Home grown BMS ideas !

How do you store and manage your electricity?
User avatar
weber
Site Admin
Posts: 2807
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Home grown BMS ideas !

Post by weber »

Tritium_James wrote: For daisy chained comms, I'd suggest driving out from each node non-isolated, and have the opto as the receiver on each node. That way you've got a nice low-impedance driver for the cable, with less chance of picking up noise.
There is that. But I was worried about that last cell board where it talks to an opto on the master. It could be at 800 V relative to the chassis (even though it's a floating supply, there's capacitance). Touch it and you'll blow up the micro as well as zapping yourself.
If you are paranoid (and who isn't, with a BMS) you could even drive a pair differentially, either use a RS485 driver, or just two outputs on your micro. So you'd be driving the opto with +1.8V or -1.8V, that way you guarantee that the opto is off when you want it off.

That's a great idea, only we'd be using +-2.5 V for reasons given above.

And clever idea about diagnosing breaks in the chain. Thanks.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).
User avatar
Johny
Senior Member
Posts: 3733
Joined: Mon, 23 Jun 2008, 16:26
Real Name: John Wright
Location: Melbourne
Contact:

Home grown BMS ideas !

Post by Johny »

weber wrote:... But I was worried about that last cell board where it talks to an opto on the master. It could be at 800 V relative to the chassis (even though it's a floating supply, there's capacitance). Touch it and you'll blow up the micro as well as zapping yourself.
Have an end-of-line daughter board you attach to the last BMS.
User avatar
weber
Site Admin
Posts: 2807
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Home grown BMS ideas !

Post by weber »

So how can we have low impedance drive without having lethal comms wires?

Of course what we really want is to put half of each opto on the two communicating boards and have only light travelling between them. But any opto communications I know of is way too expensive for one-per-cell, except a naked IR LED aimed at a naked phototransistor. Sorry if this is getting needlessly erotic Neville. Image

I suppose we could afford a few optic fibres, say between sub-packs and to and from the master.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).
User avatar
weber
Site Admin
Posts: 2807
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Home grown BMS ideas !

Post by weber »

Johny wrote:Have an end-of-line daughter board you attach to the last BMS.

OK. Or even just a part of the cell board that isn't usually populated. It would have to be fed 12 V from the accessories supply, but that's OK.

But I'm still nervous. I was taught that you never take a bare micro pin off-board. The protection diodes can only take a milliamp or so. So I suppose I have to have some series resistance. Why not make it the resistance needed to limit the LED current, and have no such resistors on the board that has the LED? Does this defeat the purpose of having low impedance drive? I doubt it. Not if it's differential drive.
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 »

weber wrote:
Nevilleh wrote:It would require a failure that turned on the Tx opto permanently. That's fairly unlikely.
Or a fault that continually transmitted gibberish. But yes, probably unlikely. Most likely failures are probably in comms wiring.

So you have only 2 comms connection pads on each PCB? We'd need 4.

When you ask how the cascaded thingy would work, what level of description are you looking for? Hardware, bits, bytes, packets? Do you see why it has to be only one opto per board with that scheme now? Opto on output, resistors on input.

We should definitely look more closely at the AVR Tiny25 and consider saving that 50c per board. Is there a free assembler development system for it?

Your packet scheme seems very sensible. Can you upload a Protel file for your master and slaves? Or point us to schematics elsewhere in this huge thread?


Yes, I really do know how the cascade thing works, I thought you might might have come up with something else, but your "resistors on input" doe NOT provide 2.5 kV isolation - in fact, none at all.

The Tiny25 is the "most bang for the buck" processor that I have been able to find. You can download a free assembler and a free C compiler (WinAVR) from Atmel's web site, plus heaps of app notes and other stuff.
The ISP feature makes it a piece of cake to get going and Atmels emulator even lets you single step your code. I've never needed anything else to make it work. Edit: PonyProg supply a good, free programmer too.
If you order the "V" of the chip, it works from 1.8V to 5.5V and you can get a 20 MHz version for the same price. Doing away with voltage regulators saves money and all you need do is filter the supply properly, like a decoupling resistor or even a small choke and an electro and a ceramic. Mind you, making the comms work at 1.8V might be a bit of a mission.

The data packet scheme I invented way back in '97 to handle comms with poker machines - it included encryption then - and it is now used in about 60% of the poker machines in NSW under the aegis of "eBet" who bought my company. So it is a pretty robust thing.

I'll dig out the schematics for the boards in the next day or so and stick them up here.

edit: Guys, I thought by now you would realise that daisy chained comms is for the birds! Let's not waste any more time and effort discussing it, if it was any good the Internet would use it Image

What does the Internet use? Something rather like the packet structure I have described - OK, a tad more complex, but pretty similar. And we all know how well that works! If I buy any more Viagra, I can start a pharmacy
Image
Last edited by Nevilleh on Wed, 09 Dec 2009, 10:41, edited 1 time in total.
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 »

weber wrote: S Sorry if this is getting needlessly erotic Neville. Image


Needlessly?

Image
User avatar
weber
Site Admin
Posts: 2807
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Home grown BMS ideas !

Post by weber »

Nevilleh wrote:Yes, I really do know how the cascade thing works, I thought you might might have come up with something else, but your "resistors on input" doe NOT provide 2.5 kV isolation - in fact, none at all.
Perhaps isolation is the wrong word, and I didn't mention 2.5 kV, only 800 V. But large enough resistors with a suitable voltage rating certainly can limit the current to non-lethal values (for you and the micro). This is what acmotor does (and coulomb and I copied it) to monitor cell voltages from the analog boards. You can see the 300k 1W 1kV-withstand "isolation" resistors in the photo above.
Doing away with voltage regulators saves money and all you need do is filter the supply properly, like a decoupling resistor or even a small choke and an electro and a ceramic.
Yes. That aspect of the AVR is of interest too. Good point.
Guys, I thought by now you would realise that daisy chained comms is for the birds! Let's not waste any more time and effort discussing it, if it was any good the Internet would use it Image
The internet does use it, in a sense. What do you think bridges gateways and routers do? Mind you, a message from the other side of the planet will pass through far fewer than 200 routers on its way to you. More like 10 to 20. When you get too many things on a bus, you start a new bus and bridge between them. But we'd like to avoid the complexity of having a two-level structure here.

There's no reason why your packet protocol can't be used with route-thru comms.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).
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 »

weber wrote: So I suppose I have to have some series resistance. Why not make it the resistance needed to limit the LED current, and have no such resistors on the board that has the LED? Does this defeat the purpose of having low impedance drive? I doubt it. Not if it's differential drive.


This is what I'd be doing if I was building what you're talking about. Split the resistance required for the opto into two and put half on each output pin. You'll then have turned the output into a differential current source, it's going to be very immune to noise pickup.

Though Nevilleh is right, driving an opto reliably from 1.8V is going to be problematic I think.

We have done a BMS with this type of comms, but we capacitively coupled the boards together, using manchester encoding on the comms to keep the DC levels on the caps right. It worked fantastically (and cheap, too!) on the bench, and failed as soon as we stuck it in the Porsche, as the ACP charging scheme slams the battery pack around at 100Hz squarewave relative to the chassis and screws with the capacitive isolation. Bummer.

The daisy chained approach sucks for debugging, but is nice from the point of view that you only have short links for your comms, and they're point-to-point, no branching, which can help hugely with signal integrity.

You can also do nice things like if a node needs to signal RIGHT NOW that it's got problems, it can stop passing on any messages and inject it's own message into the loop, rather than waiting for a free slot on a common bus. But in reality this probably isn't necessary, if something has to wait a few ms for a free comms slot it's probably fast enough.
User avatar
weber
Site Admin
Posts: 2807
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Home grown BMS ideas !

Post by weber »

Tritium_James wrote:This is what I'd be doing if I was building what you're talking about. Split the resistance required for the opto into two and put half on each output pin. You'll then have turned the output into a differential current source, it's going to be very immune to noise pickup.
Thanks for the confirmation.
Though Nevilleh is right, driving an opto reliably from 1.8V is going to be problematic I think.
Why do we need to drive it reliably from 1.8 V? We plan to use a 2.5 V regulator (or we might go to an unregulated AVR). However, when regulation has dropped out, it does need to work "just-barely" down to 1.8 V. I thought, since an opto LED is IR it only drops about 0.9 V at 1 mA?

Does this have any bearing on the route-thru vs. bus decision?
We have done a BMS with this type of comms, but we capacitively coupled the boards together, using manchester encoding on the comms to keep the DC levels on the caps right. It worked fantastically (and cheap, too!) on the bench, and failed as soon as we stuck it in the Porsche, as the ACP charging scheme slams the battery pack around at 100Hz squarewave relative to the chassis and screws with the capacitive isolation. Bummer.
I'm pleased to say that coulomb and I discussed the capacitive/manchester option on the weekend, and rejected it for similar reasons.

But glad to know you've had route-thru comms working well. How many did you have in the chain?

[Edit: Looked it up. Opto LED more like 1.2 V @ 1 mA when cold.]
Last edited by weber on Wed, 09 Dec 2009, 12:48, edited 1 time in total.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).
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 »

The nice thing with daisy-chained comms is it's cheap because you only need one opto per board. Bus comms need one for tx and one for rx.

We had it running with 28 nodes. We had a nice scheme worked out where the (isolated, logic level side) Rx connected through to the (logic level side) Tx on each board, so that messages going around the bus from the master would travel around the loop almost instantly via each board, without the micro getting involved. If the micro wanted to transmit, it could override the direct connection.

Having said all that, it might be cheap, but daisy-chained comms sure was painful for development and debug. Our BMS now uses a bus topology, CAN bus no less. Adds to the cost, but development was quick!
domEV
Noobie
Posts: 4
Joined: Mon, 07 Dec 2009, 00:36
Location: CRO
Contact:

Home grown BMS ideas !

Post by domEV »

Friend of mine (he is also after another conversion ICE to AC)and me discussed few days ago which micro to use for single battery BMSes. In the end we narrowed it down to ATtiny25 and PIC16F688-E/P ("a bit" overkill for such a BMS, but smallest and available PIC that has USART).

Regarding ATtiny there is also set of powerfull tools available at http://www.oshonsoft.com/avr.html. Altough they're not free, licenses seems to be fairly priced.

Btw, when you talk about isolation, what do you say about idea (that just crossed my mind) to go for e.g. http://www.neuhold-elektronik.at/datenblatt/E9245.pdf (for English scroll down)

Or even better, almost all-in-one low price solution like: http://ww1.microchip.com/downloads/en/D ... 70091a.pdf Image
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 »

Do it by wireless might be nice indeed!

But here's the circuits and photo of the prototype for my 8-cell BMS. I haven't included the master unit but there was a picture of on here somewhere. Edit: can someone tell me how to refer to an image I have already uploaded - in a new post, that is?

The only change to the pcb from the prototype was to swap the IDC connector for a Molex one with a bit more current capability, but I haven't made that board yet.

Been too busy getting my car finished and writing the software for both the cell monitor board and the master. My C skills have taken a bit of refreshing!

ImageImage
This is the home-made, hand assembled prototype. The 3 holes at the top centre are for the comms connector (not fitted in this photo)
Also, the 2 holes below the micro are for a 2-pin jumper that connects the power to the micro - used to switch it on for the auto-install.
Image
Last edited by Nevilleh on Thu, 10 Dec 2009, 05:43, edited 1 time in total.
User avatar
weber
Site Admin
Posts: 2807
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Home grown BMS ideas !

Post by weber »

domEV wrote: Friend of mine (he is also after another conversion ICE to AC)and me discussed few days ago which micro to use for single battery BMSes. In the end we narrowed it down to ATtiny25 and PIC16F688-E/P ("a bit" overkill for such a BMS, but smallest and available PIC that has USART).
Thanks domEV. Yes it seems if you want a cheap micro with ADC you can't also have a UART or USART. The best you can get is a so-called "Universal Serial Interface (USI)", which non-marketing types call a shift register.

Or if you want a cheap micro with a UART the best you can do is a built-in comparator. At least they don't try to call that a "Universal Analog to Digital Converter (UADC)". Yet.

So given the choice between a slow inaccurate single-channel ADC by timing an RC with your comparator, and a software UART, most folks understandably opt for the software UART.
Btw, when you talk about isolation, what do you say about idea (that just crossed my mind) to go for e.g. http://www.neuhold-elektronik.at/datenblatt/E9245.pdf (for English scroll down)

Or even better, almost all-in-one low price solution like: http://ww1.microchip.com/downloads/en/D ... 70091a.pdf Image

Wireless is a very attractive idea. Not having to wire up those 200 cell boards. Yum. This was suggested to me last weekend and I looked into it then. I didn't discover rfPICs, which are now the best option I've seen. Thanks. But once you add a crystal, and some Ls and Cs on the RF side, that price creeps up.

It's pretty hard to beat a 20 cent optocoupler and there's no danger of interference from another passing EV.
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 »

I did see an AN on the Atmel site describing how to implement a UART with the USI. I originally looked at using the Tiny25 for a 4-cell board but decided the 8-cell one with a mega48 would be better and it has TWO UARTS!

Its AVR307 if you're interested. Probably better than bit-bashing.

Come to think of it, you could do 2 cells with a Tiny25. Use 2 x adc i/ps for voltage measurement, 2 o/p pins for the shunts, on-chip temp sensor so mount the board on top of one of the cells, a couple of optos, 2 power transistors to drive the shunts, 2 power resistors and a few r's and c's and you're there. The other 2 i/o pins on the chip would be DataIn and DataOut. I reckon that would have to be pretty cheap on a per cell basis.
Last edited by Nevilleh on Thu, 10 Dec 2009, 08:44, edited 1 time in total.
User avatar
Johny
Senior Member
Posts: 3733
Joined: Mon, 23 Jun 2008, 16:26
Real Name: John Wright
Location: Melbourne
Contact:

Home grown BMS ideas !

Post by Johny »

Nevilleh wrote:.... can someone tell me how to refer to an image I have already uploaded - in a new post, that is?
I go to the Members Control Panel (in a new window) then File Manager. Click on the picture filename then right click on the filename on the right next to "Download File". Select Copy Link Location and use that to paste into your post as an image link.
Nevilleh wrote:This is the home-made, hand assembled prototype.
Have you fired it up yet Neville?

Edit: That's "Insert Image", not "Image Upload" in the full reply editor.
Last edited by Johny on Thu, 10 Dec 2009, 07:53, edited 1 time in total.
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 »

Thanks for the info on images.
Yes, I have fired up that board. I have been programming it and talking to the master as well.
I made a resistor voltage divider chain to provide the cell voltages and ran the whole thing off my bench power supply at about 28V so it isn't actually connected to batteries, but it measures voltages and sends them to the master. Needs a bit more work on the s/w yet, but I sidetracked myself into the master s/w trying to do amp-hours and stuff. That meant I had to start using long integers and that's taken a wee while to figure out the arithmetic.

Still to do the addressing stuff (I'm just using a fixed address - 01 - at present) and a few other things, but the comms works.

"There are just 10 types of people the world: those who understand binary numbers and those who don't"
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 »

I thought some more about using a Tiny25 to do 2 cells and came up with this quick sketch. It would monitor voltage, temp (of itself, so screw it to a cell) and allow shunt balancing. Isolated comms to the master.
What do you reckon?
I had to add a voltage regulator to keep the volts down, but that allows the 2.56v internal ref for the adc to be used.
Image
User avatar
weber
Site Admin
Posts: 2807
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Home grown BMS ideas !

Post by weber »

Oh wait! I didn't fully appreciate what dmEV was saying. The PIC16F688 _does_ have both 10 bit ADC and UART, and its cost is intermediate between the AT-tiny25V and the MSP430F2012, which only have a USI (and 10-bit ADC).

The PIC16F688 doesn't have a buit-in temperature sensor like the other two, so we have to add 13 cents for a thermistor and resistor, but that only takes it to AU$1.67 in 100 qty (Mouser). MSP430F2012 is AU$1.82, AT-tiny25V is AU$1.32.

Another feature we want badly is the ability to update the software on all cell boards at once, using the same serial comms we use for voltage and temperature reporting. That is, we need the micro to be able to self-program its flash. It appears that the PIC is the only one that can't do that.

The PIC only works down to 2.0 V. The others go to 1.8 V. Can we live with that?

The MSP430 can only work up to 3.6 V while the others can go to 5.5 V and so may get away without a regulator.

Any other pros and cons I've forgotten?
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).
User avatar
weber
Site Admin
Posts: 2807
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Home grown BMS ideas !

Post by weber »

Doing two cells with the one micro has some merit. But we're comitted to having the boards on the tops of the cells. We don't want lethal spaghetti running all over the place. We'd still have to have one board per cell to fit all the bypass resistors. The only question is whether it's worth leaving the micro off every second one and running power, voltage-sense and bypass-control wires (3) between them, instead of comms wires (2). And only sensing the temperature of every second cell.

My feeling is that there are already too many wires. And the special-casing of odd-numbered sub-packs, and having two different kinds of board, would be a nuisance.

I've been thinking about Johny's suggested special case on the last slave, to address my isolation concerns with TJ's opto-on-the-input scheme.

Here's a sketch:

Image

I'm having trouble finding the cheapest and most compact way of doing the differential driver shown at the bottom right. RS422 drivers seem to always come in quads, or packaged with receivers.

This extra opto and differential driver need to fit on the cell board, but only be populated on the last one in each sub-pack.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).
User avatar
weber
Site Admin
Posts: 2807
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Home grown BMS ideas !

Post by weber »

Edit: Deleted duplicated post. The first time gave me a timeout error so I assumed it had not gone through.
Last edited by weber on Thu, 10 Dec 2009, 11:01, edited 1 time in total.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).
User avatar
weber
Site Admin
Posts: 2807
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Home grown BMS ideas !

Post by weber »

Here's a side-by-side comparison of the three contenders.
                 MSP430F2012  AT-tiny25V   PIC16F688
Price 100 qty    $1.82        $1.32        $1.54      (AUD)
Min supply       1.8 V        1.8 V        2.0 V
Max supply       3.6 V        5.5 V        5.5 V
UART             No           No           Yes
USI start detect No           Yes          n/a
Temp sens        Yes          Yes          No
Self-program     Yes          Yes          No
10 bit ADC pins  8            4            8
I/O pins         10           6 (3 prog)   12 (2 prog)
Prog pins        2            4            2
Flash            2 kB         2 kB         4kW (14 bit)
RAM              128 B        128 B        256 B
EEPROM           0 B          128 B        256 B
Off current      0.1 uA       0.1 uA       0.001 uA
Active current   350 uA       480 uA       200 uA      (@ 1 MHz 3.3 V)
Packages         16-QFN       20-QFN       14-TSSOP
                 14-TSSOP     8-SOIC       14-SOIC
                 14-DIP       8-DIP        14-DIP
I note that only the QFN packages have the big pads that give good thermal contact to let us get a good cell temp measurement without an external thermistor and resistor.

[Edit: Corrected 2 omissions]
Last edited by weber on Thu, 10 Dec 2009, 12:03, edited 1 time in total.
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 »

My thinking is that you probably don't need to monitor the temperature of every cell, every other one would be pretty good. They don't change temp all that fast. Hence 2 cells per board. If the one you are monitoring gets up to - I dunno, say 45 deg to allow for the thermal resistance and lag, then you get worried as the max cell temp is only 55 deg.
You make a board that bolts onto one cell with a 2-pin connector going off to the next one and that's it.
Certainly halving the number of boards and bits is worth considering.
User avatar
weber
Site Admin
Posts: 2807
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Home grown BMS ideas !

Post by weber »

Nevilleh wrote: My thinking is that you probably don't need to monitor the temperature of every cell, every other one would be pretty good. They don't change temp all that fast. Hence 2 cells per board. If the one you are monitoring gets up to - I dunno, say 45 deg to allow for the thermal resistance and lag, then you get worried as the max cell temp is only 55 deg.
Maybe we could live with that. Not sure.
You make a board that bolts onto one cell with a 2-pin connector going off to the next one and that's it.
Certainly halving the number of boards and bits is worth considering.
We can only fit four 5W resistors on a board. If we only have one board for 2 cells then we can only use two 5W resistors per cell. That means we'd have to halve our bypass current, from 1 A to 500 mA. Again, I suppose we could live with it, but do we want to?

Connectors are expensive. I imagine we'd just have two flying leads with ring crimp-lugs to bolt to the second cell. Those are not cheap either, but they would be cheaper than a printed circuit board.
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 »

weber wrote:
Maybe we could live with that. Not sure.


Certainly halving the number of boards and bits is worth considering.

We can only fit four 5W resistors on a board. If we only have one board for 2 cells then we can only use two 5W resistors per cell. That means we'd have to halve our bypass current, from 1 A to 500 mA. Again, I suppose we could live with it, but do we want to?

Connectors are expensive. I imagine we'd just have two flying leads with ring crimp-lugs to bolt to the second cell. Those are not cheap either, but they would be cheaper than a printed circuit board.


The temp thing, have to try it and see.

You can get 7W resistors almost the same size as 5W now. In my case, bypassing at 3.60 V the dissipation is 3.9W (using 3.3R resistors) and that's a 1.09A bypass current.

A 2-pin header is not expensive!

I'd use the smallest IDC connector I can find for the comms though
Last edited by Nevilleh on Fri, 11 Dec 2009, 01:01, edited 1 time in total.
User avatar
weber
Site Admin
Posts: 2807
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Home grown BMS ideas !

Post by weber »

Nevilleh wrote:You can get 7W resistors almost the same size as 5W now.
We tried them. In terms of how much power they can dissipate at the same surface temperature, we found they are really only 5.5 W resistors and therefore not worth the significantly greater cost. We also found stocks of many values were low or non-existent, as if they are no longer produced. We figure there is no magic they can do. The dissipation for a given surface temperature and air flow rate is proportional to their (exposed) surface area.
In my case, bypassing at 3.60 V the dissipation is 3.9W (using 3.3R resistors) and that's a 1.09A bypass current.
Your resistors are widely spaced and vertically oriented. This is ideal. But unfortunately we have neither the area nor the height available on top of our cells.

You may have noticed in the photo that we have 3R3s on the outside and 3R9s in the middle (2S2P) as the outside resistors have more exposed surface and so can dissipate more power at the same temp. We need to keep our hot-spot temperature lower than the insulation temperature rating of the cables that are sharing the same space.
A 2-pin header is not expensive!

I'd use the smallest IDC connector I can find for the comms though

The header is not expensive, but what about the connector to plug onto it? Also, relying on a connector, even gold-plated, to carry low signal current for a long time in such a hostile environment (heat from bypass resistors and possibly corrosive vapours from cells) seems like asking for trouble. I think we should use soldered strain-relieved connections, as I understand Rod Dilkes uses (who has had lots of experience with cell-top boards).
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).
Post Reply