Multi-pack battery voltage monitor

I guess when you say 'pack' you are referring to a multi-cell battery (12 Volts?)

Why not just off-set the Atmel device by 7 to 10 volts or so? 3

Hi Gary,

So is that 72x12V batteries?

What sort of accuracy do you need on the measured voltage?

How much can you load those batteries ?

How about a voltage divider across each 12 V battery, producing a voltage less than 5 V ?

The voltage divider could consist of a 7 V zener (the upper leg) and a resistor (the lower leg), thus the battery voltage swing would go into the IC without attenuation..

Once upon a time ® I designed (for California Micro Devices) a LiIon charge/discharge controller that tested and equalized all the cells (3) in a pak. I think it's used internally in present-day LiIon paks.

I have previously posted the back-to-back PMOS current-direction controller that was used. ...Jim Thompson

OK, I'll bite. Use a d'Arsonval movement and a resistor. Connect probes to the terminals and poke 'em at the battery cells, one at a time.

More useful nowadays would be a bevy of thermosensors and, as voltmeter, some kind of machine-addressable multiplexer into an ADC, though. With a single sense resistor and a bunch of simple reed relays feeding a current-frequency converter, you could get accurate and monotonic readings (but not fast; heck, battery state doesn't change fast anyhow).

Please clarify 72 packs, you talking 72 cells, or 72 x 3 or 6 cells?

In any case it's a high voltage, probably opto coupling would be required.

Are you monitoring or wanting to perform balancing as well?

Grant.

Maybe measure each cell junction with appropriate divider to a multiplexed ADC, software set for appropriate gain and sensitivity; derive each cell voltage wit some simple math.

I suspect that you really want to _monitor_ the voltage. Anyway, until you answer the questions you've already been asked, here's a generic solution that does not depend on those answers.

You could use a micro & some low current telecom relays like Digikey PB1168-ND to select the batteries individually. Conceptually, something like this:

/ +---o o---A | [Bat] Rly---------- uP port A0 | +---o o---B | \ | | / +---o o---A | [Bat] Rly---------- up port A1 | +---o o---B | \ }}} | / +---o o---A | [Bat] Rly---------- uP port B7 | +---o o---B \

The A and B points all connect to whatever circuit you're using to monitor the voltage.

You could also use p and n FETs and inverters to replace the relay contacts and save on $, at the cost of more assembly work. (The relays cost $3.19 each in low quantity.)

Ed

The OP stated up to 72 batteries. At 14 volts each, it would be near 1000 volts on the end. Some thought needs to be given to safety with whatever scheme is chosen. Of course, we never got any good follow-up on some of the questions asked.

tm

--- news://freenews.netfront.net/ - complaints: snipped-for-privacy@netfront.net ---

In Europe, still within the Low Voltage Directive (1500 Vdc).

When working with batteries, I would be more concerned with the potentially very high fault currents.

In the rely example above, if there is a malfunction in the relay control and two relays are active simultaneously or if , say contact B on a relay gets stuck and an other relay is activated, one or more batteries are shorted.

Of course, if the batteries have sufficient capacity and sufficiently low source resistance, the stuck relay contact is cleared :-), with the relay itself, as well as a lot of low power wiring going up in smoke. Thus it is essential that any wires going from the battery poles to the small signal relays are protected by suitable fuses.

For a large number of batteries in series, it is simpler to have individual electronics floating across each battery and then using optoisolators for communicating the data to a processor. After all,

Too right, but voltage sensing is easily isolated by resistors to control fault current.

I don't like the relay option for reliability, but they don't draw a standby load from batteries.

V to F -> optos probably easiest? Possibly some addressable scheme to reduce battery loading, another opto to enable each V/F unit?

Grant.

If the need is to monitor 12 volt lead acid batteries, then I would think it necessary to have at least a 0.1 volt accuracy. That is non-trivial at 1000 volts above ground.

tm

--- news://freenews.netfront.net/ - complaints: snipped-for-privacy@netfront.net ---

Non trivial? Easily enough done with opto couplers, there's 10kV and up ones with a LED one end and a phototransistor other end of a fairly long plastic tube. If they're not made any more, one could make their own.

Grant.

Opto couplers need circa 3V and 4 mA to drive 'em on and off. Is there a plan to make isolated power supplies for each of the seventy-two cells? Using opto couplers is not a real solution until you provide the support circuitry. You'd also want an ADC at each node, and put the digital output onto the optocoupler's drive amplifier.

Opto couplers need circa 3V and 4 mA to drive 'em on and off. Is there a plan to make isolated power supplies for each of the seventy-two cells? Using opto couplers is not a real solution until you provide the support circuitry. You'd also want an ADC at each node, and put the digital output onto the optocoupler's drive amplifier.

Personally, I would probably just use some little PICs with enough circuitry to handle the 12V power, and measure that power for each battery. I would then use an isolated comm channel to report the status of each battery. Maybe have one of them be a little bigger unit to monitor the comms from the others, and provide supervisory functions.

Heck, if I used a PSOC, it might be even cheaper...

Charlie

ote:

ies

ow-end

an to

=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D

l, change no$pam

ieee.

=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D

If you're putting a PSoC in there, transformers are probably lower power and cheaper than Optical Isolators. Maybe even capacitors in a "ring" topology to limit voltage. The data rate isn't an issue.

I see lots of power available there. Besides you can command the things to be off most of the time, battery voltage is a fast moving target ;) Two optos plus a small PIC per cell. Cheap, easy, not much support components needed.

Grant.