Old BMS Hardware Thread

[GregsGarage]

Congratulations Peter on getting the noise issue sorted.

I got my Freeduino board today and am starting to make plans for my master. Had a thought on the watchdog IC. Instead of having a pulse from the Master I am going to use the pulse from the slave bus. So the last slave will send the pulse to the watchdog PICAXE. This frees a pin for other uses and also tests all the slaves are working along with the master. I am going to order a surface mount 08M Picaxe to go on the proto board, which has a place for a 16 pin SMT. I will order some of those varistors for mine as well.

Will be waiting for your latest master software version so that I can start converting it to work on my master.

[steiner]

Peter,

Glad to hear you got the emi issues under control. I used unshielded twisted pair on mine and never had a problem. I haven't put a scope on the lines but everything worked properly from the start.

I am curious....when your first cell starts to bypass, how long does it take before the pack is balanced? I assume when the first cell starts to bypass, the charger output is reduced to about 250mA. Is that correct?

Rick

[retepsnikrep]

Peter,

Glad to hear you got the emi issues under control. I used unshielded twisted pair on mine and never had a problem. I haven't put a scope on the lines but everything worked properly from the start.

I am curious....when your first cell starts to bypass, how long does it take before the pack is balanced? I assume when the first cell starts to bypass, the charger output is reduced to about 250mA. Is that correct?

Rick

Thanks Greg & Rick. It took a while but getting there now.

The bypass load comes on when any cell goes over 3.6V, the charger starts to throttle back when any cell goes over 3.7V, the Alarm sounds and Charger relay is cut when any cell goes over 3.8V.

You need a decent charger and my Zivan is adjusted so that it is also on the CV stage when the cells reach 3.7V.

The cells stay very well in balance. When I checked the pack the other day all were within the 10mv resolution of my DMM.

Balancing takes no more than an hour at the end of the charge cycle.

Master software has got a bit bloated so i may well take out some of the flashy bits. I'm not very happy with the way it stores everything and then displays it once a second, so I may re-write all that. I'm not happy with the way it deals with errors either as they disapear if the error goes away, so you have no idea if an error has occured whilst you have been away.

Greg I quite like the idea of using the last Slave in the string to send the pulse to the watchdog. I might get a chance to look at software over next week or so with luck. Also going to pull my cells and revert to the serial data routine I think now the EMI is sorted. The Pulse width routine is useful but not quite so useful as the serial data. And I may want to transmit other codes back to the Master at a later date.

Rick do you have your setup installed in a working vehicle? Mine worked on the bench fine right from the start, when it was exposed to the heavy currents and emi it fell over. Honda did a fair bit of work in UK when the Insight was first introduced as the emi was a problem for it's own electronics even then

[steiner]

Peter,

Yes, mine is installed in my vehicle and is working fine (although it is only monitoring two cells). However, my vehicle is an off-road 4x4. Not that it should make a difference. I probably don't have as much electronics in mine as you do. I was careful to run the sense leads away from any major cables but you really can't help running them close to the buss bars that connect the individual cells.

Just for fun, I am trying to get a slave working with a 12F683 instead of the picaxe chip. I am having communication issues right now but will hopefully get them resolved.

I wanted to ask you about the serout command on the slave. My version is using "serout MasterBus,Baud2400,(VData)" but there is a symbol command that says baud2400=N1200. I assume this is because the chip is running at 8 Mhz instead of the typical 4 Mhz. Am I correct at assuming this means that it is actually transmitting data at 1200 baud inverted?

Rick

[retepsnikrep]

I wanted to ask you about the serout command on the slave. My version is using "serout MasterBus,Baud2400,(VData)" but there is a symbol command that says baud2400=N1200. I assume this is because the chip is running at 8 Mhz instead of the typical 4 Mhz. Am I correct at assuming this means that it is actually transmitting data at 1200 baud inverted? Rick

Yes and No, the N1200 means Inverted output idle low 1200 baud, but at 8mhz it is actually transmitting at N2400.

I've been looking at my serial data routine as well, which do not make sense and am about to install the latest serial software to my complete pack this morning.

This should work at the full speed 4800 baud and is more logical.

Note Slave & Master are both changed.

Slave now uses T2400 and Master N2400 (Both 4800 baud at 8mhz)

www.solarvan.co.uk/bms/SLAVE_SERIAL_020 ... 00BAUD.txt

www.solarvan.co.uk/bms/MASTER_SERIAL_02 ... 00BAUD.txt

[retepsnikrep]

Its in and running at max speed 4800 baud and I have put the battery packs back in the car and been for a 10 mile test drive.
Cells are all within 20mv which is the limit of the resolution. No unexpected errors.

Still some issue with Master software but that's expected.

What Works.

1) Cell voltage measurement and transmission to Master
2) Current sensing & AH counting
3) Charger control Relay & Opto
4) Over/Under Voltage/Temp alarms
5) Individual cell voltage view feature.

What doesn't work yet or untested.

1) Security code feature crashes Master requiring it to be re-programmed so I am removing that for the time being.

2) I2C Temp sensors are ok when engine off but when driving giving eroneous readings due to interference. Need to scope them a bit more. May move them to seperate inputs as I have some spares.

3) Speed/Distance not connected up yet. Code is active in Master though.

So car is back together and driveable. Few minor niggles to sort and re-write some of Master code but it basically works now as I originally intended.

I think the above Slave code will stay as is for quite a while now.

The Master code will continue to evolve as I correct the issues and add/remove features.

I will add remote programming of things like Slave Load on/off voltage but at the moment you have to pre-program that in Slaves.

Thanks to all for their help and support on here, you know who you are! This project would not have ben possible without the support, advice and contributions from friends around the world

[GregsGarage]

Well done Peter!! Glad to see it is all working, at least the important bits!!

P.S. Got any screen shots of the BMS in action in your Insight?

[retepsnikrep]

Might post something pics/vid when temp & speed are up and running.

RE I2C Temp I actually have two spare digital and two analog inputs on the Master board so could split them off and have one per input or fall back to something like the old LM335 linear diode analog type sensor which should work fine as I can supress that too my hearts content.

Plus code would be much simpler but I would only get two sensors. I may well add Analog temp sensors as an option for next version of the Master Board.

With an LM335 which I think I have a packet of in shed. And a 1k bias resistor

The Master analog input resolution is 10-bits and allows a resolution of about 1 Degree Celcius.

The following formula is used to calculate temperature in Celsius.

C= ( reading-558) / 2

Accurate to about 1C

This page covers it http://www.ontrak.net/LM335.htm

I could have a much improved temp sample rate as well

[GregsGarage]

Peter,

I like the LM335 idea. My arduino/freeduino board has 6 analouge inputs, so that would allow 4 temp sensors in addition to my buttons and current sensor input and the code will be simpler as well.

[retepsnikrep]

Forgot to scope the Slave bus which runs from Master down to first Slave near all the nasty noise to initiate data transfer

Just looked at it and pretty noisy so bunged an 0.1uf poly cap across it and that's sorted.

Off for another test shortly.