Home grown BMS ideas !

How do you store and manage your electricity?
Post Reply
antiscab
Senior Member
Posts: 2353
Joined: Mon, 26 Nov 2007, 05:39
Real Name: Matthew Lacey
Location: Perth, WA

Home grown BMS ideas !

Post by antiscab » Tue, 24 Mar 2009, 02:56

Nevilleh wrote:
TS quote 4.2v as the fully charged maximum. This circuit turns fully on at 4.13v with the pot setting I use - the voltage gain is enormous with two darlingtons - so the voltage can't go any higher than that. Others are using a lower voltage for maybe two reasons: their circuit doesn't turn on so quickly, or, they are just being more conservative on the upper voltage limit. If you look at the TS charge graphs, the voltage is limited to 4.2v and held there for 70-odd minutes as the current falls to zero. At the time point where 4.2v is reached and the current starts to drop, the % charge is only 70.
The main reason I dropped it from 4.2v to 4.13v was that I read Tesla Motors have dropped their max charge voltage for 18650 cells from 4.2v to 4.15v "in the interests of increased battery life".
I have noted that if I charge the cells at 3.7V until the current drops to zero, then increasing the voltage to 4.2v causes a current flow (at the 6A limit) for only another 10 minutes or so, so they must be close to fully charged at that voltage.


the difference in recoverable AH, between charging to 3.4v and 4.2v is 2.5%, based on the bench testing i did on a single cell (TS 40AH) a while ago.

from a BMS design perspective, charging only to 3.6vpc with a 4.2v limit reduces complexity somewhat (dont need to shunt full charge current).

you can also use more cells on the same peak voltage rated controller

it also increases charge efficiency, as you wont *have* to balance on every single charge.

as far as the effect on cycle life, im still figuring that one out (ive been charging to 3.6v for more than a year now).
Nevilleh wrote: As far as my car is concerned, I also wonder if I use 4.2v per cell as the max charge voltage, the whole pack of say 46 cells now produces 193.2v instead of the nominal 3.2x46 = 147.2. Sure it's only for a few minutes, but can my 156v rated controller handle that?
I'd be interested to other opinions.


if the logisystem guys are still using 200v parts, then starting at 193.2v is indeed a little dangerous. a little overshoot on the powerstage and BANG.
that takes about 1 sec.

if you let your bms pull the pack voltage down before starting, you should be fine.
Id ask the logisystem guys what the max operating voltage of their controller is.

Matt
Matt
2017 Renault zoe - 25'000km
2007 vectrix - 156'000km
1998 prius - needs Batt
1999 Prius - needs batt
2000 prius - has 200 x headway 38120 cells

User avatar
acmotor
Senior Member
Posts: 3595
Joined: Thu, 26 Apr 2007, 03:30
Real Name: Tuarn
Location: Perth,Australia

Home grown BMS ideas !

Post by acmotor » Tue, 24 Mar 2009, 05:55

Still pondering this recharge voltage myself.

This post raises lithium top voltage vs cycle life issue.
Although generic lithium, I guess even Tesla have tugged the top back a bit. viewtopic.php?t=983

BTW Nevilleh, the circuit at the start of this post turns full on in .05V
The 431 is op-amp output stage.
iMiEV MY12     105,131km in pure Electric and loving it !

Nevilleh
Senior Member
Posts: 773
Joined: Thu, 15 Jan 2009, 18:09
Real Name: Neville Harlick
Location: Tauranga NZ

Home grown BMS ideas !

Post by Nevilleh » Thu, 26 Mar 2009, 23:38

acmotor's clamp turns on in 50 mV, mine turns on in 3 mV. I also tried it with an MJE3055 instead of the TIP142 ( I don't have enough '142s) and it works nearly as well, switching full on in 5 mV.
The question is, how fast does it need to switch? Is the reduced parts count worth the slower turn on?
I tend to think of it as the automatic safety device to prevent the cells from over charging, but the charge voltage seems to be OK anywhere from about 3.6v up to 4.2v, so if you have a clamp that turns on on 50 mV, that's probably quite sufficient and if you have to make hundreds of the things, a few less bits is very desirable.
But acmotor's transistor is only half an amp Ic and I think one should allow for a lot more than that, up to the charger current limit. I am inclined to think that the LM431 driving a pnp Darlington with an Ic of at least 6A would be the best solution. That's what I will try next!
After all, I want to stick it on an LFP160 - eventually.
Last edited by Nevilleh on Thu, 26 Mar 2009, 12:43, edited 1 time in total.

User avatar
Johny
Senior Member
Posts: 3729
Joined: Mon, 23 Jun 2008, 16:26
Real Name: John Wright
Location: Melbourne
Contact:

Home grown BMS ideas !

Post by Johny » Thu, 26 Mar 2009, 23:51

IMO your PNP darlington (or NPN on the anode side) would be fine. Just don't use a super fast one that has a built-in low value resistor like the TIP14? family.

BTW I know I am coming from a 2 way Radio perspective here but I notice that no-one is including RFI suppression. The CBer or Taxi/Police car/Fire truck that pulls up next to you and transmits might have horrible effects on your BMS. Even the odd cell phone might be a disaster.

User avatar
acmotor
Senior Member
Posts: 3595
Joined: Thu, 26 Apr 2007, 03:30
Real Name: Tuarn
Location: Perth,Australia

Home grown BMS ideas !

Post by acmotor » Fri, 27 Mar 2009, 03:09

Johny, true about RFI but at least all my battery boxes are metal so it can only get in via wiring. You could start adding some ceramic caps but take care, it could cause more problems than it fixes. You are welcome to make suggestions !

The hundreds of amps flowing in the battery wiring is more likely to be a problem ( I found it was with pic BMS ).

Nevilleh, I was a bit concerned about too sharp a turn on as it may result in oscillations since the BMS shares the same terminals as the actual battery working load.

Spotted as well....
I ended using KSH127 a surface mount darlington as the shunt transistor.data sheet for 2 reasons. First, the Ic, second the price at digikey. Image I'll update the circuit on the post.
iMiEV MY12     105,131km in pure Electric and loving it !

User avatar
Johny
Senior Member
Posts: 3729
Joined: Mon, 23 Jun 2008, 16:26
Real Name: John Wright
Location: Melbourne
Contact:

Home grown BMS ideas !

Post by Johny » Fri, 27 Mar 2009, 03:29

Yes it's all a nasty environment. Standard electronics in cars is not usually a problem but we have destroyed after-market mosfet car-stereo amplifiers with 25W of RF at 160MHz into properly matched antennas. rural Ambulance is an interesting proving ground.
Since the wiring would be the main coupling point I would suggest a 1nF straight across the battery terminals on each BMS. As you say this isn't a big issue but better safe...

User avatar
acmotor
Senior Member
Posts: 3595
Joined: Thu, 26 Apr 2007, 03:30
Real Name: Tuarn
Location: Perth,Australia

Home grown BMS ideas !

Post by acmotor » Fri, 27 Mar 2009, 03:39

Johny, Good point you raise. It should be included in home grown EV conversion design / testing.
OK, your job is to specify the test proceedure !
iMiEV MY12     105,131km in pure Electric and loving it !

User avatar
weber
Site Admin
Posts: 2566
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Home grown BMS ideas !

Post by weber » Wed, 01 Apr 2009, 06:56

acmotor wrote:Here is my circuit again just to start the topic here.
Image

I can't help thinking that instead of only a single bit of information, we could have an analog quantity represented on that single optocoupled daisy chain, so the motor controller or charger (including regenerating VF drive) could have some warning that things are getting bad and maybe even use the feedback in a linear control loop.

The idea would be to use linear optos and have the opto with the lowest output current calling the tune on the daisy-chain. When every cell is at say 3.3 volts then the daisy-chain current might be at a maximum value, say 20 mA. This might drop by 2 mA for every 0.1 V deviation either above or below 3.3 V. So if, during discharge the lowest cell was at 2.5 V the daisy-chain current would go down to 4 mA. The same would occur if, during charge or regen, the highest cell went to 4.1 V.

Temperature measurement could be included in this. 25^C might be considered optimal and a 30 kelvin temperature change considered to be just as bad as a 1 volt voltage change. Whichever was badder, temperature or voltage, would determine how low the daisy-chain current went.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).

User avatar
acmotor
Senior Member
Posts: 3595
Joined: Thu, 26 Apr 2007, 03:30
Real Name: Tuarn
Location: Perth,Australia

Home grown BMS ideas !

Post by acmotor » Wed, 01 Apr 2009, 18:40

OK so the most worserer of the badders sets the shutdown. I get it Image
Keep thinking on it. There is potential.
Final cost must also be kept in mind.
iMiEV MY12     105,131km in pure Electric and loving it !

User avatar
Johny
Senior Member
Posts: 3729
Joined: Mon, 23 Jun 2008, 16:26
Real Name: John Wright
Location: Melbourne
Contact:

Home grown BMS ideas !

Post by Johny » Wed, 01 Apr 2009, 18:53

The problem is that linear opto-couplers are fickle devices. In order to get good linearity and defined gain stability they usually require a "servo-loop" back to the driving LED. This would be a pain when trying to keep things simple.
Maybe there is something better than what I have used in the past but they left me wishing I had "done it" another way.

User avatar
coulomb
Site Admin
Posts: 3637
Joined: Thu, 22 Jan 2009, 20:32
Real Name: Mike Van Emmerik
Location: Brisbane
Contact:

Home grown BMS ideas !

Post by coulomb » Wed, 01 Apr 2009, 19:03

As a point of interest, are the FET output opto isolators any more linear? They would presumably also have less voltage drop (when conducting fully), important when a large number of optos will be in series.

Also, I'm not sure that linearity is necessary, just monotonicity. In other words, when things go from badder to badderest, it's important that the resistance increases (assuming low resistance = least badness), not that 37% extra badness will result in 37% +/- 5% increased resistance. (Weber: no, you won't ever live that one down Image   )

The overall feedback loop should still settle down to keep all the batteries fairly happy (small badness), while the vehicle has maximum performance.
Nissan Leaf 2012 with new battery May 2019.
5650 W solar, PIP-4048MS inverter, 16 kWh battery.
1.4 kW solar with 1.2 kW Latronics inverter and FIT.
160 W solar, 2.5 kWh 24 V battery for lights.
Patching PIP-4048/5048 inverter-chargers.

User avatar
coulomb
Site Admin
Posts: 3637
Joined: Thu, 22 Jan 2009, 20:32
Real Name: Mike Van Emmerik
Location: Brisbane
Contact:

Home grown BMS ideas !

Post by coulomb » Wed, 01 Apr 2009, 19:14

acmotor wrote: Final cost must also be kept in mind.

Yea, verily.

I'm thinking that diodes rather than low voltage comparators or op-amps could achieve the required "analogue logic". And still be cheap. Perhaps the analogularity (TM) could extend to the red LED: a dull glow from the LED means that this cell is the one causing the cutback of regen or output, but is only a little way from its happy zone (either temperature, over, or under voltage). A bright red LED indicates "I'm really bad, help me!".

It would be good to have a fair range of voltage where there was zero badness reported, e.g. 2.7 to 4.0 volts should be zero badness, with moderately steep ramps at both ends. Same for temperature (probably single ended, so extreme cold is not considered bad). Otherwise, as soon as you drive off, you will start tapering off performance. I believe that the programmable zeners used in acmotor's design can be easily configured to do this.

Edit: The point of this is of course to prevent the controller or charger from continually "hunting" about the badness threshold. Suppose one cell is getting hot; let's say it's at 140°C and we want to keep the batteries under 150°C at all times. With a digital output, you have to choose a safe threshold, say 140°C. The controller doesn't know how much to cut back, so we cut back 10%. The cell rises to 145°C, and the output is still bad. So the controller cuts back 20%, eventually the cell falls to 140°C but by then the controller has cut back 50%. Now the controller goes back to 100% output, and the cycle repeats. In this case, it's a slow cycle, because the time constant for temperature is long. For regen, you might get a bumpy ride.

In the analogue case, we have a ramp that goes from minimum to maximum badness over the 140°C to 150°C range, say. As soon as the cell gets just a but hot, the controller backs off a bit, say 5%. It might settle there, at 142°C and 5% less power. This is a lot better than cutting in and out with wild overshoots.

I guess I'm not being fair to the digital case here; if in the digital case we cut back in 5% increments, the same outcome might be achieved, though it still has to hunt. But what if a cell is rocketing up in temperature? In the digital case, the best you can do is cut back 5%, wait 10 seconds, and cut back another 5% if the badness remains. Meanwhile, the cell could have burned. To prevent that, you have to cut back quicker, at the expense of possible wild overshooting.

In the analogue case, if the temperature shoots up, the output goes quickly to maximum badness, and the controller can shut down quickly. So I'm saying that the analogue output could provide much smoother operation, and/or protect the battery better.
Last edited by coulomb on Wed, 01 Apr 2009, 08:31, edited 1 time in total.
Nissan Leaf 2012 with new battery May 2019.
5650 W solar, PIP-4048MS inverter, 16 kWh battery.
1.4 kW solar with 1.2 kW Latronics inverter and FIT.
160 W solar, 2.5 kWh 24 V battery for lights.
Patching PIP-4048/5048 inverter-chargers.

User avatar
acmotor
Senior Member
Posts: 3595
Joined: Thu, 26 Apr 2007, 03:30
Real Name: Tuarn
Location: Perth,Australia

Home grown BMS ideas !

Post by acmotor » Wed, 01 Apr 2009, 20:03

We do need to keep in mind what we are actually trying to achieve with the BMS, otherwise we will all have to buy BMI from LBB ! and leave your phone on so it can send you an SMS ! Image

You must stop or (at least reduce) charge or discharge when you get O/V or U/V. You must provide some charge eq.
After that, you are just playing with data and if temperature for instance is an issue then you are most likely pushing the cells way beyond their design rating. IMHO

Understood about digital vs analogue but you know the old saying, there is no such thing as a little bit ........ !

But don't let me put you off. If you can collect more data and still keep it simple, low cost and reliable then got for it. Image
iMiEV MY12     105,131km in pure Electric and loving it !

User avatar
weber
Site Admin
Posts: 2566
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Home grown BMS ideas !

Post by weber » Wed, 01 Apr 2009, 20:16

Ah'm back and ah'm bad(-er). Image
coulomb wrote: As a point of interest, are the FET output opto isolators any more linear? They would presumably also have less voltage drop (when conducting fully), important when a large number of optos will be in series.
FET output is resistance output, which merely adds when wired in series, rather than giving the required MIN or MAX function. Bipolar transistor outputs and photodiode outputs are current outputs and when wired in series these give an (approximate) MIN function. But the two have to be sensed in different ways.

The bipolar transistors require you to put about n * 0.3 V across the string and measure the current. So for 200 cells that's 60 V. If one opto wants to pull the current down to zero it has to take the full 60 volts across itself. That's not a problem.

The photodiodes (essentially tiny photovoltaic cells) don't need an external applied voltage, they produce their own approx 0.3 volts each. But they produce a lower current than the transistors, and if one of them wants to reduce the current to zero, it has to take the full 60 volts generated by all the others, across itself _in_reverse_. This aint gonna happen. They can typically only take 5 V reverse. So can't wire more than about 16 of them in series. But that might be OK. 16 LiFePO4's is a 48 V module.
Also, I'm not sure that linearity is necessary, just monotonicity.
I agree, but as Johny mentioned, what is important is that they are all about the same and that they don't change with temperature etc. Typical transistor gains, LED efficiencies, PV efficiencies taken together, vary +-100%. A standard "linear opto" deals with this by having two identical photodiodes. One is your output. The other is your feedback, which you then need an op-amp to deal with.

And you're stuck with a photodiode output in that case, because if you amplify its current with a transistor you then have the highly variable transistor gain unaccounted for.

What did you have in mind re diodes?
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).

User avatar
weber
Site Admin
Posts: 2566
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Home grown BMS ideas !

Post by weber » Wed, 01 Apr 2009, 20:43

acmotor wrote:If you can collect more data and still keep it simple, low cost and reliable then got for it. Image

Your circuit is a masterpiece of minimalism. The best I can do to reduce it further is save maybe 2 resistors. One by taking the bypass and overvoltage sensing off a single 3-resistor divider. The other (maybe) by putting the LED where the transistor's base resistor should be.

In theory you could also get rid of the two power resistors and heatsink the transistor instead, e.g. heatsinking it to the battery terminals with fat copper strips as in Rod Dilkes BMS. But I'm sure the resistors are cheaper and simpler and they can run at a higher temperature and so radiate more. It's a shame you can't buy 25 or 50 watt 4 volt lamps. They would be even better than resistors and you wouldn't need the LED. With a clear cover on the batteries and a black painted underside of boot or bonnet above that, they would get the heat right out of the battery box without needing a fan.

I just want to see if we can make some kind of minimal changes such as replacing some precision zeners with a quad op-amp and using a different kind of opto, to get a little more info out on that single wire, to avoid the overshoots by knowing in advance when the badness is coming.

Your bypass circuit could stay as it is now, however if temperature sensing was included it might also be used to temperature-compensate the bypass voltage. Fixed-bypass-voltage boards might overvoltage the cells if the vehicle was sold and moved to a hotter climate. And even in one place, summer winter variation may be significant.

The apparently high voltages specified by Thunder Sky might be due to low ambient temperatures in their part of China.
Last edited by weber on Wed, 01 Apr 2009, 09:46, edited 1 time in total.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).

User avatar
Johny
Senior Member
Posts: 3729
Joined: Mon, 23 Jun 2008, 16:26
Real Name: John Wright
Location: Melbourne
Contact:

Home grown BMS ideas !

Post by Johny » Wed, 01 Apr 2009, 20:50

If anyone wants to get a lot more complex this might be of interest.
A single chip that monitors up to 12 Lithium cells. US$9.95 in 1000 lots.
Can also be used to drive FETs for current sink.
LTC6802 Info

User avatar
acmotor
Senior Member
Posts: 3595
Joined: Thu, 26 Apr 2007, 03:30
Real Name: Tuarn
Location: Perth,Australia

Home grown BMS ideas !

Post by acmotor » Wed, 01 Apr 2009, 21:08

Nice but not KISS.

It starts off OK but look at page 36.

It is definitely a sign of good things to come though !

edit: added link to data sheet.
LTC6802 data sheet
Last edited by acmotor on Wed, 01 Apr 2009, 10:51, edited 1 time in total.
iMiEV MY12     105,131km in pure Electric and loving it !

User avatar
Johny
Senior Member
Posts: 3729
Joined: Mon, 23 Jun 2008, 16:26
Real Name: John Wright
Location: Melbourne
Contact:

Home grown BMS ideas !

Post by Johny » Wed, 01 Apr 2009, 21:14

I did say a LOT more complex. Image

User avatar
acmotor
Senior Member
Posts: 3595
Joined: Thu, 26 Apr 2007, 03:30
Real Name: Tuarn
Location: Perth,Australia

Home grown BMS ideas !

Post by acmotor » Wed, 01 Apr 2009, 21:49

Maybe the new UWA Lotus EV could go this 'lots a data' direction ? Image
iMiEV MY12     105,131km in pure Electric and loving it !

User avatar
weber
Site Admin
Posts: 2566
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Home grown BMS ideas !

Post by weber » Wed, 01 Apr 2009, 22:30

Here's another idea, in the spirit of brainstorming.

Instead of analog current we use PWM current. We'd need optos going in the opposite direction (two optos per cell) to synchronise. But we'd still use only the single-wire loop.

The LEDs of the input optos could be in parallel with the collector-emitter of the output optos.

Start with all output optos off (o/c) and n * 0.3 V across the loop. Then the master unit puts a big enough voltage pulse (but very narrow) across the loop, to pulse all the input LEDS. This would take something like 1.5 V per cell, so maybe 300 V, but current and energy limited. Done using an inductor flyback circuit. After the n * 1.5 V pulse the master drops the voltage back to n * 0.3 V and watches the current.

All the cell boards respond to this by immediately turning on their outputs (s/c). Then they wait a time proportional to their goodness before turning off again.

goodness = 1 - badness
goodness and badness have values between 0 and 1 (or 0% and 100%)

So the baddest one determines when the whole loop goes o/c. The master waits a fixed time between n*1.5 V pulses. If all cells are 100% good (0% bad) then their output optos all go o/c just before the master sends the big pulse again. If the loop hasn't gone o/c by that time, the master does not send the pulse but registers a fault. Or it might be smart enough to wait just a bit longer and alter its cycle time accordingly.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).

Nevilleh
Senior Member
Posts: 773
Joined: Thu, 15 Jan 2009, 18:09
Real Name: Neville Harlick
Location: Tauranga NZ

Home grown BMS ideas !

Post by Nevilleh » Fri, 03 Apr 2009, 16:33

Lots of interesting ideas here!
But I am going with a simple clamp to stop any cell going too high and using fixed resistors in my original circuit sets the activation volyage at 3.93.
In addition, I use an Atmel ATTiny25 to monitor the voltage of each cell. The Tiny 25 has a 4 channel, 10 bit ADC, so one micro does 4 cells. It also has a built in temperature sensor and I plan to use that as the pcb will be attached to the top of one cell in each block of 4.
The Tiny25 is about $2.50 - 3.00 depending on where you get it from.
Using it's SPI capability lets me run a single line, multi-drop comms setup, at the moment using optocouplers ($1.00 or so each) but I might try just capacitive coupling instead - save a few dollars.
The system uses a single byte for addresses. Each micro is programmed as $FF initially. The first one to be turned on talks to the master which allocates it an address in sequence starting at $01 ($00 is reserved for the master). Thus they all receive their own unique address simply by turning them on one at a time and putting a jumper on the pcb is the easiest way to do that.
The master will then just poll them all in turn and receive the 4 cell voltages associated with that particular micro, plus it's temperature.
The comms is async and around 9600 baud is more than adequate. That is about a millisecond per byte (10 bits with start and stop bits) and each data packet will contain an address, 4 cell voltages in 5 bytes, temperature and a CRC byte ie 8 bytes, so 8 msecs per device or 125 per second. I think that is fast enough!
Being a polled system, there are no data collisions and the CRC will provide enough protection against external corruption, although 9600 bps is fairly incorruptible.
Anyone else who wants to try this, drop me a note and I'll give you the software - when its finished!
The master software will collect and store the cell voltages and raise an alarm if any one (or more) goes above or below whatever limits I put in. This is actually the biggest part of the job as I want to use my laptop as the master and my VB skills are very rusty. Any experts out there want to help out?
We could make it an open source thing if enough people are interested.

I tried to upload the circuit diagram, but the image is too big. I'll have to find another way.
Last edited by Nevilleh on Fri, 03 Apr 2009, 05:42, edited 1 time in total.

User avatar
coulomb
Site Admin
Posts: 3637
Joined: Thu, 22 Jan 2009, 20:32
Real Name: Mike Van Emmerik
Location: Brisbane
Contact:

Home grown BMS ideas !

Post by coulomb » Fri, 03 Apr 2009, 18:19

weber wrote: What did you have in mind re diodes?

I've already discussed this face to face with Weber, so this is for others.

The idea is that internally you have a voltage representing temperature badness, say 1-3 V, and another representing voltage badness (also 1-3 V). Maybe two voltage badnesses, one for under, one for over.

Then you just have 2 or 3 diodes that have common cathodes, anodes connected to the internal badness generators. Whichever voltage is the highest gets its diode to conduct (the others will be reverse biased). So the common cathodes have the highest badness, less one diode drop. This is used to generate the badness current.

In practice, we want 0 badness to be max current (so a break in the daisy chain is an error, not "all OK"), so we would invert all this. Each badness signal would be 3-1 V (3V = no badness, 1V = max badness), and common the anodes. Now the signal with the most badness has the lowest voltage, and will conduct its diode, so the output will be the same as the baddest signal, plus one diode drop. A resistor turns this voltage into a current into the opto's LED.
Nissan Leaf 2012 with new battery May 2019.
5650 W solar, PIP-4048MS inverter, 16 kWh battery.
1.4 kW solar with 1.2 kW Latronics inverter and FIT.
160 W solar, 2.5 kWh 24 V battery for lights.
Patching PIP-4048/5048 inverter-chargers.

Tritium_James
Senior Member
Posts: 683
Joined: Wed, 04 Mar 2009, 17:15
Real Name: James Kennedy
Contact:

Home grown BMS ideas !

Post by Tritium_James » Fri, 03 Apr 2009, 19:05

I can almost guarantee you a capacitive coupling scheme for the comms won't work - I've tried it in the BMS for the Porsche and it runs fine until you have either the battery charger running or the motor running. In the charger case it moves the whole pack around at 1/2 the mains voltage and 100Hz (this will be ACP system specific) and when it's motoring there's fast transients from the PWM switching. The capacitive comms stopped working fairly rapidly. We were doing it properly too, with manchester encoded comms to keep a constant DC value on the caps, idle time filled with 50% transitions, shielded twisted pair cable, all that sort of thing.

User avatar
Johny
Senior Member
Posts: 3729
Joined: Mon, 23 Jun 2008, 16:26
Real Name: John Wright
Location: Melbourne
Contact:

Home grown BMS ideas !

Post by Johny » Fri, 03 Apr 2009, 19:09

In addition you would have to provide ground-supply diode clamping which could well introduce RFI issues (diodes acting as detectors). Reasonable speed optos are very cheap when bought in quantity - even 100 or so.

User avatar
coulomb
Site Admin
Posts: 3637
Joined: Thu, 22 Jan 2009, 20:32
Real Name: Mike Van Emmerik
Location: Brisbane
Contact:

Home grown BMS ideas !

Post by coulomb » Fri, 03 Apr 2009, 19:15

weber wrote: Here's another idea, in the spirit of brainstorming.

Instead of analog current we use PWM current. We'd need optos going in the opposite direction (two optos per cell) to synchronise. But we'd still use only the single-wire loop.

The LEDs of the input optos could be in parallel with the collector-emitter of the output optos.

I like this idea much better than attempting to use optos in linear mode. The characteristics of optos vary so much. For example, current transfer might be minimum 20% (so 10mA diode -> 2mA transistor), but is typically 70%, and there is no maximum value. Similarly, forward voltage drop is typically 1.15 (for a 4N25), but has a max of 1.5v. It is also temperature dependent, which could be used to advantage, but I suspect it will vary too much from device to device. I think I'd really prefer not to have to adjust 200+ BMS boards.

So the PWM idea is good; it uses optos in digital mode (where they work best), yet we get analogue information.

Another consideration: we'll likely need a quad CMOS op-amp device to generate the various "ramps" we require. A typical device (e.g. TS914) might require a minimum rail voltage of 2.7v, yet we expect the circuit not to go nuts with the cell at 2.5v or (hopefully rarely) less. There are low voltage op amps (e.g. 1.8v), but these may be expensive and possibly hard to find (I haven't looked).

A single programmable zener device could suffice; the 1.4 V reference could be used at one input of all 4 op-amps. The four op-amps would presumably handle low volts, high volts, high temperature, and bypass. That's assuming we don't need an op-amp to handle the timing. An LMC555 works down to 1.5 V, so that's a likely contender.
Last edited by coulomb on Fri, 03 Apr 2009, 08:16, edited 1 time in total.
Nissan Leaf 2012 with new battery May 2019.
5650 W solar, PIP-4048MS inverter, 16 kWh battery.
1.4 kW solar with 1.2 kW Latronics inverter and FIT.
160 W solar, 2.5 kWh 24 V battery for lights.
Patching PIP-4048/5048 inverter-chargers.

Post Reply