Who's online
There are currently 0 users online.
Who's new
- eric01
- Norberto
- sarim
- Edd
- OlaOst
Support V is for Voltage
Electric vehicle related discussion, advocacy, and information.
There are currently 0 users online.
This applet is a fantastic tool - I can only hope that it was programmed properly and therefore gives reliable results.
But, seriously, there are likely several problems that need sorting out before this is ready to be tested out on a real 150V battery!
The applet simulation is easy! Just a drop-and-drag and a copy-and-paste exercise. The applet writes all the code, I have no idea how! You instantly see the results of any changes to the circuit, it's a bit like playing a computer game - trying to find the way to the next level! If you get zapped, you just get a new life....
Trying to work this out with a real 150V battery would have taken months or years (for me)!
The difficult work was done by the inventor of the original circuit and by Mikemitbike who found it and adapted it for use with three battery segments.
The work done so far is based on a circuit by G. La Rooy, Christchurch, New Zealand, described here: http://www.siliconchip.com.au/cms/A_30610/article.html and here: http://www.extremecircuits.net/2009/...r-circuit.html
This circuit was then extended to work for three battery segments by "Mikemitbike" here: http://visforvoltage.org/forum/9675-...#comment-54426
The idea to consider opto-relays (if that is the right name...) came from "SgtWookie" at http://forum.allaboutcircuits.com/showpost.php?p=289978&postcount=5 .
I did just play around with various values in the applet to tweak it a bit so that transistors with a hFE of 50 can be used, but the simulation was otherwise already complete. It was made using the applet at http://www.falstad.com/circuit/
I spent a bunch of hours to learn by trial and error which combinations of resistor values etc. do what to the simulated circuit. I do not really understand how it works!
We really really need people who have the needed know-how to help with this. The transistors might be overkill (wrt their 250V rating) while the opto-relays might possibly just fry (rated for 60V max, I think).
I hope The_Laird will find time to have a good look at it all.
The biggest question mark hovering over all of it is of course if the basic idea for the IDeA is viable! For all I know, it might have been patented years ago, or it might be rubbish, or both! It seems to work with this one used battery on my garage floor at 24degC, but who knows what would happen at -3degC or at 50degc in real road use? The whole concept might be a disappointment, but the core idea might also be a breakthrough in NiMH battery management. The chances are that it will be a flop, because I cannot believe that no-one else has come up with it and tried it before. It probably does not work well for some reason. Or it is being used already in the Prius, but needs much shorter battery segments to work, like the 12-cell segments in the Prius.
The core idea of the IDeA is this:
1) It is too difficult to consider all the factors that influence the voltage of a NiMH battery. Temperature, current draw, SOC, age, etc...
2) Ignoring all these factors is much easier than trying to properly integrate them.
3) Comparing the voltage of large numbers of similar cells under similar circumstances (similar for temperature, identical for current draw) at exactly the same time should detect the sudden voltage reversal of the one cell with the lowest SOC.
4) That will hopefully allow to largely prevent one of the two most damaging effects on NiMH batteries - cell reversal. The other main damaging event, over-charge, is more easily managed because it usually occurs parked, out of traffic (except maybe for regen braking with a full battery).
This information may be used entirely at your own risk.
There is always a way if there is no other way!
of course, you're right.. It isn't all your work, but still :-) give yourself a little credit, man!
I understand the workings of the applet (or the idea of it), but still, this must be a dream for you :-)
I follow this thread eagerly, though I don't fully understand the circuit itself (any more, it's getting too complicated for me)
you said:
"3) Comparing the voltage of large numbers of similar cells under similar circumstances (similar for temperature, identical for current draw) at exactly the same time should detect the sudden voltage reversal of the one cell with the lowest SOC."
That is the key to it I think! No matter the circumstances, age, etc.. the IdeA will sort out differences in voltage anyway!
(when finally working properly of course)
I'd be hoping that besides the_Laird, other electronical geniusses would read this thread and give you some feedback that I can't, but I realize that this is a serious amount of information to process, and many would leave just by seeing the length and complexity of this discussion (while I'm adding more shit to it than needed:-)) I read EVERYTHING on this forum (V-related that is), also the "boring" parts, but I think not many would do the same..
Don't worry about patents though; when you're thinking Vectrix, you're not endangering other (bigger) markets (yet?)
I hope it would, because it could be your success!
that all said,
in opposition to your other invention, the ABCool (which is unneeded for me in my cool country), I (and others) would be helped a LOT with this IdeA-addition, and I'd be among the first to adopt it, preferrably built by you, and I'd be very happy to pay for such an addition! (hint)
Besides that, I still realize that you're a great person with his heart in the right place, because you don't HAVE to share all this with us, and still you do that! I know that you also learn a lot here, but you have given FAR FAR FAR FAAAR more than you take! (unless you will become rich in the future from it of course)
And that is why you deserve the credit!
My vectrix has more value to me thanks to you!
I don't wanna sound mellow, but I think I speak for all of us, when I say I love you and thank you indefinitely (for all your work alltogether)!
Great work pal!
"doing nothin = doing nothing wrong" is invalid when the subject is environment
Thanks Turok, I'll have to make sure I polish my halo before logging on now... HAHA!
I think what goes around comes around and the rewards don't necessarily come from the direction of the effort, if you know what I mean.
About patenting: I did not mention patent because I want to patent this or anything else. I think patents are a thing of the past, the Chinese don't give a toss about them anyway. And I will not b able to build this for anyone but myself, I don't want the liability etc and I do not have the appropriate qualifications for it. It would be asking for trouble!
If we manage to develop the IDeA and it actually works, and it is not already protected by a patent to someone else, then anyone can build it for themselves, or sell kits, or complete devices (including the Chinese)!
This information may be used entirely at your own risk.
There is always a way if there is no other way!
Hi Mik,
I don't wish to dampen your enthusiasm but you might just be getting ahead of yourself a little here.
Theses simulation programmes are not invincible. Your suggestion of a ten ohm inrush current limiter in your recent post is great except that the instantaneous power will be approximately 2250watts (yes, that's 2.25 KWatts) at a current of 15 amperes. I know that it is only for a fraction of a second but your resistor won't last long with that kind of beating up.
By the way, I have just sent you an e-mail response to your circuit, hope that it helps.
Keep trying and smiling:-)
The Laird
Thanks! It sure helps! I have read it and am trying to process and understand...
The 10 ohm resistor I have in mind (and own 7 of!) weighs about 300 grams (guessing) and is used in the NHW10 Prius as the inrush current limiter for the 240s NiMH battery. Similarly, the contactor would be from a NHW10 Prius because I have a few already.
This information may be used entirely at your own risk.
There is always a way if there is no other way!
Another option is to use a normal 240v incandescent light globe.
The pre-charge won't be all that long, but it should be reasonably reliable.
Matt
Daily Ride:
2007 Vectrix, modified with 42 x Thundersky 60Ah in July 2010. Done 194'000km
That works much better as an ICL for maintenance, particularly because the correct functioning is visible as the glow of the globe.
But, permanently installed, the risk of wrong connection or other malfunction is much lower.
It depends very much on how soon the ABCool SMPS (or DC/DC converter) starts to put out 12V. There needs to be enough time for the capacitors to charge up before the full (low resistance) connection is made, otherwise you end up with inrush current again.
Another problem is that a permanently installed light globe might not last long, due to the vibrations during on-road use.
Some sort of time-delay should be relatively easy to rig up - maybe a resistor in series with the contactor's coil, in combination with a capacitor. If the resistor only lets 500mA through (12V / 0.5A = 6 ohm; 0.5x0.5x6= 1.5W), and if the contactor needs 450mA to close, then someone smarter than me could calculate how big a capacitor one needs to install in order to delay the contactors closure by about 5 seconds. This would then allow to use a higher value resistor as ICL.
A 100 ohm resistor would experience only 150V / 100 ohm = 1.5A peak current, and 1.5A^2 x 100 ohm = 225W peak power - but only for a fraction of a second. A relatively cheap resistor network of 5 x 5 equal resistors would bring the peak power down to 9W, I think they would last for decades!. 3 x 3 resistors would experience 225W / 9 = 25W peak power for a fraction of a second. I assume 10W rated resistors ( 100 ohm x9 ) would last longer than the scooter.
This information may be used entirely at your own risk.
There is always a way if there is no other way!
I have been snooping around in transistor data sheets a bit in the last few days, and it appears to me that the transistors to be used need to be low current ones.
Much of it I don't understand, so it is entirely possible that I am getting it wrong. But anyway, unperturbed by this, here are my half-baked conclusions and assumptions so far:
Those transistors with a higher power rating do not have a linear (or flat) hFE vs Ic current graph in the area of 1mA to 10mA (The opto-relays switch on when they receive 3mA from the transistors and off again at 0.8mA . See http://www.farnell.com/datasheets/358205.pdf).
Darlington-type transistors have higher hFE values, but they need typically hundreds or thousands of mA to operate in their specified high-hFE area.
It seems a bit tricky to find PNP transistors and NPN transistors that have otherwise exactly the same characteristics. I assume that it would not work well to use PNP and NPN transistors with different hFE ratings in this circuit, because this would cause different sensitivities for the imbalance between different IDeA segments. But I have not tested this yet.
Regarding the required voltage rating for the transistors: After playing around with the above "Fault Simulator" circuit http://visforvoltage.org/forum/9675-how-improve-nimh-vectrix-battery-it-becomes-damaged#comment-54947 I figured that a 60V rating (for collector-emitter and collector-base voltage) is fine, because I could not cause any voltages higher than 51V across any of the transistors using the simulation. The transistors used in the previous simulations were rated for 250V, to be on the safe side, but they had only a hFE of 50. A low hFE makes it much more tricky to choose the resistors in the simulation appropriately, without exceeding power ratings and without running into other problems. The main difficulty is that the sensitivity for imbalance is different at high SOC and at low SOC.
It seems that with low-value resistors in the circuit, there is much current for the transistors to function well enough even if they have a low hFE. But it means that the resistors will be hot all the time and that the power draw is increased. Using higher value resistors reduces power consumption and heat generation, but makes it difficult to find combinations that have a sensitivity of about 0.4V at low SOC without permanently triggering the imbalance LED at high SOC (even if perfectly balanced).
In the end, I just went and bought some transistors because they were the only ones in the shop that had the same hFE rating for PNP and NPN versions, and a 65V / 80V rating. They are however surface-mount devices and very tiny. But I think I'll manage to solder them, somehow....
Here they are:
http://www.nxp.com/documents/data_sheet/BC846_BC546_SER.pdf
http://www.nxp.com/documents/data_sheet/BC856_BC857_BC858.pdf
.
.
.
Below is the code for the latest version of the simulation. It took many hours of trial and error to get there.
The "specs" that I have used to evaluate various permutations of this circuit are these (results shown for the below circuit):
Sensitivity (mV) at various battery SOCs / voltages: For no imbalance, for 0.8mA current (=> opto-relay opening) and for 3mA current to opto-relays (=> closing):
At 154V (51.33V/segment 25.67Vx2/segment 1.51V/cell ): (Battery does not get that high - just to test with a safety margin)
without imbalance: 46microA to opto-relays; just <800microA @ 180mV ; 3mA @ 370-390mV imbalance.
At 151V 50.33V/segment 25.17Vx2/segment 1.48V/cell :
without imbalance: 44microA to opto-relays; just <800microA @ 180mV ; 3mA @ 370mV imbalance.
145V 48.33V/segment 24.17Vx2/segment 1.42V/cell: (This is about the final charge voltage for Freddy charge at 0.3A @ 32degC)
without imbalance: 41microA to opto-relays; just <800microA @ 190mV ; 3mA @ 370-390mV imbalance.
140V 46.67V/segment 23.33Vx2/segment 1.37V/cell : (Voltage settles about here after a normal charge)
without imbalance: 38microA to opto-relays; just <800microA @ 190mV ; 3mA @ 380mV imbalance.
125V 41.67V/segment 20.83Vx2/segment 1.23V/cell :
without imbalance: 32microA to opto-relays; just <800microA @ 190mV ; 3mA @ 390mV imbalance.
91.8V 30.6V/segment 15.3Vx2/segment 0.9V/cell :
without imbalance: 16-19microA to opto-relays; 745mA at 230mV imbalance ; 3mA at 400mV - 420mV imbalance.
Imbalance causing segment currents at 154V (mA):
Segment 1 : 175.27mA
Segment 2 : 175.24mA
Segment 3 : 175.27mA
Max power at balancing resistors at 154V : 263mW
Here is the code:
This information may be used entirely at your own risk.
There is always a way if there is no other way!
Here is a circuit showing one suggestion by TheLaird: Removing two of the transistors - and it still works!
However, I have not been able to find combinations of components that will work across the entire SOC range.
The code below shows what happens when two of the transistors are replaced with 1.38k resistors. I chose 1.38k resistors because this causes 3mA to flow though the LED's when there is 400mV imbalance between segments at 0.9V/cell.
The second lot of code shows the same circuit when the battery voltage is 145V: All LED's are permanently lit, irrespective of the presence of any imbalance!
.
.
.
This information may be used entirely at your own risk.
There is always a way if there is no other way!
Hi Mik,
I had a look at the chanced circuit. In my opinion the circuit with the values you addet in your post Fri, 10/08/2010 - 18:38
would be the better solution. The reduced circuit works only in a small voltage-range which might not be satisfying our needs.
I try to find the post where I wrote, which transistors I´ll use for my prototype. They stand high voltage with a hfe around
50. I´ll have a second look at the datasheet when they have which hfe and if the vale is high enough for our needs.
Greetings Mike
But the fast and strong response is only one part of the picture - the other part is the power dissipation across the various components!
The 2.5k resistors in the post you mention would dissipate over 1W in heat when the battery is at full SOC. You could use a 0.5W resistor network, (or 5W wirewound resistors) to bring the power per resistor down, but it adds to the component count (or size) and will take up space on the PCB. It also makes the entire device run hot all the time, which may well affect it's longevity.
The power dissipation of the transistors would also be quite high, probably too high to keep their temperature stable, so that the hFE remains stable. If the transistors dissipate too much power, then their temperature will be much higher at high SOC than at low SOC. Have a look at the data sheet and compare the curves for 25degC, -55degC and 150degC. http://www.nxp.com/documents/data_sheet/BC846_BC546_SER.pdf
I finally got some general advice on how to choose the components, see http://forum.allaboutcircuits.com/showpost.php?p=294859&postcount=17
and am in the process of assessing the circuit components with this in mind. The circuit from a month ago would fail on several counts!
Below is the code from Oct 8th, adjusted to show the situation with a (over-) full battery. Check the power dissipation at the resistors and calculate it for the transistors.
This information may be used entirely at your own risk.
There is always a way if there is no other way!
Hi Mik, and the others offcourse! ;-)
I had a look at the onlinestore (Conrad-at) where I usualy buy my stuff. 5w resistors have 10% varition which would be insuficient.
Using 1W resistors are available with 1% var. and they are cheap. I would have no problem with using 2 resistors in series. It´s not
VERY elegant but would be an acceptable solution for me.
I read it, it a good advice indeed. I usualy try to find components this way already. With the resistors I´ll work as descibed above.
I checked the simulation and you are right with the high powerratings. I only ment the behavior of the Oct 8th design not the powerratings themselve. With the transistors I´ll have to check some datasheets too.... but I´m still optimistic to find some "right" in the printcataloge from Conrad.
Greetings Mike
I found that the power consumption and the ability to detect imbalance with high sensitivity and similar sensitivity across the entire SOC range are intrinsically linked.
When I choose the resistor values in the circuit so that it draws a lot of power, then it is capable of sensing more effectively, more accurately and more evenly across the entire SOC range.
I am trying to include simulations from 91.8V (=0.9V/cell) to 154V (=1.51V/cell).
But, the more different approaches we try, the better!
Strangely enough, my basic electronics book (I have not read much of it, yet...) says this about wirewound resistors:
"Wirewound resistors can be manufactured to have values within a very close range. They are precision components. Also, wirewound resistors can be made to handle large amounts of power." But you are right, the 10W wirewound resistors in the shops have poor tolerances for some reason.
We need to keep in mind that there is very little space available in the battery compartment. Not much at all will fit in there without impeding the air flow. And if the device always ends up being 100degC hot, then it might heat the closest cells to it and cause imbalance instead of just detecting it.
This information may be used entirely at your own risk.
There is always a way if there is no other way!
Hi Mik,
I have studied the circuit as a whole and it seems to me that you are trying to achieve the impossible.
Your circuit is requiring transistors to behave as switches, it requires opto coupled 'relays' to behave as conventional coil relays and it is also exceedingly complicated and operates at high voltages.
Your transistor problem is caused by the 'knee' of the hfe curve at low current operation of the transistor. This is an inherent feature of all transistors and is made use of for automatic gain control.
You are having difficulty with the switching 'levels' of the opto coupled 'relays'. I think that this is a drawback of these devices. If you were to measure exactly the current at which switchover takes place you will probably find that there are differences between individual devices. You may also find that the hysteresis varies between devices, this too will cause problems when using the devices to obtain precise switching points. If there is no hysteresis that too can cause problems in your circuit.
Looking at the whole problem again, what is needed is a three/two? stage comparator from which you require an output on imbalance (irrespective of where that imbalance occurs). You also require a means of running the impellers when riding or on imbalance or on deep discharge. That would seem to be all you are trying to achieve. Correct me if I am wrong.
It would be advantageous to reduce the voltages being compared down to a more easily managed level, the voltages could then be compared in an operational amplifier comparator system and the op-amp outputs could be used to operate conventional coil relays giving more predicable and controllable results.
I have looked briefly at a simplified version of what is required and believe that a practical circuit is possible (to achieve your objectives).
The design of such a circuit as indicated above will require considerable development before a working, practical, and repeatable (copyable, mass production) circuit is finalised.
I have a feeling that I am not being very helpful, but that is how I see things. Sorry if this makes depressing reading.
I will post this on VisforVoltage, it may save others from beating their heads against that same brick wall and it may bring out some more (better?) ideas than mine.
Your comment on wire wound resistor tolerances. The tolerance figures given simply mean that all of the resistors in that value category will be within the limits of tolerance. Any individual resistor will have specific measurable values, these will not change (unless it is seriously abused or it fails (which is almost invariably open circuit)
I will try to provide you with some outline circuits in the next few days.
Keep smiling:-)
The Laird
Hi Mik, there is still enough calculation left for us ;-)
Well by using two 1W resistors in series each has only 0,5W to handle, that should keep the tempperature moderate.
The size is around 5-6 mm lenght and 3mm diameter (DIN 0207) (Source:http://www.token.com.tw/pdf/melf-resistor.pdf)
The size DIN 0414 I haven´t found yet. Found it 04 is diameter 14 is lenght, both in mm. Should be still able to
handle such sizes in the safty-container.
Mik check these transistors (PNP high-voltage transistor 2N5401 and NpN Version 2N5551.), they should work (its
still tricky for me t read english datasheets as I sometimes mix up text and meanings...
Greetings Mike
Hi Laird,
as both of us Mik and me ar rather basic in electronics your imput is highly wellcome.
Even if you don´t think the design will work well, could you have a look at the simpler
design of the circuit (only the detection-part). Mik´s post from Tue, 11/09/2010 - 04:00.
One of the reasons we split up the battery in three parts: With this configuration the
IdeA can be attached to the battery without pulling it out of the scooter, which would
mean a lot of work if done.
Greetings Mike
Guess what...The Laird is right again, as usual!
I should have known better than to use SSR's of any description - the solid state relay I used for the NHW10 Special Freddy charger gave me a lot of grief, too! They do weird stuff that makes no sense to mere mortals like myself....
I read the data sheet for the ASSR-1228 Form A, Solid State Relay (Photo MOSFET) at http://www.farnell.com/datasheets/358205.pdf a little more thoroughly today, and found that I misread some details. And, that the 3mA switching current mentioned in the overview is later described as 0.5mA in the fine print.
So, I built this circuit on a breadboard today to test it:
I had to change the value of the resistors a wee little bit, though....until I eventually ran out of 1Mohm resistors, but the LED was still lighting up! At 10V, with 10Mohm in series!!!
This ASSR-1228 may be good for rapid switching, when three seconds are an eternity, but it seems like it will always close the switch eventually even if just one micro-A is applied. The switch closes slowly under these conditions, i.e. the LED comes on gradually, but eventually it will be fully lit!
I managed to reduce the sensitivity to 2.8mA by putting a resistor wheel in parallel with pins 1 and 2 (input side) of the ASSR-1228. Dialing up 330 ohm gave reliable results with switching on gradually around 2.8mA (and sudden switching off when the trimpot was being adjusted up so that the current dropped).
But, guess what, the sensitivity depends on if I have my fluorescent lamp magnifying glass on, or not! It seems to me that the solid state relay is so sensitive to electro-magnetic disturbance that it will most likely mis-behave if placed next to the battery and motor controller in an EV.
I'll have a re-think and try out if "normal" relays can be used somehow. The ones that go "Click"! It seems they are the only electronic component that I have a reasonable understanding of, anyway! (Except for forgetting the diodes that redirect the voltage spike when they are being turned off. HAHA!)
I have not given up on this particular approach to solve the IDeA circuit, yet! It should be fairly simple to direct the small output of the transistors to another transistor that has enough OOmphhf to make a good old (bunch of) relays go "Click"!
This information may be used entirely at your own risk.
There is always a way if there is no other way!
I think I'll have to take this back this time! While it seems that The Laird was quite right insomuch that the Opto-Mosfets may not behave the way I expected, they might be particularly suitable for the IDeA circuit! I'll probably have to eat my words sooner rather than later, but I think I'm close to getting this to work!
Following below is a diagram showing the 10Mohm in series with the 10V supply (the opto-Mosfet still closes after a few seconds under these conditions!)
And it shows where to put a resistor (labelled "R adjust") that allows to fine-tune when the opto-mosfet will turn on. That makes it possible to eliminate the problem that The Laird was concerned about: That the individual devices might significantly differ from each other!
I think they are in fact identical in the sense that they will all turn on eventually if they receive even a small current flow. The mosfet gate has some kind of a capacitance. Even if the amount of "capacitance" between individual devices is significantly different, it will not matter much for the IDeA circuit! That's because it does not matter much if it takes one millisecond or 3 seconds for the opto-mosfet to switch to "ON" (and I expect that the differences between individual devices will be much less than three orders of magnitude.)
Here is the diagram to make it clearer:
This information may be used entirely at your own risk.
There is always a way if there is no other way!
Hello All,
Well, I have been looking at the possibilities. First things first. The object of the exercise is to have some indication of the onset of cell failure. This, of course, manifests itself in the form of an imbalance between the voltage of the cells.
There are many ways of determining when an imbalance exists, I have considered two possibilities, both of which will work but one is a lot simpler (in all respects) than the other.
The circuit below shows the Vectrix battery with a potential divider chain of resistors attached.
The following table list the voltages at all points under 'balanced' battery condition. A and C are always equal and B and D are also always equal.
The top line lists the cell voltages as would be expected on a discharging battery
The second line lists the total battery voltage at the above cell voltages.
The remaining lines list the voltages at points A, B, C, D.
The are the voltages of a perfectly balanced battery with all cells performing equally.
The next table is laid out exactly as the one above but I have introduced a single cell whose voltage is o.2 volts lower than the others. The list of voltages in lines A, B, C and D now show small differences which could be used to indicate imbalance.
Two more tables showing the effects of the faulty cell being in other areas of the battery.
The circuit above and the voltages under the listed conditions are attainable. But why not simplify things?
The next circuit is a simplified version of the above but works just as well. Here we have a resistor chain, but less of it and the battery has been 'center tapped' (divided electrically into two rather than three.
Tables showing the voltages at all points and under normal and fault conditions are shown below.
(Sorry about the scribbled out bit, but I am a bit pressed for time)
It is much easier to identify the imbalance in the second /last /simpler circuit because there are only two points to compare with each other.
I have almost completed some working drawings of a circuit which can pick out the imbalance and provide a switched output which could be used to achieve the desired results.
The circuit I would use (the one I am developing) operates at voltages of less than 10volts Derived from the Vectrix battery, which I feel is a much safer level to be involved with. The current version of the circuit of the detector requires a 30 volts power supply (-15.0v - 0v +15v), but this could possibly be halved.
Without going further, Is there any interest out there in what I have done and the way I am doing it? If there is, then i shall try to have a circuit available within a day or two.
Anyone out there got any better ideas?
By the way, the all Vectrix scooters already have a 'battery management system' fitted (Honest). You didn't know that did you?
Keep at it folks, it's all good clean fun:-)
The Laird
Hi The Laird,
At least on Vectrix (kept in original shape) is running without BMS only with temperaturmeasurement - mine. I had a look at the "BMS" boards
when last battery revision was done. There is no wiring for the voltage-sensores in my battery, only 12 temp. sensors...
By the way off course I´m highly interested in your solution (circuit), even if it means dismanteling the whole battery.
Greetings Mike
Thanks for taking the time to help with this, much appreciated!
If this is your objective, then it is quite different from what I am intending to do with the Imbalance Detection Apparatus (IDeA). The IDeA is about detecting SOC imbalance, not cell damage. By the time a cell shows a reduced voltage while it is still charged, it is badly damaged already. The weakest cells in the Vectux have half the capacity of the best cells, but are practically identical in voltage until they near their individual empty state!
But it might be that the difference is purely academic and that the circuit is suitable for both those purposes, anyway! I assume that there will be some part in the circuit that allows to adjust the sensitivity to imbalance for particular needs or scenarios?
My aim for the IDeA is to detect imbalance by sensing the temporary reversal of a single cell anywhere in the battery, so that the discharge current can be immediately reduced or altogether stopped (if traffic allows it). This should prevent most of the rapid battery cell damage and keep the battery good, at least if it is coupled with very gentle charging, so to also avoid over-charging events.
The avoidance of over charging depends on the presence of some type of IDeA: If there is an IDeA to warn / terminate discharge when it gets hairy for the lowest-SOC cell, then one does not need to over-charge the battery very often.
My intent is not to allow maximum range for each ride, but to allow to keep the battery in the 30%-70% SOC range most of the time, by using less than half of the maximum possible range most of the time.
In practice, the Vectrix will be discharged to about 5 bars left, and recharged to about 12 bars only. The top 5 bars and the bottom 5 bars remain unused in order to achieve battery longevity. Doing this will cause the real and perceived SOC to slowly drift apart in many circumstances.
And individual cells will also slowly become imbalanced, because there are no regular over-charge situations in this scenario. Eventually, after months of riding around 25km or less per partial charge, one cell might be so much out of SOC balance with the others that it reverses under load.
That's when the IDeA comes in: The warning lamp comes on and the rider knows it's time to gently ride to the next available charging station. The IDeA warning light should go out again when the current draw is being reduced enough to allow the low-SOC cell to bounce back up. The rider will have real-time feedback telling him/her how fast the bike can still move without damaging the weak cell. The stock system will probably still allow near full power to be drawn at that time, unless the imbalance is so mild that the first reversal only occurred when the overall battery voltage was already so low that the stock system was limiting the power.
Once the IDeA warning light has come on at a time when the battery voltage was still comparatively high (i.e. there is significant SOC imbalance), it will be time to do a slow and long EQ charge relatively soon. This could then be followed by a deep discharge, with the last part of it done automatically with the Vectrix parked.
Because the IDeA warning light would eventually come on when there are about 5/17th left on the fuel gauge, this means that the same per-charge range remains available if an EQ charge cannot be done immediately. Just charge from 5 bars to 13 bars, then down to 6 bars, and then cycle the battery between 6 bars and 13 bars until there is a good opportunity for an EQ charge. Once the EQ charge is done, ride it down to 5 bars, then only charge up to 12 bars each time again.
7 bars of charging will take about 12min x 7 bars = 1h:24min to fill in CP mode. The CC charging mode would only ever be used when an EQ charge is desired or needed.
For someone who wants to use their Vectrix to the full range frequently, the IDeA would be also useful, by providing a warning when the capability of the weakest cell is being exceeded.
But the main function of the IDeA would be to allow prolonged shallow cycling without risk of silent cell reversal due to the inevitable drift between real and perceived SOC and slightly differing self-discharge rates.
My rationale for proposing three rather than two monitored battery segments is based on two main points:
1) Tabs 35 and 68 are easily accessible and the IDeA could be installed easily, without any need to dismantle the battery.
2) This is a hunch, or an educated guess, or maybe speculation; I have not had the time and inclination to do the mathematical analysis properly, but I am simply convinced that the following statement is close to the truth: The likelihood for identical numbers of cells beginning to reverse at almost the same time in all battery segments is an exponential function of the number of battery segments that are being monitored.
In other words: If one cell in each battery segment reverses simultaneously, the IDeA will be as blind to it as the stock system is! This event is much more likely if there are fewer monitored segments.
Oh yes, please!
My understanding is that there were plans for inclusion of a BMS, but they were canned to save costs and to bring the VX-1 to market, regardless of if it not being quite ready at the time. So instead of finishing the work on the BMS, they just started to sell the VX-1 without a functioning BMS.
I think it was X Vectrix who mentioned that the stock wires to tabs 28 and 76 are not being used for anything. This is supported by mikemitbike's observation that his Vectrix runs without these wires (and was sold "stock" like that), as well as by antiscabs and others experience that the VX-1 runs without much complaining even when the temp-sensor and voltage sensor boards are completely removed. And, last not least, in my Vectux all the weak cells are deliberately concentrated into the 27-cell segment at the positive end of the battery - but the stock system seems utterly oblivious to this!
This information may be used entirely at your own risk.
There is always a way if there is no other way!
Hi The Laird.
So we will have two options and everyone can decide which solution he will take: Simpler circuit but dismantle the whole battery,or more complicated circuit but only 4 cell connections to open. As mentioned in some posts before I´m tending to the second option, but am still high interested how you will solve the problem.
Greetings Mike
I was working on a Idea of using 1 microprocessor per battery cell that would keep a central computer updated as to the voltage status of all the cells. During battery charging this system could indicate when to stop charging before any cells were over charged. While driving the scooter it could inform you of the voltage of the weakest cells to prevent you from overdischarging them and would also allow you to discharge all the cells to a particular voltage for refreshing the battery. If this was perfected then the circuitry could be miniaturized by using the SMD version of the chips on a circuit board. Although I've bought some of the parts I haven't been able do any work yet for a number of reasons.
Thanks for the help from the Parallax forum for the serial comunications theory see;
http://forums.parallax.com/showthread.php?t=125444&highlight=TIL111
I don't recommend Parallax microprocessors for this project unless you only use a few of them because they would be to expensive, I myself prefer them because I've had experience with them.
Here's some more detail of how this would work; an ultra small PC using a two wire cable activates a 5 volt relay on #1 battery cell's circuit board that connects the booster (TPS612020) to the battery cell. The booster converts the .3-1.5 volts into 5 volts that powers up the microprocessor (Picaxe-08M) and after a 1 second delay then serially sends voltage data through an opto isolator and to a diode to a two wire serial cable connected to the PC. Then the microprocessor activates a 5 volt relay on #2 battery cell's circuit board using a two wires of a 3 wire cable. This cycle repeats itself again an again until all 102 cells have sent there voltage data at which time the PC computer deactivates the relay on #1 battery cell which in turn powers down the rest of them.
The battery drain part would work by using a resister circuit on each circuit board to drain the battery cell to a predetermined voltage. This is activated by the microprocessor after receiving input from another bucket system using a 3rd wire in the cable.
If this seems like an interesting alternative to the other proposed ideas than anybody is welcome to make suggestions or use any of these ideas they want to.
Hi Anderson, sounds like "state of the art BMS solution", for my part I´m afraid I would not be able to build such modules myself (I´m not
able in programming microcontrolles (Picaxe-08M) and soldering smd parts). Another part of wich I´m afraid of is price (est. ~1000 Euro) and
time for building the modules and connecting them to each battery.
Greetings Mike
Regarding price, one thing it has going for it if it worked and I don't know if it would is economy of scale. Imagine 10 Vectrix's having this system, that's 1020 identical circuit boards all with the same components and all programmed the same way. Maybe that would be enough so it could be mass produced on an assembly line. Anyway the booster chip only comes in a 3 mm x 3 mm QFN-10 Package which means the work would have to be out sourced.
Maybe you could use this opto-isolated serial connection principle to measure segments of 3 or 4 (or even more) cells. That will drastically reduce the power consumption and component cost.
This information may be used entirely at your own risk.
There is always a way if there is no other way!
I have done some more testing, experimenting and simulating for the IDeA circuit using the six transistors and 8 opto-Darlingtons.
The result is that I am convinced this is likely going to work well! And I don't think this circuit is very complex. It has relatively few parts to go onto the PCB.
Looking at the transistor data sheets again, I found that the best linearity and similarity between hFE values for the NPN and PNP (BC846 + BC856) versions are between 0.5mA and 1mA Ic. The hFE vs Ic curves are flat and almost equal with hFE=190 across that area.
Therefore I chose 0.75mA as the target current to be delivered to the opto-darlingtons at the imbalance-trigger point.
The next step was to find the best resistor value to make the opto-darlingtons switch the output side "ON" at 0.75mA. That was determined with the breadboard test circuit shown earlier at 10V.
Testing just one ASSR-1228 I found that with a 1.2K (1191 ohm measured) resistor across it's input side it will reliably switch "ON" when more than 712uA are flowing through it (and the resistor). It barely switches on at 712uA, with 727uA it turns on slowly but eventually fully.
At 0.75mA it takes 18s until it begins to close enough to dimly light the LED in the test circuit; after 35s the LED will be fully lit.
At 0.76mA it takes 8s to dimly light, and 20s to fully light the test LED.
At 0.8mA the LED lights up fully within 3s.
The (single tested) opto-darlington (with 1.2k resistor) will switch off again when the current is slowly reduced to 695uA. If the current is reduced suddenly, it will turn off temporarily even if the current is above the threshold for turning the output side on. (I don't yet understand why that is happening, but it will potentially be a nice feature - giving feedback to the rider that quickly easing off on the throttle does help! If the new, lower throttle demand is still too high, the warning LED will come on again a few seconds later.)
I have not tested the other opto-darlingtons, because I initially bought the SMD "Gull Wing" version - it's a hassle to solder wire to the pins so that it fits onto the breadboard! But I have a few 8-pin DIP through-board versions on order now. Once they arrive I will test them all to see if they will all function with the same 1.2k resistor value across the input side. Something I had not realized earlier is that the ASSR-1228 actually contains two devices in the one package. Less components to be fitted on the PCB!
If there are significant differences between individual opto-darlingtons, I expect that I would be able to fully compensate for this by using slightly different resistors across their input side.
The next step was to adjust the "On-current" parameter for the SSR1-4 in the simulation applet to 750uA. That way I think they represent the ASSR-1228 with 1.2k resistor behavior. The resistor is not present in the simulation because a coil-relay is all I could find in the simulation applet. But the properties for the coil relays SSR1-4 (and SSR 5-8) are set to represent the behavior of an ASSR-1228 opto-darlington (or at least that's what I've tried to do, it might be very different in a real device, that remains to be seen!)
Next step was to adjust the 4 trimpot values to tune the circuit so that imbalance of 400mV or more between any of the three battery segments (at 91.8V battery voltage = 0.9V/cell) will trigger the imbalance detection.
The values turned out to be 785 ohm and 2.92k ohm. Therefore, a 2k-multiturn and a 5k-multiturn trimpot should be sufficient to allow adjusting of the sensitivity in the real prototype device.
Then I checked the simulated sensitivity for imbalance at other battery voltages (up to 154V), the maximum power dissipation at 154V at the 10k resistors, the (potentially imbalance causing) current through battery segments 1,2 and 3, and the maximum current flowing through the SSR's 1-4 when imbalance is present (including simulated disconnection of any of the 4 tab wires due to failures).
This was done using the simulation code shown at the end of this post; it records max/min currents and power levels in the scope traces. Cut any combination of the tab wires and change the voltages of the six battery segments to anything remotely possible for fault simulation - and let me know if you can break it!
All of the results seem to be well within specs, across the entire expected SOC range and a bit beyond it, too!
I would appreciate it very much if you could double and triple check the results before I start building the prototype!
Results of my simulations:
The imbalance sensitivity ranges from 370mV (full) to 400mV (empty) across the entire plausible SOC range.
Max current at SSR's 1-4: 5.13mA.
Max power at any of the 10k resistors (at 154V): 263mW.
Imbalance causing current (battery balanced at 151V) (mA): 40uA differential current.
Segment 1 (mA): 171.82
Segment 2 (mA): 171.78
Segment 3 (mA): 171.82
Imbalance causing current (when unbalanced at trigger point at 151V) (mA): 190uA differential current.
Segment 1 (mA): 172.15
Segment 2 (mA): 171.34
Segment 3 (mA): 172.14
The only power rating (apart from the ones I have not thought of!) which I have not checked yet are those for the transistors. I bet they are well within specs, because of the maximum 5.13mA current demanded of them under any conditions, while their maximum rating is 100mA. However, I do not understand how to calculate the maximum power dissipation for the transistors, yet. I need to read up on it until I can understand the explanation given here: http://forum.allaboutcircuits.com/showpost.php?p=295084&postcount=19 by SgtWookie:
Here is the code for the applet at: http://www.falstad.com/circuit/
This information may be used entirely at your own risk.
There is always a way if there is no other way!
2-3 cells would work but 4 cells would be close to the the booster's input limit of 5.5 volts. Higher than that and you should look for a booster in a higher voltage input range and eventually a 5 volt voltage regulator I guess.
The cheapest microprocessor I could find is model MSP430G2131 with an ADC (analog to digital converter) and operates at 1.8 - 3.6 volts, sells for .49 cents. starting at quantity's of 1000 ( I don't know what the price of lesser quantity's are). Another economic incentive for using the MSP430G2131 is you could use a more economical 3.3 volt voltage regulator in place of the booster (TPS612020) for 4-6 cells but I don't know how good it is for prototype development purposes.
I think the Picaxe-08M microprocessor which operates a 5 volts is good for that purposes but costs over 2 dollars each.
Here's more on the MSP430G2131;
http://www.ti.com/ww/en/mcu/valueline/index.shtml?DCMP=Value_Line&HQS=Other+OT+430value
About the part of your question to "reduce the power consumption and component cost";
That's a good question you asked and I realize something I didn't before and that is if the scooter is under power at the time of the voltage reading then the battery's voltage is lower than it would be when it isn't so you would have to factor in the current draw at the time for the voltage reading to be useful. This could possibly be done by hooking up a portable clamp type amp meter that has a serial port to the central computer. That way you could do the conversion using the amp gauge input or save all the data and do it latter. Maybe instead of leaving the BMS on continuously or occasionally just try get one reading of all the cells at a steady load would be best, and if there are any problem cells then try to be conservative with the battery power until you can refresh the battery.
Note when your not monitoring the battery then all microprocessors are off. The time it takes for all the microprocessors to report the cell's voltages could be between 1-3 minutes I guess. If the central computer used two serial ports that could halve the time.
I drew a minimally detailed schematic below; (not to be used as an actual schematic) The booster has 10 terminals and requires external circuitry that I didn't show so please don't use this as an actual schematic. All relays I used are mechanical but I might change some to solid state.
I'll go over some of the basics again now that there is a schematic. Note all the NiMH cell BMS boards are the same including the software. The central computer turns on the BMS by activating the K2 relay on #1 Cell board hereafter called K2-1. If the others are turned on before K2 then nothing happens because the K2-1 relay supplies power to the to #1 cell board. Also when switched on K2-1 sends power to K2-2 relay on the Cell #2 board that in turns on the power to Cell #2 board which in turn turns on K2-3 relay and on and on until all 102 cell boards have been turned on. when the power is turned off to K2-1 the dominoes effect works in reverse and all boards are turned off. So now all the BMS boards are now on the central computer turns on K1-1 relay which the microprocessor monitors and responds to by sampling the NiMH cell's voltage, when that's done it serially sends that data out, then after about a .5-1 second delay the microprocessor turns on the k1-2 relay on #2 board to repeat what happened on #1 board to #2 board and so forth until all BMS have reported there cells voltages. When the 102 serial data has been received then central computer knows to turns off the power then because it's been counting the times the data has been sent. So it turns off relay K2-1 which turns off the rest.
I'll try to explain about draining the battery cells to a predetermined voltage to refresh the battery some time soon.
Those points highlight the reasons why I hope the IDeA circuit works.
Here is a link to a post on a yahoo forum where someone explains how the Prius ECU measures battery segment voltages at 10Hz.
An excerpt:
And a link to the labeled Prius ECU: http://ingineerix.com/pic/?priusg2batecu
You'd probably have to sign up the the yahoo group to read it all.
This information may be used entirely at your own risk.
There is always a way if there is no other way!
Hello again,
Long delay but I have a circuit which will/can be made to work.
Using the simple system of dividing the battery in two sections as already demonstrated, the tapings A and B are taken into an op-amp comparator.
Points A and B are taken to the comparator input which is shown below.
This comparator will sense a 0.2 volt drop in either section of the Vectrix battery and produce an output voltage swing to either negative or positive (depends which section loses the 0.2volts).
The comparator output is taken on to the 'self latching fault signal output stage' which provides outputs according to the relay used.
The relay could be a 24volt multipole changeover type so there are possibilities for all sorts of uses.
The circuits above are not tested, but I have every reason to believe that they will work as required. They can easily be reconfigured if necessary to sense lower volt drops or greater voltdrops.
Note that, once the battery output has passed through the ten megohm resistors, there is no longer any danger from the Vectrix battery (voltage or current). The voltage at the op-amp input is only 14volts and the current in the resistor chain is a mere 5 microamps.
Whilst the circuit is a practical one and whilst in theory the whole idea should work. That is, by noting a voltage drop in either section of the battery you have an indication that an imbalance exists, the reason for the imbalance is not certain.
An imbalance could be caused by one cell failing to produce the same current as the others. This could be the result of a lower state of charge or it could be a result of that cell having a slightly higher internal resistance which will show up under load as a larger volt drop than the others.
It has also occurred to me that it is almost certain that the cells will all have very slightly different internal resistances from each other. Under load this difference in resistance will result in different volt drops per cell. As the battery ages, different cells will age differently again resulting in variations of the internal resistance. There is no way to account for these potential problems and the more I consider these problems, the more complicated the game becomes.
I will leave you all to play with the various circuits, I still have some coding problems to solve.
Best wishes and good luck,
The Laird.
Pages