Monday, July 25, 2011

Battery Pack Balance Monitor

It has been 19 months since I installed my LiFePO4 battery pack made up of 40 (originally 36) Thunder Sky TS-LFP100AHA cells in a 2p20s arrangement for a nominal pack voltage of 64V and an energy capacity of 12.8kWh. From January 2010 until the end of July 2010 I ran with 18 cell pairs and top balanced the pack at 4.00vpc. At this point I received my reprogrammed charger back from Elcon set for a saturation voltage of 69.3V which, until recently, I had bumped up to 69.7V. At this point I also installed the remaining 4 cells for a total of 20 cell pairs. I also did my final top balance at this point in time since I had noted that there was very little if any balancing going on and that it wasn't the same cells which hit top voltage first. I left the Black Sheep Technology BMS boards in place but they did not do any shunt balancing since my charge was ending at 3.485vpc and the boards don't shunt until 4.00V.

With the BMS boards, however, I had no way of knowing if any cell was going high or low relative to the others. When the BMS alarm would sound all I knew was that at least one cell was low. I wanted some way to narrow down if I had a weak cell or a bad connection. Lee Hart has posted on the EVDL Library a Batt-Bridge Battery Balance Alarm. This device used resistors selected for a particular pack voltage and a series of LEDs. I emailed Lee about some alternatives to his circuit. I wanted to have some sort of meter which would deflect to show me what was going on rather than wait for a large enough voltage swing to occur to light one of the LEDs. You can read about my dialog with Lee and some other discussion on the DIYElectriccar site. What I ended up doing is ordering a 9V battery powered volt meter from eBay seller clinia. He will setup the meter to what ever specs you want. Here is a picture from his listing of the meter:
With a battery balance circuit such as this the voltage reading will be exactly half of what the difference is between the two halves. Because of this I had the meter setup to display a reading double what the input voltage was. Now I can glance at the meter and see the actual voltage difference between the most negative half of the pack and the most positive half of the pack.

The circuit is constructed by connecting a resistor between the most negative post of the battery pack, to one end of a potentiometer, and the other end of the potentiometer to another identical resistor and then to the most positive end of the pack. A volt meter is then connected between the middle post of the battery pack and the wiper on the potentiometer. The potentiometer is used to calibrate the device so that when each half of the battery pack is exactly the same voltage the meter will read zero. One of the challenges with hooking wiring up to high voltage DC in a vehicle is the potential for a short which could cause all kinds of damage, including fire. The challenge is then to find small, low amperage, high voltage fuses. Radio Shack fuses won't cut it. I then searched for name brand fuses and holders but didn't know that I wanted to have to mount them on PC boards and such. Next is to determine where and how to mount them. I then inquired Lee about using a resistor as a safety device. Since I wanted this device to draw as little power as possible I used resistors with a very high resistance. This meant that any short would cause only a very low current and wouldn't burn something. With this in mind I went about coming up with a way to mount the fuses right at the terminals so that no matter where any wire might short the current would be very low and not cause any problems. Naturally I wanted to secure things where there would be no shorts but I wanted to be safe.

In playing around with Ohm's Law, V=IR, and trying a couple of fuses I had laying around and my DMM, I determined that it really doesn't matter what resistance I use as long as it is high enough. I ended up getting 100K Ohm, 1/2 Watt, 5% resistors. Since the max my pack voltage would ever go is 80V, if I were to top balance at 4.00vpc, this means that the maximum current would be I=V/R=80V/100KOhm=0.0008A=0.8mA. This is only 0.064W so well under the resistor's rating. This is if one wire shorted between the extreme most posts which is highly unlikely to happen! For the potentiometer I picked up a 1KOhm 15 turn pot. This means that between the negative post and the most positive post would be a total of 201KOhm of resistance and with a fully charged pack at 70V (I'm really having charging end at closer to 69.3V now) the current draw of the device is under 0.35mA. This is quite acceptable and means that even if I left for a year and didn't charge that the pack would only be drained about 3Ah. Furthermore, this is evenly across the full pack so there will be no imbalance caused by the device. Perfect!

These resistors take care of protecting two wires but what about the center tap? What if it shorts to something? I played around with different resistors to see what difference they made to voltage readings. They do make some difference so I decided to go with a 10KOhm resistor on the center tap wire. This only changed the reading by about 0.01V which is acceptable to me. The most voltage the center tap could ever see was 40V so the current would be at most 4mA and 0.16W, again well under the 0.5W rating of the resistor.

For mounting I bought some double clad PC board material and cut out 25mm squares. At the moment I'm not using the BMS boards so I can use the screw mounts for them. If you have read earlier posts on the BMS boards you know I am using brass bolts with a tapped hole in the center. I drilled an off center hole for the screw to go through and then used a dremel tool to grind away a gap in the copper on both sides of the board. I used a #57 Wire Gauge drill bit to drill three holes, two for the resistor and one for the wire to attach. Below are the three boards I built. The two at the top are the most positive/negative boards with one showing the back side. The one marked with an M is for the center tap.
Terminal connections




The small unfilled hole is where I attached the wire to the board. Note that I soldered both sides of the board. I did this to provide as many path ways for conduction as I could to hopefully minimize the chance for a bad connection. I haven't done it but I probably should conformal coat the boards except for the solder pad for the screw. It was very easy to make the pad and draw the M. The copper is textured somewhat and is very ready for soldering.

The next thing to do was to decide how to connect everything. While I like the good connection achieved with soldering I didn't want such a permanent hookup since I would have to do the soldering on the vehicle or in the battery compartment. These Gizmo's don't have much extra room for things, especially after all the things I have added over the years. I decided to get a 4.5cm square PC Board from Radio Shack and mount the pot and some PCB terminals to it. The set of four terminals is for hooking up the battery connections (only 3 are used) and the double terminal is for hooking up the volt meter. This makes a simple and hopefully reliable connection point. Time will tell how well this will hold up.
Front view
Rear view

Except for the insulated jumper wire I just used the leads I cut off of the resistors to make the connections. For the wire connections from the terminals to this board I used 24 gauge aircraft wiring I got from Aircraft Spruce. The wire is tinned and rated to 600V. Most importantly the insulation is very tough. I accidentally set some on a hot soldering iron and it barely made a mark on it! The wire is rated to 150°C which explains the minor damage from touching the soldering iron.

Next came the need to mount the volt meter. Since a Gizmo dash is curved it is very difficult to mount flat meters. I also don't want big holes in the dash, especially if I change something in the future. I decided to mount this meter the same way I mounted the CycleAnalyst, by carving the back of an ABS project box to match the contour of the dash and then silicon the unit to the dash. I first needed a pattern to go by so I used some air dry clay as I did before. I used a 4"x2"x1" Radio Shack project box wrapped by a strip from a cereal box and pressed the clay onto the dash and then let it dry.
Unfortunately the dry, sunny, warm days we were supposed to get turned into cool wet days so it took a while to dry. After a couple of days I finally carved out the center and then used a hair dryer to finish the drying so I could remove it from the dash. Here is the result:

I carved the edges so I could place this inside the project box. After cutting the back off the project box I put this inside and held the sides firmly while I used my Dremel tool to cut the project box to match the contour.

While the clay was drying I took the screws which hold the lid to the project box in place and screwed them into a piece of cardboard. I sprayed them with metal etching primer and then sprayed them with a couple coats of gloss black paint. This seems to work well and the paint doesn't chip when inserting the screws. I did have to cut the screws shorter since they would have hit the dash otherwise. On the thinnest corner of the box I reinforced the screw hole by gluing in a piece of the ABS box I cut out between the screw hole cylinder and the wall of the box. Regular ABS glue for black sewer pipe works great for this. Without the reinforcement the screw mount would tend to twist and deform the edge of the box.

To mount the display I placed the box cover face down on a piece of #2 plastic scrap, scored the opening I wanted with a small tipped Xacto knife and then went to work repeatedly running the knife down the groves until I cut through the front. Working from the back makes sure that any slips don't show through on the front. The hole was then filed with a standard 8" mill file until it was smooth and the correct size to hold the bezel. This part took me about an hour to do. Here is the finished result:
The ammeter really doesn't look like it is crooked. When sitting in the seat it all does look just fine. I have the meter hooked up so that a positive display represents how low the most positive part of the pack is from the most negative part and a negative reading is the opposite. I'm still trying to decide how I want to label the meter so that my wife and any one else who drives my Gizmo will know how to interpret the reading. As you can see the front (most positive) half of the pack is 0.02V lower than the back (most negative) half. When I did my last pack balance I had to do each half separately. I believe this is why there is a difference. However, under a >1.5CA load the difference is +0.13V and when regen current is over 50A I get a -0.00 reading so I may have a poor connection or weak cell pair in the front half of the pack.

Here is a parts list:

  • Low Cost LCD Volt meter ($16.21)
  • 4"x2"x1" Project Box ($3.19)
  • 2-sided PC Board material ($4.19)
  • 45mm square PC board ($2.19)
  • 2 packages of 2-pin 5mm PCB Terminals ($2.39 each)
  • 5 pack of 10KOhm resistors ($1.19)
  • 5 pack of 100KOhm resistors ($1.19)
  • 1KOhm 15-turn potentiometer ($3.19)
  • #57 Wire Gauge Drill bit ($2.89)
  • 24 gauge wire ~ 25 feet total (~$6.00)
  • Air Dry clay ($???)
  • 9-volt battery (from my smoke detector) [Note: Since I don't turn off the meter with the key the batteries were going dead too fast so I installed a 12V-9V isolated DC-DC. See the post at http://2003gizmo.blogspot.com/2011/10/battery-balance-9v-battery-replacement.html]
  • miscellaneous zip ties (had on hand)
  • a donation to Lee Hart for all his help. Have you benefited from his help? Send him a Thank you donation for the Sunrise EV2 project via paypal to leeahart_at_earthlink.net
I had the wire, clay, and zip-ties in stock and the 9V battery is from one of my smoke detectors. The meter will run on as low as 7V or so and has a very low drain. I decided it wasn't worth trying to get an isolated power supply for it. I just zip tied it to a small shelf under the dash. I put a piece of weather stripping under it to keep it from sliding out and it should be good to go for several months. The total cost to me for the materials I didn't have on hand was $33.44 excluding tax. For paying all retail pricing I'd say this isn't too bad.

Now if I have a cell going bad I'll see the voltage difference show up right away unless, of course, I have a cell in each half of the pack doing the same thing at the same time.

In my next blog post I plan on relaying the results of not doing any pack balancing for the past 11 months.

Edit: I remembered to take some pictures the other day when I was checking pack balance. Below are different views of the batt-bridge installation. The pack really looks dirty in these pics!
Main Board

Main Board and Center tap

Negative Terminal connection

Positive Terminal connection