DIY Electric Car Forums banner

1 - 20 of 54 Posts

·
Registered
Joined
·
2,217 Posts
Discussion Starter · #1 ·
Since the Lee Hart Battery Bridge has come up a number of times it has been suggested I start a thread on the topic.

For the Buggy I'm looking into some type of Lee Hart style battery bridge so the car can be shut down easily with no parasitic pack loads. Just disconnecting the bridge pack + and pack - connections would remove the entire load from the system. I've been thinking about 2 different approaches to this.

One would be a 2 comparators forming a voltage window detection circuit. If the common input was the center of the pack and the references where about 1/4 volt above and below the virtual center of the pack the voltage window would be exited if any one cell swung 1/2 volt out of line. The output of the comparators would drive a optical isolator and then a typical 12v system control setup could drive the alerts. My Clean Power Auto BMS head end board could be used.

The other idea is to build the system with diodes like Lee designed it but instead of red LEDs use the LEDs in a 2 channel optical isolator. Then the isolated output of those LEDs could drive the warning/shutdown systems. Since this would operate the isolator output transistors is a more analog mode I would guess each channel would require a trim pot to adjust an output stage that would most likely be a TLV431.
What about starting a thread for something like this. I have a couple of emails from Lee Heart about two other ways to do this. One of them is to use a zero centered ammeter in with a +-50 uA range. The meter could be calibrated to represent any voltage difference desired. The other circuit would be some sort of comparison circuit. A couple of designs could be created with possible source of components for each one.

What do you think?
It probably is a good idea. Quite a few different approaches have been suggested. I've posted a couple of basic ideas above for how one might be constructed. I'd like to hear more ideas on how to accomplish this, as well as how to use and/or display the results.
 

·
Registered
Joined
·
903 Posts
Below is an email exchange I had with Lee Heart on a modification to his Batt-Bridge.

On 1/4/2011 7:19 PM, David Nelson wrote:
Hi Lee,

I was wondering how hard it would be to have some sort of analog
display like a typical auto ammeter or galvanometer with zero at the
center to measure the difference between the two halves of a battery
pack?
That's even easier than the existing circuit. View with a fixed width font:
HTML:
   ________/\/\__________
  |         R1           |
__|__+                   |
 ___ upper half of pack  |
  |  -      __           <
  |________/  \_________\< R3
  |        \__/ meter   /< trimpot
__|__+                   <
 ___ lower half of pack  |
  |  -                   |
  |________/\/\__________|
            R2
Use a zero-center analog meter. R1 and R2 are equal values, about 10 times higher than R3, and limit the meter current in case R3 is turned all the way to one end. R3 is a trimpot to set the balance. Adjust it so the meter reads zero when the two half packs are at exactly the same voltage.

The meter should be something like +/-50 microamps full scale, to allow the resistors to be large values and so not draw much current or dissipate much power. Set the pot *first* and then connect the meter.
Another idea I had was to have some sort of comparator circuit
that reads the two halves and outputs a + or - voltage reading to a
display. I would also want this device to be powered directly from the
pack it is measuring and not rely on the 12V system. I really like
that the CycleAnalyst runs entirely off the pack since it won't put an
imbalance on the pack.
That's not hard, either. Connect the LEDs of an optocoupler with a series resistor across each half-pack. Connect the transistor output of the two optocouplers in series. Pick the series resistors for each LED so the voltage between the transistors is half of the power supply when the two half-packs balance. It will then swing up and down as the pack's voltages are different.

With either circuit, you'll have to pick your values experimentally, based on the specific parts you use.
=======================================

Thanks Lee, I think I followed that. Just to be sure see notes below.

On Wed, Jan 5, 2011 at 10:08 AM, Lee Hart <[email protected]> wrote:
> On 1/4/2011 7:19 PM, David Nelson wrote:
>>
>> Hi Lee,
>>
>> I was wondering how hard it would be to have some sort of analog
>> display like a typical auto ammeter or galvanometer with zero at the
>> center to measure the difference between the two halves of a battery
>> pack?
>
> That's even easier than the existing circuit. View with a fixed width font:
>[see circuit diagram above]
> Use a zero-center analog meter. R1 and R2 are equal values, about 10 times
> higher than R3, and limit the meter current in case R3 is turned all the way
> to one end. R3 is a trimpot to set the balance. Adjust it so the meter reads
> zero when the two half packs are at exactly the same voltage.
>
> The meter should be something like +/-50 microamps full scale, to allow the
> resistors to be large values and so not draw much current or dissipate much
> power. Set the pot *first* and then connect the meter.


Ok so the meter is a microamp meter with an internal shunt and is
connected to the wiper of the trimpot. The meter will indirectly
measure voltage difference where the meter scale would be some
constant times the current through the meter depending on R1, R2, R3
and the total pack voltage. So if the parts were selected such that at
a pack voltage of 64V a full swing of 50uA would mean say 2V
difference that with a pack voltage sagging down to 55V a full swing
of 50uA might be something like 2.2V (not calculated, just a guess).

>> Another idea I had was to have some sort of comparator circuit
>> that reads the two halves and outputs a + or - voltage reading to a
>> display. I would also want this device to be powered directly from the
>> pack it is measuring and not rely on the 12V system. I really like
>> that the CycleAnalyst runs entirely off the pack since it won't put an
>> imbalance on the pack.
>
> That's not hard, either. Connect the LEDs of an optocoupler with a series
> resistor across each half-pack. Connect the transistor output of the two
> optocouplers in series. Pick the series resistors for each LED so the
> voltage between the transistors is half of the power supply when the two
> half-packs balance. It will then swing up and down as the pack's voltages
> are different.
>
> With either circuit, you'll have to pick your values experimentally, based
> on the specific parts you use.


So the advantage with this circuit would be that the output would be
the true voltage difference between the two halves regardless of the
changes in total pack voltage assuming that the device was
built/calibrated properly, right? This circuit will take me a little
longer to study out and understand. It is times like this I wish I had
spent more time in my Physical Electronics classes to understand what
really was going on and how everything inter-related.

================================
On 1/6/2011 1:21 AM, David Nelson wrote:
Thanks Lee, I think I followed that. Just to be sure see notes below.


Ok so the meter is a microamp meter with an internal shunt

Yes, though these meters have no internal shunt (and don't need one). Their coil is many turns of very fine wire, having 2K or so total resistance. Full scale is therefore about 50 microamps at 0.1 volts.
and is connected to the wiper of the trimpot.
Yes... meter between center tap of pack and wiper of the trimpot. It directly measures the voltage difference between these two points (with full scale being a 0.1v difference. Too sensitive; but you fix that with R1 and R2.
So if the parts were selected such that at a pack voltage of 64V a full swing of 50uA would mean say 2V difference that with a pack voltage sagging down to 55V a full swing of 50uA might be something like 2.2V (not calculated, just a guess).
Yes; that's the idea.

To calculate it, start with the equivalent circuit at the wiper. It is a voltage of half the pack, with a series resistance of the parallel combination of the upper and lower resistances (R1 + R3/2 and and R2 + R3/2). Since R1=R2 and R3 is split exactly in half at balance, this is just Req = (R1 + R3/2)/2.

Suppose we want a 2v imbalance to produce full-scale on the meter, and the meter's resistance is 2000 ohms. When one half of the pack is 2v different, the center tap is 1v different from the voltage on the wiper. To get 50ua to flow with a 1v difference, the total resistance must be Rtotal = 1v / 50ua = 20k. The meter has 2k of this, so we need 18k more.

Thus, Req = 18k, so R1 + R3/2 = 36k and R2 + R3/2 = 36k. You can use any combination that adds up to 72k; two 36k and no pot, or two 35k and a 2k pot, etc.

The pot sets the balance point (compensates for R1 and R2 not being identical values). With perfect 0% resistors you wouldn't need it. But as a practical matter, if you use 5% resistors the pot needs to be at least 5% of their value to be able to compensate for the difference between them.

With precision resistors, the meter can be calibrated in volts difference. (Remember that a 1v difference at the meter means a 2v difference in the two half-pack voltages). If you don't use precision resistors, you can add a second pot in series with the meter to adjust the meter so full scale is exactly a 2v difference (or whatever you want).
So the advantage with this circuit would be that the output would be
the true voltage difference between the two halves regardless of the
changes in total pack voltage assuming that the device was
built/calibrated properly, right? This circuit will take me a little
longer to study out and understand. It is times like this I wish I had
spent more time in my Physical Electronics classes to understand what
really was going on and how everything inter-related.
The meter circuit measures true voltage difference, regardless of the pack voltages.

The circuit with the optocouplers has an isolated output, but won't be as accurate because the optocouplers are not precision components. Their characteristics change with temperature, with age, and each part is a little different from the others. But it should work "good enough" for spotting a cell in danger.

Warning... remember that this is a very simple circuit; it can't spot a problem is TWO cells are both low in voltage, one in the upper half and one in the lower half of the pack.

--
Lee A. Hart | Ring the bells that still can ring
814 8th Ave N | Forget the perfect offering
Sartell MN 56377 | There is a crack in everything
leeahart earthlink.net | That's how the light gets in -- Leonard Cohen
 

·
Registered
Joined
·
21 Posts
I have read and understand the Lee Heart Battery Bridge, but if you are driving and the LED comes on, what do you do? Does this indicate that you need to rebalance that half of the pack? I'm afraid I don't quite understand that aspect.
 

·
Registered
Joined
·
937 Posts
I have read and understand the Lee Heart Battery Bridge, but if you are driving and the LED comes on, what do you do? Does this indicate that you need to rebalance that half of the pack? I'm afraid I don't quite understand that aspect.
No, you never (re)balance a part of your pack. Per definition you balance the complete pack. Or all to the same low voltage at the low knee to bottom balance, or to the same high voltage at the top knee. The knees can be seen in charge/discharge curves in the graphs that comes with the brand.

The theory behind this device is that if there occurs an inbalance between the pack halves, you probably have a bad cell. And the half with the slightly lower voltage is the place to look for it.
 

·
Registered
Joined
·
21 Posts
Just had a quick thought, since this is a relatively simple circuit, why not have the pack divided further, maybe into fourths, to narrow down your search for the underperforming cell?
 

·
Registered
Joined
·
2,217 Posts
Discussion Starter · #9 · (Edited)
I like the meter, but that doesn't fit with the Buggy. I would like a system with a digital output, if you will. That is a normally closed connection isolated from the traction pack (like what is typically seen with cell monitors) when all is good and having that open when the pack half difference exceeds a set value. This could be used to drive one of Dimitri's head end boards just like is currently done with cell monitors.

I see this type of monitor as a substitute for cell level monitoring (or possibly a backup to it when cell boards have load resistors.) I really think that first the charger and controller should be programmed to take care of the pack (not rely on the BMS.) Then the battery monitoring system should provide an on/off output to insist a charge be stopped or drive be cut short. I don't want to wait until I teakettle a Lithium cell to find out something went wrong.

Code:
                pack center           
P                   |                p
a           D1 //   |                a
c           --->|---)-------         c
k     R1    |       |      |  R2     k
+ ----^^^------>|------>|----^^^---- -
               \\      \\             
            O1 ___     ___ O2         
               / \     / \            
              |   |----)--|           
              |--------|  |           
         R3   |           |           
+12 -----^^^--|           |           
              | Z1        |           
output  -----|< -------------- -12
Pack +, center, and - should be self explanatory. Remember that the pack + and - connections needs fuses at the cell terminals.

R1 and R2 are precision resistors of the same value to allow the needed current to flow, perhaps 10 milliamps at 2.5 vpc. For a 40 cell pack that would be 5000 ohms each.

O1 and O2 are a pair of optical isolators, transistor type. The IR LEDs are part of the pack circuit and the isolated photo transistors are part of the vehicle 12 volt system.

D1 is a red LED with a turn on point about 1/4 volt higher than the IR LEDs. If any one cell drops or raises 1/2 outside of its neighbors one of the photo transistors will be turned on.

Z1 is a TLV431 programmable zener. R3 keeps Z1 on unless one of the photo transistors turns on and grounds Z1 turning it off. This can be used to turn off a common DPDT PC board relay. This provides useful outputs to turn down a controller or turn off a charger or just scream warning alarms or whatever you think a battery monitor should do.

This is so quick and so crude I'm sure it won't quite work yet. Tear it apart please. :)
 

·
Registered
Joined
·
1,900 Posts
Good thread for a good idea. I'd like it to be a little more sensitive, maybe a volt or 1-1.5V per led and have about 9 led total, four each side and center. This way you get a little more data. Ie a high resistance connection will show up sooner with better sensitivity and only when the pack is under load. Let off the accelerator and it would return to normal. As it gets worse more led would light up under load. However a toasted cell would light up most of one side.

I've looked around for an led driver to do such but haven't really found the answer.
 

·
Registered
Joined
·
2,217 Posts
Discussion Starter · #11 · (Edited)
I think a display would be great as well. The analog gauge is easy to set up. Using an IC to do this is more difficult because most of the chips cannot take anywhere near pack voltage. The clamping needs to be good while not interfering with measurement.

The simple on/off single output I was drawing was intended as a "do something" output. The kind of thing that might shut off a charger. The output on the one I drew should alarm at about +/- 0.5 volt (a single cell going up or down by a half volt should move the center tap voltage a quarter volt off from half the pack voltage.)
 

·
Registered
Joined
·
1,900 Posts
I like the meter, but that doesn't fit with the Buggy. I would like a system with a digital output, if you will. That is a normally closed connection isolated from the traction pack (like what is typically seen with cell monitors) when all is good and having that open when the pack half difference exceeds a set value. This could be used to drive one of Dimitri's head end boards just like is currently done with cell monitors.

I see this type of monitor as a substitute for cell level monitoring (or possibly a backup to it when cell boards have load resistors.) I really think that first the charger and controller should be programmed to take care of the pack (not rely on the BMS.) Then the battery monitoring system should provide an on/off output to insist a charge be stopped or drive be cut short. I don't want to wait until I teakettle a Lithium cell to find out something went wrong.

Code:
                pack center           
P                   |                p
a           D1 //   |                a
c           --->|---)-------         c
k     R1    |       |      |  R2     k
+ ----^^^------>|------>|----^^^---- -
               \\      \\             
            O1 ___     ___ O2         
               / \     / \            
              |   |----)--|           
              |--------|  |           
         R3   |           |           
+12 -----^^^--|           |           
              | Z1        |           
output  -----|< -------------- -12
Pack +, center, and - should be self explanatory. Remember that the pack + and - connections needs fuses at the cell terminals.

R1 and R2 are precision resistors of the same value to allow the needed current to flow, perhaps 10 milliamps at 2.5 vpc. For a 40 cell pack that would be 5000 ohms each.

O1 and O2 are a pair of optical isolators, transistor type. The IR LEDs are part of the pack circuit and the isolated photo transistors are part of the vehicle 12 volt system.

D1 is a red LED with a turn on point about 1/4 volt higher than the IR LEDs. If any one cell drops or raises 1/2 outside of its neighbors one of the photo transistors will be turned on.

Z1 is a TLV431 programmable zener. R3 keeps Z1 on unless one of the photo transistors turns on and grounds Z1 turning it off. This can be used to turn off a common DPDT PC board relay. This provides useful outputs to turn down a controller or turn off a charger or just scream warning alarms or whatever you think a battery monitor should do.

This is so quick and so crude I'm sure it won't quite work yet. Tear it apart please. :)
It's time for me to build some variant of this now. I'm off to a good start, I have a wire connected to the center of my new pack.:D

With this circuit I think I'm in the neighborhood of what I want to do. It's been over a decade since I worked in electronics daily and I'm unbelievably inept now compared to when I worked with op-amps frequently. How can we connect a comparator in this this circuit in place of the transistors? Or can the transistors be biased so that they turn on in a linear fashion as the voltage imbalance grows?

As I said earlier, I'd like to light an led ie for each volt or two of imbalance. For example at 1V resolution, a 3V variation would light 3 led's. The Evision meter does that so it can be done, somehow. The Evision is just so frigging expensive, $1000 or so.
 

·
Registered
Joined
·
1,900 Posts
I wrote Lee last night and got a reply this morning. Here's the text as an FYI.
"3/31/2011 7:19 PM:
Hi Lee, I hope this finds you well.
Very well, thank you. And yourself?

I got your email from diyelectriccar.com.
I figured I'd contact you about what I'm looking to build... I'm hoping you can help with this.

We're kicking around ideas and variants over there to your original
led circuit. Metricmind has something like that which has a bank of
led's for each half of a pack. Ie one led may represent 1V imbalance
etc but their product is about $1000!


Yes, I have one of his EVisions. Nice, but pricey! Also, it is really more of a tool for an engineer than a useful gauge for driving.
This way with more resolution, more data can be derived with more led's. I'm thinking it can be done somehow with a comparator circuit but I know there's issues with high pack voltages and op amps. I'm thinking a voltage divider or such may be the answer.
Here's some background on the Batt-Bridge that may help.

I've been tinkering with EVs for over 30 years. I've lost count of the number of different gauges I've tried in my EVs. I'm an engineer, so most of my gauges read out what I thought was useful information; voltages, currents, watts, amphours, temperatures, etc. They are useful for designing and testing an EV.

But I found they weren't very useful for *driving* an EV. For example, the driver doesn't want to know the pack voltage or amphours in the pack; he wants to know how many miles he can still go (will he make it home), or if there is some problem with the battery pack that requires immediate attention.

A true "fuel" gauge for an EV is a hard problem. There have been many attempts, and almost all of them fail. Most are about as useful for telling how much farther you can drive as a normal car's gas gauge is for telling you how many gallons are in the tank. I.e. they are both "guess gauges".

Since most EVs lack any real driving gauges, it is very common for drivers to murder their batteries. They just keep driving until it won't go any more. This runs at least one of their poor batteries completely dead, which destroys it. They might have a voltmeter that says the pack is at half voltage, but there was no indicator that says "Stop, you idiot!"

The Batt-Bridge was an attempt to provide an "idiot light". When green, it means OK you have power. When it turns red, it means "STOP WHAT YOU'RE DOING NOW!"

The exact voltage differential between the half-packs is not very significant. It doesn't carry any useful information. It is sufficient to know that something has caused 1v to 2v difference between the two half-packs.

If you want to watch the degree of imbalance between the two halves of the pack, it's easy to replace the LEDs with a voltmeter. Connect two identical value resistors in series across the whole pack, and connect your voltmeter between the center tap of the pack and the center tap between the two resistors. You can watch how it changes as you drive. It normally stays pretty close to zero, and only increases dramatically when something goes wrong (overcharging, or too deep a discharge, or too high a discharge current, or a loose connection somewhere).

Obviously, you can make the circuit as complicated as you want, with a precision digital readout, or a bargraph of LEDs showing +/-X volts. But I don't think this will add to the usefulness of the gauge."


The bold above is what I'm looking at. Infinitely easier yet gives me actual voltage difference between halves. The polarity will reference which side the issue is on. I have a digital volt meter good to 200V but not sure it will display both polarities.
 

·
Registered
Joined
·
2,229 Posts
I've been using the ammeter version for charge monitoring this past week. works well. Two resistors , a pot and one 100-0-100uA ammeter in a box.
 

·
Registered
Joined
·
1,900 Posts
1 - 20 of 54 Posts
Top