Thunder Sky + Fechter/Goodrum BMS + PakTrakr + Zaurus = ???

[AndyH]

This morning's quiz is: What do you get when you combine a 21-cell pack of 60Ah Thunder Sky cells, a Fechter/Goodrum BMS kit, a PakTrakr, a Sharp Zaurus, and an assortment of wire, solder, and shrink tubing?

Geek Art! - a fusion of solder, flux, matches on shrink tube, colorful squigglies, lithium, and Linux!

And hopefully an interesting view of what a BMS keeps from happening...

It started with this on one table:

bms.jpg (73.18 KiB) Viewed 2372 times

And this on another table:

lotsa_wires.jpg (96.93 KiB) Viewed 2369 times

The first one became this:

Complete_BMS.jpg (59.93 KiB) Viewed 2362 times

While the other, with the addition of these:

connectors.jpg (51.2 KiB) Viewed 2360 times

Became this:

fuse_blocks_installed.jpg (83.92 KiB) Viewed 2359 times

[AndyH]

... On to stage Two ...

We turned this:

bagged_paktrakr.jpg (54.92 KiB) Viewed 2339 times

Into This:

paktrakr_connections.jpg (66.71 KiB) Viewed 2336 times

Sorta Like This:

wiring_plan.jpg (37.3 KiB) Viewed 2331 times

Under the watchful eyes of her:

boss.jpg (44.53 KiB) Viewed 2331 times

Then we remembered where we hid this:

data_logger1.jpg (32.77 KiB) Viewed 2330 times

And made a serial cable with the magic null-modem bits and jumpers - and got this:

flowing_data.jpg (56.27 KiB) Viewed 2325 times

And that's when strange things started to happen...

[pgt400]

Andy, you do nice work! Question, did you have to modify the BMS to match the TS stated low and high voltage points (2.5 low, 4.2 high) ??

[GGoodrum]

Andy, you do nice work! Question, did you have to modify the BMS to match the TS stated low and high voltage points (2.5 low, 4.2 high) ??

These TS cells are LiFePO4-based, and like all LiFePO4 cells, has a CC/CV crossover point of 3.65-3.70V. No BMS changes are required.

Andy, I've not seen this PakTraker before. What exactly does it do?

-- Gary

[AndyH]

The cells were charged with the T-S charger once but were not balanced. The first 3-mile ride with the default PakTrakr's 1-second data rate gave us a look at the 60Ah cells at rest and at about 1.6C spikes. The controller is set to 100A max and it worked pretty well - we saw a couple of 113A spikes that were quickly pulled back to about 98A.

(The PakTrakr's lithium sensors report voltage x100 and current x10.)

first_ride.jpg (78.37 KiB) Viewed 2268 times

Here's the T-S charge profile. I've confirmed so far that the charger will run for more than 8 hours and will continue the 2A pulses.

TS_charging_curve.jpg (21.64 KiB) Viewed 2265 times

Here's what charging looked like for this pack:

Initial Charge with Current2.jpg (38.07 KiB) Viewed 2268 times

The main charge is over and now it's doing the 2A 'charge blocks'. These cells can go up to 4.3V, but I want to keep them below 3.8V if possible. There's a weak teal line at 3.7V for reference. Notice how the pulse pushes the nearly charged cells into the low 4s but doesn't appear to do anything to the low cells. Notice also the gentle curve as voltage drops off after the charger stops pushing electrons.

No_BMS_3.jpg (38.67 KiB) Viewed 2263 times

[pgt400]

But they do state a low cut off of 2.5v not 2.1? Are they being overly conservative?

Andy, you do nice work! Question, did you have to modify the BMS to match the TS stated low and high voltage points (2.5 low, 4.2 high) ??

These TS cells are LiFePO4-based, and like all LiFePO4 cells, has a CC/CV crossover point of 3.65-3.70V. No BMS changes are required.

Andy, I've not seen this PakTraker before. What exactly does it do?

-- Gary

[AndyH]

The plan was to connect the BMS into the 'daisy chain' so it's time to see if it works. The board was dumped into the bin and plugged in and the shunts immediately pulled the high cells back down to earth. So far so good!

bms_take_1.jpg (72.14 KiB) Viewed 2242 times

At this point, some of the cells in the pack were fairly well charged but more than 1/2 the pack was still pretty low. Pack voltage was about 73V and the charger stops at 75. The charger seems to have more brains than I did at 2:45 in the morning, so the best it would do was settle into it's 'cell balancing' pulse process. As we've already seen, without a BMS each pulse pushes charged cells to and above 4V while not appearing to do much for the low cells. I expect they're charging...slowly...but they're not making the lines move - we have to fix that!

The charger ran for just over 8 hours. I'll break the pictures up a bit so we can see what's happening.

(Remember that the PakTrakr doesn't log decimal points - current is displayed 'times 10'; voltage 'times 100'.)

overnight_charge_current.jpg (48.03 KiB) Viewed 2240 times

overnight_pack_Voltage.jpg (36.67 KiB) Viewed 2241 times

overnight_Remote_C.jpg (46.47 KiB) Viewed 2239 times

overnight_Remote_B.jpg (71.89 KiB) Viewed 2239 times

Finally some signs of life! Apparently, the T-S charger is programmed to just keep pulsing until eventually the low cells collect enough charge to move up.

overnight_Remote_A.jpg (78.27 KiB) Viewed 2236 times

close_pulses.jpg (45.66 KiB) Viewed 2228 times

[AndyH]

Andy, you do nice work! Question, did you have to modify the BMS to match the TS stated low and high voltage points (2.5 low, 4.2 high) ??

Thanks! No BMS mods yet. This is a V2.2 board with 1/2A shunts.

Thunder Sky's battery manual (http://www.thunder-sky.com/pdf/2007030222.pdf) says the cells can range from 2.0 to 4.3V. For 2000 charges, they say to keep things between 2.5 and 4.3. So far, it appears that the BMS will keep things below 3.9. The controller has a minimum pack voltage of 52.5V - 2.5V per cell. If the pack stays fairly balanced (once we get it there...), and with the maximum 1.6C discharges, I expect to hit the controller's LVC more often than hitting the BMS LVC.

That's the plan, anyway...we'll see if that's the way it happens in the real world. The next BMS version might be reset to 2.5V and will probably be built for a 1A shunt.

[AndyH]

...Andy, I've not seen this PakTraker before. What exactly does it do?

-- Gary

Hi Gary,

The PakTrakr is a cell monitor that has a few alarm functions and spits out a serial data stream. http://www.paktrakr.com/

http://www.eghia.com/Bloghia/Entries/20 ... _Hell.html and http://www.evconvert.com/article/dougs- ... ith-zaurus is my longer-term plan for displaying the data.

Andy

[fechter]

Cool.

What's the time base on those charging graphs? I'm trying to get an idea for the time between pulses there (or frequency).

It might take a really long time to bring the lowest cells up the first time. You could speed things by using a single cell charger on the low ones.

Once they all get equalized, the equalization time should drop quite a bit on the next cycle.

[AndyH]

Time base? Interesting question!

The PakTrakr puts out one stream of data per second. The 'pulse charts' for the 'overnight...' charges are 8 hours and 5 minutes long. Yeah - 29K seconds.

Here are a couple of pulses:

pulses.jpg (41.31 KiB) Viewed 2116 times

And one pulse. There's no data smoothing, and data is reported one point per second.

pulse_detail.jpg (28.55 KiB) Viewed 2114 times

[fechter]

Interesting. If you looked at the voltage with an oscilloscope, my guess is it is actually oscillating at a much higher frequency. The PackTrakr samples once a second, so once in a while it will coincide with a peak and generate an artifact.

With large cells, depending on the charger voltage, you can sometimes see the orange LEDs flicker. With mine, the flicker is too fast to see and just looks steady.

The actual cell voltage when balancing is very 'noisy' and generally causes inaccurate readings with a digital voltmeter. If you stop the charging current, you can get an accurate reading.

[AndyH]

Fechter - I think the charger's pulse is more like 2 minutes 'wide'. I'll verify that next time I'm sitting next to the charger.

Should/will/is the waveform from the 3.65-3.72V 'shunt turn-on' point be oscillating? Any thoughts on what the frequency range might be?

[fechter]

The larger the cell, the slower it will oscillate. 2 min would be much slower than my 10ah pack runs. With no cells attached, the upper limit on the frequency will be around 10khz.
I don't think the frequency is too important as long as it oscillates.

The maximum charger current will affect the speed as well. A higher charger current will pump the cells up faster and they will take longer to bleed back down. Is your charger running at 10 amps?

When the charge control circuit is on and you've reached threshold for at least one cell, the main LED should switch from red to green. When it's oscillating too fast to see, it just looks orangish. At a slower speed, you should be able to see it switch back and forth from red to green. So you're saying this is happening at about a 2 min interval?

Your graphs look much like the oscilloscope traces I saw on my pack, other than the speed.

An interesting observation from your discharge curves is that under high load, the cell voltage sags considerably. This is why we think the lower LVC voltage on the BMS will be fine. Once the load drops, the cell voltage bounces back up into the recommended range.

[AndyH]

The charger will push 75V @ 15A all-out, but it's only 2A during these 'balancing pulses'.

I ran another 'overnight' charge thru the BMS last night. I caught one of the pulses this morning - 17 VERY bright orange LEDs and a fair amount of green on the 'charge control' LED. I don't think the charger to give me enough power to latch the BMS, but some of the pulses have pushed the pack over the 75V max listed on the charger.

I look forward to getting the pack balanced and taking another ride. I didn't expect to see quite as much sag from a 1.6C max discharge, but it's difficult to pull any one cell out of the 'colorful mess' of that discharge chart, and more than 1/2 the pack was at a much lower state of charge for that ride.