Battery monitoring system

Confirm that the 200 battery cells are all in series and that you need to measure the voltage across each one. This can lead you into quite a few problem areas unless you get your system organised very well to start with. Having confirmed this assumption then I am certain we will be able to steer your thoughts in the most appropriate direction.

There is nothing wrong in using the PC104 form factor and any ready built systems based on it for this type of task but your initial attention should really be in the interfacing aspects of this project. For the most part in projects like this a processor is a processor is a processor is a.......

--
********************************************************************
Paul E. Bennett ....................
Forth based HIDECS Consultancy .....
Mob: +44 (0)7811-639972 .........NOW AVAILABLE:- HIDECS COURSE......
Tel: +44 (0)1235-811095 .... see http://www.feabhas.com for details.
Going Forth Safely ..... EBA. www.electric-boat-association.org.uk..
********************************************************************

how many batteries in series? watch out for the common mode charactistics of your inputs. This sounds like a job for a flying capacitor input multiplexer like we used to use on RTUs.

Tom Woodrow

omar wrote:

Hi and Thanks!, Yes, the number of batteries is actually 420 in the string. But were using two data acquisition modules which would divide the number of cells in half (210). A single data acquisition module with a 16 channel A/D is currently how I'm conceptualizing each Data Acq. Module, so I would switch in 16 channels at a time from the string, through some multiplexing scheme into the D/A.

I'm curious about the "flying capacitor input multiplexer" Tom Woodrow mentioned. Tom, would you mind describing this in some more detail? I'll make an attempt to write up my ideas on the front end interface, which hopefully you'd be willing to comment upon.

Thanks, Omar

----== Posted via Newsfeeds.Com - Unlimited-Uncensored-Secure Usenet News==----

The #1 Newsgroup Service in the World! >100,000 Newsgroups

---= East/West-Coast Server Farms - Total Privacy via Encryption =---

With that many batteries in series, and ground assumed to be the bottom battery, you need some sort of isolation between both inputs and ground.

For example the 145th battery up the chain is floating at what (1.5 *

145) or about 220 volts and it must be measured from its negative to its positive terminal. This means that the (-) ADC input must be at 217.5 VDC and the (+) terminal at 219VDC.

The common mode limit if most ADC inputs is way less than 20VDC. The flying capacitor uses a method by which the capacitir is connected across an individual cell and charged and the capacitir is then switched away from the battery to the input of the ADC for processing.

Tom Woodrow

omar wrote:

Newsgroups

That brings up the question of what you can use to switch the capacitor. If you use solid state switches of some kind, then you still will have a common mode problem. Of course you could use dry relay contacts, but that would be messy.

-Robert Scott Ypsilanti, Michigan (Reply through this forum, not by direct e-mail to me, as automatic reply address is fake.)

No, can't use solid state. it has got to be relays or opto-isolated analog awitches.

In any case you MUST use break before make. We once used Thermosen relays and then switched CP Claire opto-isolated analog switches. We used four opto devices, two on the battery side and two on the measurement side.

BTW on a battery system like the one being talked about, the maximum float voltage during charge is way above the actual battery voltage. For instance a 130VDC substation battery float voltage during charging can be as high as 170+ volts. Gotta take that into consideration during any design.

Tom Woodrow

Robert Scott wrote:

address is fake.)

First thing I'd ask is, "What are you gonnna do with the information once you've acquired it?" It doesn't do any good to monitor something you can't control. If there's a control mechanism for each cell, is there some part of that that can be exploited in the measurement?

Do you REALLY, REALLY need to measure every cell. Can you get 80% of the value by measuring groups of cells? If you're only looking for "BAD" cells, you may not need much accuracy at all.

Some things don't scale very well...like this one. How about a completely different approach? Generate a ramp and use 200 comparators that output a pulse when the ramp passes a battery tap voltage. Measure the time between pulses and do the math. Hooking up 200 wires will cost you about as much as the 200 comparators. I have a vague vision of a floating comparator with inductively coupled pulse output somehow riding the ramp, but the details are still stuck in my xtal ball.

Or maybe a pair of optocouplers. One gets turned on to charge a cap from the cell thru the emitter of the second one that gets fed back to the measurement system. Probably have to measure more than one cell at a time to get enough volts to make it work. Put 'em in a circle and rotate the scanning light source ;-)

mike

--
Return address is VALID.
500MHz Tek DSOscilloscope TDS540 $2200
http://nm7u.tripod.com/homepage/te.html
Wanted, 12.1" LCD for Gateway Solo 5300. Samsung LT121SU-121
Bunch of stuff For Sale and Wanted at the link below.
http://www.geocities.com/SiliconValley/Monitor/4710/

Good heavens. 420 Pb/SO_4 cells in a single string? That's roughly one kV --- what on earth would you need *that* for?

I would strongly advise against any centralized ADC solution in this kind of setup. You'll be way better off with lots of small, local ADCs each monitoring a group of cells, and an opto-isolated or optical-signal field bus to connect them all to the central data collector. You *don't* want 1kV of potential difference driven by lead cells anywhere near your PC104.

--
Hans-Bernhard Broeker (broeker@physik.rwth-aachen.de)
Even if all the snow were burnt, ashes would remain.

That would give you problems in isolation between the two units. You do need to ensure proper isolation between the channels. Flying capacitor techniques or separate fully isolated ADC modules will be required for this.

[%X]

The usual way of applying a flying capacitor is by using a double pole double throw mercury wetted relay to switch the capacitor first across the battery cell then to across the ADC input. This way you can use one ADC and, by selecting each relay in turn, transfer the charge from the battery to the ADC in complete isolation. With about 1kV of battery you should pay particular attention to the wiring details and insulation requirements. As you stated that you will have a 16 channel A/D unit available you will be able to do the measuring in groups of 16.

Personally, for this sort of task, I would be inclined to design the best individual isolated ADC unit that could be networked and make enough of them to suit your needs. Pay close attention to the input to network isolation factors.

If you go with the flying capacitor method then pay close attention to insulation, creepage and clearance details on your circuit board. Spend out a biyt more for quality mercury wetted relays as skimping on these will lead to poorer performance.

As for the temperature measurements, you should ensure these are isolated from the voltage measuring circuits. If it is battery temperatures you are measuring then I expect either thermocouples or Pt100's might be the ones you choose.

--
********************************************************************
Paul E. Bennett ....................
Forth based HIDECS Consultancy .....
Mob: +44 (0)7811-639972 .........NOW AVAILABLE:- HIDECS COURSE......
Tel: +44 (0)1235-811095 .... see http://www.feabhas.com for details.
Going Forth Safely ..... EBA. www.electric-boat-association.org.uk..
********************************************************************

Sounds that you are going to need 25 completely isolated custom made boards, each handling 16 inputs.

With 16 batteries in series, the maximum tap voltage would be 48 V relative to the "local" ground tap. It would be practical to use simple resistive voltage dividers from each tap to the local ground tap, so that the voltage divider output would be in the range 0..5 V relative to the local ground tap, or whatever the ADC mux input range is. With 100 kohm : 10 kohm voltage divider, the ADC input voltage would be 1/11 of the tap voltage. With such high resistances, the power dissipation in each resistor would be quite low.

Powering each module would be easy, since you could get the power from

3-4 taps above the local ground tap.

To get the data out, asynchronous serial data could be used. The simplest form would be to use a current loop with optoisolators. Put all photo transistors in series (check for max Vce) in each module and feed the string by a constant current source. On each module, the transmit data controls the LED with the idle "1" keeping the LED on to maintain continuity in the series chain. You may have to use two separate chains, if the max Vce of the photo transistor of the optoisolator is less than about 75-100 V.

An other series chain would be needed to send the read requests individually to each module with some simple series protocol that can address each module individually.

Thus you would need 25 completely isolated custom made modules, each containing a simple processor with an ADC and a UART (or use bit banging with some parallel ports), a 16:1 multiplexor, sixteen

100kohm:10kohm voltage dividers, and two optoisolators and a voltage regulator.

The serial data current loop(s) would then be connected to your PC-104 or whatever processor board using one or two serial ports through optoisolators.

Paul

Why have a cap at all? Connect the cell to a optoA-led in series with a resistor and a second optoB-transistor. Driving the optoB-transistor will signal a optoA-transistor current at the measuring side.

This seems the simplest solution. It all boils down to the accuracy required. Most likely this needs calibration per cell/opty unit, but this can be handled from then on in software.

My first worry would be the variation of temperature on the led current and/or the current transfer ratio. Having said that, was one of your requirements not also to measure temperature? This makes the calibration/temperature compensation not too simple, but hey, the circuit will be.

By the way, minimum cell voltage that can be measured is probably around 1.5 to 2V or so if you keep the optoA-led current low when on.

Also watch out for the 'dark current' if you would connect all the optoB-transistors in parallel at the measuring side. Select low dark current ones or measure several separately.

Joop

I meant optoA-transistor here of course

Omar,

You might be interested in the "Exploring QNX Neutrino" article in the Jan. 2002 issue of Circuit Cellar Magazine that talks about QNX on a Versalogic PC104 Panther board.

The files associated with the article. talk about booting the system from compact flash and asynchronous serial communications. QNXBootCF.pdf includes additional references. QNXSerialio.pdf asynchronous serial communications. timer.c and watchdog.c are two demo files.

ftp://ftp.circuitcellar.com/pub/Circuit_Cellar/2001/138/mattern138.zip

As an alternative to a centralized measurement, have you considered a network of microcontrollers for measuring battery voltages, (RS422 network for example). This would allow you to scale the system and avoid messing with relay switching. Node costs would be relatively inexpensive, (example MicroChip PICKit board sells at $36 quantity 1) and it includes 4 10bit ADCs. You could measure temperature at each node. And you could easily replace the PC104 system in the future and still use the existing network as long as you continued to support the communications protocol. I've worked with such a system. Drop me a line if you want more details. AntiSPAM snipped-for-privacy@yahoo.com

RS422 would not be able to "practically" handle 100 channels at

100kHz, but I suspect this isn't really what you are doing since you only want to measure 16 channels at a time. I don't know your spec so I don't know how long you have to stream this data. I suppose if you wanted to use USB as a network to also power the micros that this might be possible, but I've not used USB under QNX. The USB1 would add to the per node cost, but would not give you a huge advantage over a well designed RS422 network..

Why do you have to measure battery voltage at 100kHz? What's the purpose of the PC104 form factor, (mainly size?)