Since this is a well defined system, in the sense that the composition
of components is normally static, it seems reasonable to have a
centrally controlled communication protocol. The BMS for each cell would
normally be quiet and only send something when prompted by a central BMS
component. You can imagine building some additonal layers of safety such
as the cell automatically shutting down if no poll comes from the
central BMS for some period of time.
TCP/IP is probably overkill but it's well established and could be used
to ensure fail-safe communications. If one is to embed a radio in each
cell with some logic processing, it seems a small extension to add the
bottom layers of comm.
I suppose I'm regurgitating what's been done with wired systems, but it
seems straight forward enough.
Peri
------ Original Message ------
From: "Bill Dube via EV" <[email protected]>
To: [email protected]
Cc: "Bill Dube" <[email protected]>
Sent: 28-Apr-20 7:46:26 PM
Subject: Re: [EVDL] Communicating over "mains" (Was: Minibms?)
Indeed, you would have to design the entire system, inverter, charger, etc. to
accommodate this communication scheme. Not an "add-on" type BMS my any means.
All of the "noise" comes from the components, and the designers have control over each
and every component in that environment. A regularly spaced, "moment of silence" is not
difficult to coordinate if you talk to every component with the CAN bus. It can be so short that
the user will not be able to sense that it has happened.
You can also pick your communication band without a care, because you are on a private
"network". Perhaps have the cell BMS sweep to find a band that works best
during commissioning.
The input caps might block the cell BMS RF signal from _entering_ the inverter
(or charger). However, the signal will still be present and detectable as a
_current_ in the traction wiring, rather than as a voltage on the input
terminals of the inverter (or charger.)
Bill D.
On 4/29/2020 1:48 PM, Lee Hart via EV wrote:
Bill Dube via EV wrote:
I have thought about doing this for perhaps 20 years. It may well be
possible to communicate via the traction conductors. It is probably
worth the effort to do so because it would allow you to incorporate the
BMS in the cell. Sealing the BMS inside each cell could be very useful,
especially from a warranty/liability angle. The cell manufacturer would
love to have a log of the SOC history of the cell.
It is certainly possible. The problem is whether it is practical.
If you are an automaker, with control over every aspect of the vehicle, the situation may
be manageable. You can pick a part of the RF spectrum for your BMS communications where
you know (or create) a "hole" in the noise from the other parts of the vehicle.
You can also route your wiring so as not to create any "dead spots". When you
don't have a controlled impedance (known capacitance and inductance in the wiring), RF
systems will have peaks and nulls that can prevent certain locations from communicating,
where moving it a foot down the wire either way works.
But I think the situation is nearly hopeless in an open-source hobby EV. It
would boil down to trial and error, where the installer doesn't know what noise
the pieces are producing, and can't do anything to change them, and can't
change the RF spectrum that the BMS is trying to use.
That's why providing a separate communication channel is almost universal. It
might be wired, or optical, or RF (not relying on the traction wiring to carry
the signal). You have a far better chance of it working.
Communicating _to_ the BMS is simple. You have two VERY large
transmitters, the charger and the inverter. Simply frequency modulate
the pwm of the inverter and/or the charger and put an FM detector in
each BMS on the cell level. You turn the "noise" source into the
communication transmitter. Done.
That works if you designed the charger and inverter and BMS specifically to
work together to do this.
Communication _from_ the cells is not quite as simple, but doable. Use
the by-pass circuit to talk to the outside world. Put a capacitor in
parallel with the by-pass resistor so that when you switch on the
resistor, you get a spike.
Perhaps; but the batteries themselves still have a huge equivalent capacitance.
The charger and controller are also likely to have huge low-ESR filter
capacitors across them, which try to short out any RF signals present.
It sounds easy; but put a spectrum analyzer on your battery leads to see what's
*really* there. I think you'd be shocked at the noise level.
There are a few clever tricks you can employ. During charging, you can
have the charger pause for a regular "moment of silence" in which the
BMS can communicate quickly and in the clear without having to "shout"
over the charger PWM. Perhaps the same thing could occur to a lesser
extent with the inverter.
Same as above. Are you going to design a special charger that must be used with
your BMS?
There are lots of solutions that work *some* of the time. There are a few that
work *most* of the time. But it gets damnably difficult to find schemes that
work *all* of the time.
The problem is that a BMS is a safety system that you want to work *all* of the
time.
Lee Hart
_______________________________________________
UNSUBSCRIBE: http://www.evdl.org/help/index.html#usub
ARCHIVE: http://www.evdl.org/archive/index.html
INFO: http://lists.evdl.org/listinfo.cgi/ev-evdl.org
Please discuss EV drag racing at NEDRA (http://groups.yahoo.com/group/NEDRA)
_______________________________________________
UNSUBSCRIBE: http://www.evdl.org/help/index.html#usub
ARCHIVE: http://www.evdl.org/archive/index.html
INFO: http://lists.evdl.org/listinfo.cgi/ev-evdl.org
Please discuss EV drag racing at NEDRA (http://groups.yahoo.com/group/NEDRA)
