Re: [eaa-phev] Open Source Battery Management System
914 views
Skip to first unread message
Chris Ewert
Jan 3, 2008, 3:46:05 PM1/3/08
to eaa-phev, fre...@googlegroups.com
Per Hassel Sørensen wrote:
> I would like to shed some light on a impressive pair of BMS front end
> ICs, namely ISL9216 and ISL9217 from intersil. With a few extra parts,
> they protect, balance and monitor 12 Li-ion cells. Seven such sets can
> handle 84 cells, more than enough for a PHEV battery pack. The front
> end IC cost alone is approximately 1 US$ per cell.
>
> I imagine bulding several boards each handling monitoring and
> balancing of 12 cells in series. 7 such balancing & monitoring boards
> (BMB) communicate directly with the master BMS PCB using plastic fibre
> optic links. The master BMS control the charge and discharge FETs,
> temperature, and handles communication to/from the PHEV battery pack.
> This design keeps all the high current items on the master. The BMB
> modules does not handle high current and should not cost much.
This is interesting. If I understand correctly from the datasheet, the
chips basically do the level shifting and provide an output that goes to
an A/D converter on a microcontroller. Then you can control which cell
you want to measure via an I2C link and then turn on and off the
balancing shunt via I2C as well? So then one microcontroller would
still be responsible for the algorithm and would do the actual measuring
and determining of when to turn on the balancing FETs? The actual
balancing algorithm would still have to be written and run on the
microcontroller? The spec sheet says that the voltage measuring is
accurate to within 25mV (it looks like the majority of the error is at
cold or hot temperatures.)
> I would like to shed some light on a impressive pair of BMS front end
> ICs, namely ISL9216 and ISL9217 from intersil. With a few extra parts,
> they protect, balance and monitor 12 Li-ion cells. Seven such sets can
> handle 84 cells, more than enough for a PHEV battery pack. The front
> end IC cost alone is approximately 1 US$ per cell.
>
> I imagine bulding several boards each handling monitoring and
> balancing of 12 cells in series. 7 such balancing & monitoring boards
> (BMB) communicate directly with the master BMS PCB using plastic fibre
> optic links. The master BMS control the charge and discharge FETs,
> temperature, and handles communication to/from the PHEV battery pack.
> This design keeps all the high current items on the master. The BMB
> modules does not handle high current and should not cost much.
This is interesting. If I understand correctly from the datasheet, the
chips basically do the level shifting and provide an output that goes to
an A/D converter on a microcontroller. Then you can control which cell
you want to measure via an I2C link and then turn on and off the
balancing shunt via I2C as well? So then one microcontroller would
still be responsible for the algorithm and would do the actual measuring
and determining of when to turn on the balancing FETs? The actual
balancing algorithm would still have to be written and run on the
microcontroller? The spec sheet says that the voltage measuring is
accurate to within 25mV (it looks like the majority of the error is at
cold or hot temperatures.)
That looks interesting because it would significantly reduce the number
of PCBs, microcontrollers, electronics and the length of the
communication buses. How would EMI be managed with such a setup -
especially since the HV pack has to be isolated from the chassis
ground? I've been told in the past that I2C wasn't particularly good
when theres lots of noise.
(I'm cross posting this on the FreeBMS group since this is relevant
there as well.)
Chris
Chris Ewert
Jan 3, 2008, 6:03:42 PM1/3/08
to eaa-phev, fre...@googlegroups.com
Davide Andrea wrote:
>
> Balancing algorithm: see the
> ISL9208_16_17 Microcode Ref Guide 072707.pdf
> file in the zip archive that comes with the chips' documentation.
> Especially figure 13.
Thanks for finding that. Its very helpful, not just for this chip, but
for understanding how battery balancing systems in general work. A link
directly to the download is:
http://www.intersil.com/data/ev/isl92xx_eval_kit_software_and_docs_rel_v.1.41.zip
(the file is under the Software directory)
>
> This chip set seems to be ideal for low power battery packs, but I
> would be concerned about how these chips can handle the significant
> voltage fluctuations in the cell voltages in EV / PHEV applications.
> The instantaneous cell current jumps up and down at the PMW frequency
> of the motor inverters. I am afraid that the noise would create havoc
> with the communications between the 2 chips, if not worse. In a Prius,
> there's the additional 5 kHz, 200 Vpp noise from the inverters;
> dealing with it is not easy. (That's not the case with the Escape.)
The other thought that I had is that with this chip, you would be
reading 12 cells, but 1 at a time (since there is only 1 A/D converter
per 12 cells), what happens if the driver goes from accelerating to
regen braking from the time that cell #1's voltage was read to when cell
#12's voltage was read (or just if the current draw varies?) If you try
to do balancing while discharging and used sequentially collected data
as a premise for deciding which cells to balancing, that seems like it
would screw up the balance (though I think it would work well for
charging.) If you have one micro per cell and they are all on a
communication bus, a supervisor (or controller) could broadcast a
command, which would result in all the individual micros doing an A/D
conversion at the same time and then storing the data until the bus was
clear for sending to the supervisor. I would think that would probably
at least minimize the issue. Or perhaps the A/D conversions happen fast
enough that its not a problem, I'm not sure (but I would think even a
small difference would cause problems.)
>
> This chip set is designed to drive MOSFETs in series with the battery
> pack current. Of course, that's not likely to be the case in a PHEV.
> In a Prius or an Escape, you wouldn't turn off a MOSFET, you would
> tell the car to take easy for a bit (Charge Current Limit, Discharge
> Current Limit messages).
>
> Also, this chip set is designed to sense the current through a
> resistor. In a PHEV you're more likely to use a Hall Effect sensor.
It looks like the current sensing and the series MOSFET could be omitted
and the rest of the chip could be used.
>
> Balancing algorithm: see the
> ISL9208_16_17 Microcode Ref Guide 072707.pdf
> file in the zip archive that comes with the chips' documentation.
> Especially figure 13.
Thanks for finding that. Its very helpful, not just for this chip, but
for understanding how battery balancing systems in general work. A link
directly to the download is:
http://www.intersil.com/data/ev/isl92xx_eval_kit_software_and_docs_rel_v.1.41.zip
(the file is under the Software directory)
>
> This chip set seems to be ideal for low power battery packs, but I
> would be concerned about how these chips can handle the significant
> voltage fluctuations in the cell voltages in EV / PHEV applications.
> The instantaneous cell current jumps up and down at the PMW frequency
> of the motor inverters. I am afraid that the noise would create havoc
> with the communications between the 2 chips, if not worse. In a Prius,
> there's the additional 5 kHz, 200 Vpp noise from the inverters;
> dealing with it is not easy. (That's not the case with the Escape.)
The other thought that I had is that with this chip, you would be
reading 12 cells, but 1 at a time (since there is only 1 A/D converter
per 12 cells), what happens if the driver goes from accelerating to
regen braking from the time that cell #1's voltage was read to when cell
#12's voltage was read (or just if the current draw varies?) If you try
to do balancing while discharging and used sequentially collected data
as a premise for deciding which cells to balancing, that seems like it
would screw up the balance (though I think it would work well for
charging.) If you have one micro per cell and they are all on a
communication bus, a supervisor (or controller) could broadcast a
command, which would result in all the individual micros doing an A/D
conversion at the same time and then storing the data until the bus was
clear for sending to the supervisor. I would think that would probably
at least minimize the issue. Or perhaps the A/D conversions happen fast
enough that its not a problem, I'm not sure (but I would think even a
small difference would cause problems.)
>
> This chip set is designed to drive MOSFETs in series with the battery
> pack current. Of course, that's not likely to be the case in a PHEV.
> In a Prius or an Escape, you wouldn't turn off a MOSFET, you would
> tell the car to take easy for a bit (Charge Current Limit, Discharge
> Current Limit messages).
>
> Also, this chip set is designed to sense the current through a
> resistor. In a PHEV you're more likely to use a Hall Effect sensor.
It looks like the current sensing and the series MOSFET could be omitted
and the rest of the chip could be used.
Chris
Mike
Jan 5, 2008, 2:15:43 AM1/5/08
to FreeBMS
Good work Chris and Hassel Sørensen, this chip is just what the doctor
ordered, however after reading Intersill's documentation I noticed
that the chipset is "in development", and none of their vendors have
stock, or even the part number in their inventory system. I have not
been able to get any information on when "real" chips will be
available. It is very common in the silicon Valley to announce a
product months or even years before it ships. About the second time
they slip the shipping date there is mention designing out the
Unobtanium before starting volume production.
Mike
ordered, however after reading Intersill's documentation I noticed
that the chipset is "in development", and none of their vendors have
stock, or even the part number in their inventory system. I have not
been able to get any information on when "real" chips will be
available. It is very common in the silicon Valley to announce a
product months or even years before it ships. About the second time
they slip the shipping date there is mention designing out the
Unobtanium before starting volume production.
Mike
Per Hassel Sørensen
Jan 5, 2008, 3:11:19 PM1/5/08
to fre...@googlegroups.com
2008/1/5, Mike <toms...@cruzio.com>:
Good work Chris and Hassel Sørensen, this chip is just what the doctor
ordered, however after reading Intersill's documentation I noticed
that the chipset is "in development", and none of their vendors have
stock, or even the part number in their inventory system. I have not
been able to get any information on when "real" chips will be
available.
Mike,
That is not correct. These parts became available in May 2007. I received a few samples of both ISL 9216 and ISL9217 half a year ago. They are available in quantities from Rose electronics, search for ISL9216 and ISL9217at
http://www.roseelectronicsinc.com/search.html They had today more than 6000 pieces of each. Alternatively, more than 25 thousand chips is in stock of the single chip version ISL9208 from the same chip set family. This chip has the same functionality for up to 7 cells.
Per
Mike
Jan 6, 2008, 1:44:34 AM1/6/08
to FreeBMS
OK if they are available we can use them to greatly improve our cost/
performance ratio. For those using NiCad or NiMHD it may be possible
to monitor them in sets of three with these chips. Software will have
to handle the problem of one cell of a set going bad, or way out of
balance. It is interesting that Intersill's site sent me to a vendor
that didn't know about this part number?
I will start another thread on balancing cells.
Mike
On Jan 5, 12:11 pm, "Per Hassel Sørensen" <perhas...@gmail.com> wrote:
> 2008/1/5, Mike <tomsw...@cruzio.com>:
performance ratio. For those using NiCad or NiMHD it may be possible
to monitor them in sets of three with these chips. Software will have
to handle the problem of one cell of a set going bad, or way out of
balance. It is interesting that Intersill's site sent me to a vendor
that didn't know about this part number?
I will start another thread on balancing cells.
Mike
On Jan 5, 12:11 pm, "Per Hassel Sørensen" <perhas...@gmail.com> wrote:
> 2008/1/5, Mike <tomsw...@cruzio.com>:
>
>
>
> > Good work Chris and Hassel Sørensen, this chip is just what the doctor
> > ordered, however after reading Intersill's documentation I noticed
> > that the chipset is "in development", and none of their vendors have
> > stock, or even the part number in their inventory system. I have not
> > been able to get any information on when "real" chips will be
> > available.
>
> Mike,
> That is not correct. These parts became available in May 2007. I received a
> few samples of both ISL 9216 and ISL9217 half a year ago. They are available
> in quantities from Rose electronics, search for ISL9216 and ISL9217athttp://www.roseelectronicsinc.com/search.htmlThey had today more than 6000
>
>
> > Good work Chris and Hassel Sørensen, this chip is just what the doctor
> > ordered, however after reading Intersill's documentation I noticed
> > that the chipset is "in development", and none of their vendors have
> > stock, or even the part number in their inventory system. I have not
> > been able to get any information on when "real" chips will be
> > available.
>
> Mike,
> That is not correct. These parts became available in May 2007. I received a
> few samples of both ISL 9216 and ISL9217 half a year ago. They are available
Mike
Jan 19, 2008, 1:04:46 AM1/19/08
to FreeBMS
This time delay is not a problem. The microprocessor can select a
cell, read its voltage, and filter it several times in one ms. If any
significant change in battery current happens in less than 200 ms
there is a design problem with the motor controller, or the throttle
response system. There should be some analog filtering on the front
end for high frequency noise then the micro can add more if necessary.
The communications system should be able to handle two or three reads
per second, although we probably will not use that much speed.
On Jan 3, 3:03 pm, Chris Ewert <ch...@infolaunch.com> wrote:
> Davide Andrea wrote:
>
> > Balancing algorithm: see the
> > ISL9208_16_17 Microcode Ref Guide 072707.pdf
> > file in the zip archive that comes with the chips' documentation.
> > Especially figure 13.
>
> Thanks for finding that. Its very helpful, not just for this chip, but
> for understanding how battery balancing systems in general work. A link
> directly to the download is:http://www.intersil.com/data/ev/isl92xx_eval_kit_software_and_docs_re...
cell, read its voltage, and filter it several times in one ms. If any
significant change in battery current happens in less than 200 ms
there is a design problem with the motor controller, or the throttle
response system. There should be some analog filtering on the front
end for high frequency noise then the micro can add more if necessary.
The communications system should be able to handle two or three reads
per second, although we probably will not use that much speed.
On Jan 3, 3:03 pm, Chris Ewert <ch...@infolaunch.com> wrote:
> Davide Andrea wrote:
>
> > Balancing algorithm: see the
> > ISL9208_16_17 Microcode Ref Guide 072707.pdf
> > file in the zip archive that comes with the chips' documentation.
> > Especially figure 13.
>
> Thanks for finding that. Its very helpful, not just for this chip, but
> for understanding how battery balancing systems in general work. A link
Per Hassel Sørensen
Jul 10, 2008, 4:49:36 AM7/10/08
to fre...@googlegroups.com
Is someone is still listening here on FreeBMS?, let's give it a go:
I feel the existing homemade BMS systems are to complex and expensive and have reliability issues when used in large batteries. I want to build a BMS tailored for Li-ion using ISL9216/17 chips. That will make it less complex and expensive and hopefully more reliable. I am planning on building a modular BMS using these ISL chips. The BMS will first be tested on an e-bike before being used on PHEV and EV. Plan on starting this fall but I need help from a software guy, anyone out there?
The Intersil ISL9208 and ISL9216/17 have some very nice features; see a short interview with one of their application engineers at: http://www.youtube.com/watch?v=oiRhChE6rt8 who actually mentions PHEVs!
Cheers,
Per Hassel
Joe Ellerbach
Jul 29, 2008, 12:03:22 AM7/29/08
to FreeBMS
Per Hassel,
I am just getting involved in PHEV's now. I may be able to help. I'm
currently trying to decide what battery technology to pursue. I want
to use Li-ion if I can find a supplier at a reasonable cost. I'm
still trying to understand the full impact on the electronics as well
before I lock myself into Li-ion. Anyway, let me know if you are
still looking for help.
Thanks,
Joe
> short interview with one of their application engineers at:http://www.youtube.com/watch?v=oiRhChE6rt8who actually mentions PHEVs!
>
> Cheers,
> Per Hassel
I am just getting involved in PHEV's now. I may be able to help. I'm
currently trying to decide what battery technology to pursue. I want
to use Li-ion if I can find a supplier at a reasonable cost. I'm
still trying to understand the full impact on the electronics as well
before I lock myself into Li-ion. Anyway, let me know if you are
still looking for help.
Thanks,
Joe
>
> Cheers,
> Per Hassel
Reply all
Reply to author
Forward
0 new messages