel MOSFETs to switch each section.

The entire electronic assemebly I'm aiming to get for under GPB5.00 including PIC (16F687 ~GBP1.10), small PCB (~GBP1.20), bunch of LEDs, piezo sounder, MOSFETS etc.  I've probably absorbed another 1.50 in the rest of the design, leaving about GPB1.20 for the voltage monitoring (not including switching MOSFETS if required).

Regards

Mike

=======================================================================
This e-mail is intended for the person it is addressed to only. The
information contained in it may be confidential and/or protected by
law. If you are not the intended recipient of this message, you must
not make any use of this information, or copy or show it to any
person. Please contact us immediately to tell us that you have
received this e-mail, and return the original to us. Any use,
forwarding, printing or copying of this message is strictly prohibited.
No part of this message can be considered a request for goods or
services.
=======================================================================

2005\08\08@083828 by Michael Rigby-Jones

picon face


{Quote hidden}

I have thought about this, but you haven't mentioned the expensive bit...a decent voltage reference!  With one PIC, you need only one voltage reference (which could be Vdd if the regulator is good enough), but with multiple floating PICs/ADCs, each needs it's own reference.

The cheapest PIC with an ADC seems to be the 12F675 at ~GBP0.80 and they also sell a voltage refence for 0.25. (0.80+0.25)*6 = 6.30, entire budget blown.

Using a 10F part with a comparator (for delta sigma ADC) on buy.microchip is GPB 0.39, plus the same reference comes to 3.84, still too expensive.

Regards

Mike

=======================================================================
This e-mail is intended for the person it is addressed to only. The
information contained in it may be confidential and/or protected by
law. If you are not the intended recipient of this message, you must
not make any use of this information, or copy or show it to any
person. Please contact us immediately to tell us that you have
received this e-mail, and return the original to us. Any use,
forwarding, printing or copying of this message is strictly prohibited.
No part of this message can be considered a request for goods or
services.
=======================================================================

2005\08\08@090158 by Gerhard Fiedler

picon face
Michael Rigby-Jones wrote:

>>Sigma Delta ADCs.

> Oops, forgot to mention that one, been a long way down this path.  Main
> problem is simply the length of time it takes to take a sample. [...]
> With 6 cells, temperature monitor and current monitor the PIC is going
> to be spending a lot of time awake which means higher discharge when the
> battery is idle.

Can't you assume that the voltage doesn't change too fast? Temperature is
probably quite slow. You didn't say how fast/frequently you need to monitor
the cells.

Gerhard

2005\08\08@090542 by Michael Rigby-Jones

picon face


{Quote hidden}

Once per second (ish).

Regards

Mike

=======================================================================
This e-mail is intended for the person it is addressed to only. The
information contained in it may be confidential and/or protected by
law. If you are not the intended recipient of this message, you must
not make any use of this information, or copy or show it to any
person. Please contact us immediately to tell us that you have
received this e-mail, and return the original to us. Any use,
forwarding, printing or copying of this message is strictly prohibited.
No part of this message can be considered a request for goods or
services.
=======================================================================

2005\08\08@090747 by Michael Rigby-Jones

picon face


{Quote hidden}

The other snag with Sigma-Delta is the precharge time , which can exceed the conversion time by quite a margin if the input voltage is near the limits of the converter.  You can avoid this to some extent by designing the converter voltage limits to be outside the normal input range, but then you lose resolution so have to perform more conversion phases.

Regards

Mike

=======================================================================
This e-mail is intended for the person it is addressed to only. The
information contained in it may be confidential and/or protected by
law. If you are not the intended recipient of this message, you must
not make any use of this information, or copy or show it to any
person. Please contact us immediately to tell us that you have
received this e-mail, and return the original to us. Any use,
forwarding, printing or copying of this message is strictly prohibited.
No part of this message can be considered a request for goods or
services.
=======================================================================

2005\08\08@092258 by Christian A. Weagle

flavicon
face
> 2) Differential amplifiers wired across each cell.  Works well, but
cost
> becomes prohibitive with decent op-amps, and the voltage levels in the
> upper cells gets problematic for many cheap rail-rail op-amps e.g.
> Microchips offerings.

Analog switches to connect the positive and negative of each battery to
_a_ diff amp, sequentially.  Or, if a diff amp is not in the cards, use
a capacitor to hold the voltage, and then transfer it to
ground-referenced.  Needs more switches, but less common-mode worries.

2005\08\08@093026 by Michael Rigby-Jones

picon face


{Quote hidden}

Using a cap and switches to convert from floating to ground referenced is a clever idea, I like it.  Leakage currents will have to be considered quite carefully though.

Regards

Mike

=======================================================================
This e-mail is intended for the person it is addressed to only. The
information contained in it may be confidential and/or protected by
law. If you are not the intended recipient of this message, you must
not make any use of this information, or copy or show it to any
person. Please contact us immediately to tell us that you have
received this e-mail, and return the original to us. Any use,
forwarding, printing or copying of this message is strictly prohibited.
No part of this message can be considered a request for goods or
services.
=======================================================================

2005\08\08@101743 by Christian A. Weagle

flavicon
face
> Using a cap and switches to convert from floating to ground referenced
> is a clever idea, I like it.  Leakage currents will have to be
> considered quite carefully though.

Sure, but ADCs are pretty fast these days.  ;)

You may also want to parallel a bunch of the switches to get lower Rs
(for quick cap charging), especially if you are using cheap 4000 logic.

I think Linear has an app note that mentions this technique, ISTR that
it is used for the LT2400 24-bit SE ADC (to allow diff signals).  Try
searching their site for that...

2005\08\08@172946 by Peter

picon face

6) flying capacitor sampler switched with cmos switches that withstand
20V (you said 4 cells ?). With some ingenuity you could 'lift' the
ground of the circuit to the middle of the battery and then the switches
only need to withstand 2 cells. That puts it in range for older CMOS
parts like CD4052 & co. 2 cells will never see more than 10V across
them, leaving some margin for a 3V powered PIC floating above them
(makes 13V total supply). Now you only have to shift the logic control
for the lower CMOS switches (the upper one is direct driven), and that
is easy. Low power too, use only 4 chips (2 x 4052, pic, 3V voltage
regulator), and no divider networks. The power drain should be just the
static cmos quiescent current when not driven (and the lower cmos logic
pulled low and drawing no current through the level shifters).

Of course 'some work' is required but the concept seems sound (at 0024
hours here).

Peter

2005\08\08@173335 by Peter

picon face

On Mon, 8 Aug 2005, Christian A. Weagle wrote:

> Analog switches to connect the positive and negative of each battery to
> _a_ diff amp, sequentially.  Or, if a diff amp is not in the cards, use
> a capacitor to hold the voltage, and then transfer it to
> ground-referenced.  Needs more switches, but less common-mode worries.

Hmm I just posted such a solution before reading your message.

Peter

More... (looser matching)
- Last day of these posts
- In 2005 , 2006 only
- Today
- New search...