Using the Universal grid charger

[Mike Dabrowski 2000]

The sawtooth in the slope dif is not noise at all, it is samples / bit, if you watch the slope you will see that the sawtooth raises with each sample that does not cause a bit change until it finally changes a bit and catches up the slower the rate of change the bigger the difference.Once the bit changes, the dif drops, and begins waiting for the next bit change, which is why I have the average graphed as well.

[bluesight]

Mike,

The smoothed version is still pretty noisy (that is the graph I lost when my notebook locked up). The smoothing Tech variable (Smoothing) is set at 10, though I'm not sure what that does (10 point rolling average?). What are the dangers in increasing that variable? I was thinking about trying 20 for some test cycle.

Also, I put some recovered graphs and screen images on my previous post #629. The question I could really use feedback on is whether upping the 1A charger voltage stopping variable (Cut-backVP1) to 170V from 166V might cause a disaster? The overall graph seems to show room for about 4 volts more at the higher current, and might make the top-off part of the curve look a little better (constant climb rather than inflection climb) up to what looks like a 175V plateau. This may not be good for all packs, but might be OK for packs like mine where the graphs are pretty well defined.

B (s/n 0166)

PS...I tentatively set up to try Cut-backVP1 of 168...should be safe. I can change it back at later this eve if it turns out to be a really bad idea...

Originally Posted by Mike Dabrowski 2000

The sawtooth in the slope dif is not noise at all, it is samples / bit, if you watch the slope you will see that the sawtooth raises with each sample that does not cause a bit change until it finally changes a bit and catches up the slower the rate of change the bigger the difference.Once the bit changes, the dif drops, and begins waiting for the next bit change, which is why I have the average graphed as well.

[Mike Dabrowski 2000]

If I am in a hurry, I have set the cutback higher than the pack will ever reach, and let the temp dif stop the 1A charge , probably not a good thing on an unbalanced pack, but with fan running the 1A heat can easily be dissipated.


The smoothing is only for the battery V and current signals,
Smoothing + 10 samples average, new sample added to total 10 sample accumulator, average value subtracted. average of the total is displayed.

The difference trace will always make a sawtooth because of the algorithm.
The sawtooth is usually a small part of the slope level, and should not effect the detection.

[bluesight]

Looks like not so great news. It looks like I am getting the patterns for cell dropouts on the slope graph. They are almost impossible to detect on the voltage graph and don't look much like what was on the cell testing video. More like a 2 step down at the sample point rather than 1 step down. They also maybe weren't so easy to detect before because the SlopeDifMax of 30 stopped the charge when those happened in the past, though I do see one pretty early on this time. At the end, I got a code 2 at 138.2V, so still something with the slope graph, but not sure what the trigger was exactly. It looks like there might have been a sampling glitch just before the very end. The volts/slope graphs are not well aligned, so it is a little hard to see the correlations.

I don't know if it will be worthwhile to try the "test" method to bring the pack any further down. If there are dropping cells, I don't want to reverse them for too long.

We'll see how the charge cycle goes. I went back up to 170V for Cut-backVP1.

After this, I will likely try a 3 cycle full conditioning charge. Unless there are other recommendations for one more test cycle with different settings.

B

[bluesight]

Finished test cycle 2. Charge looks really good using 170V for Cut-backVP1. Seems pretty repeatable for the topping voltage around 174V. The 7000mah limit for MaxChg_mAH was a problem, so I will stick to something over that, maybe 8000, for future runs.

Here are the overall results in graphs and "picts", including the detail on the charge switchover. Since only 5 attachments are allowed, I'll focus on the discharge stuff in the next post.

B

[Mike Dabrowski 2000]

The cell that I purposely reversed charged at 1A for 14 hours and then put in series with a non reversed cell of the same initial condition is still going strong after 40 or more15-20A discharges. The other way to look at it is that the cell dropping out is the lowest capacity cell in the pack, you will probably want to replace it anyways.
I would be running the disabled discharge stop test on the bench with the pack ends exposed so we could see if this is one stick, or if the whole pack is not far behind that stick as shown in the video.

[Mike Dabrowski 2000]

You may want to download the new V 2.6 of the datalogger, as it has the battery V and slope/temp graphs lined up so you can see the relationships better.

[bluesight]

I am still skeptical about what the discharge cycle is telling me. Looking at the slope curves, there are hints of cells dropping. But there is virtually nothing in the voltage graph that confirms that. It looks more like high sensitivity to sampling noise than real dropouts. But I still don't know what to look for.

I also still cannot quite see what caused the discharge to stop. The slope started to steepen, but the actual voltage sampling still was looking OK. The slope data did not appear to reach my setting of 50, so maybe the DischSmpFac setting is coming back in. The effect of that setting and where it is used is still something I don't understand.

I will likely do one more test discharge (starting at 6 tonight) to hopefully see obvious cell dropouts or to get to 132V. Hopefully, I won't do damage with reversing cells. The basic attempt will be to further increase SlopeDifMax (to 80), further decrease DischSmpFac (to 5) and try Smoothing at 23 (just to see what it looks like). If that doesn't automatically stop at 132V (while carefully watching for obvious cell drops), I will continue the discharge using the Tech Test mode to get to 132V (or cell drops). After that, I will stop instead of continuing into the charge part of the cycle, then restart a 3 cycle conditioning, beginning with a Soak charge. I'll put that into Bank 0 as my new baseline.

Should be interesting...

B (S/N 0166)

PS..I tried my darndest to correlate the horizontal axes on the graphs. It would be nice to get the horizontal scales to be the same so slope/temp can be directly compared with voltage. This method kind of works, but isn't exact. I guess to be accurate it would be better to use the actual csv data, but the slope info would need to be derived in the same way as with the charger, so more steps would be needed.

[bluesight]

That is the version I am running (top of screen says Genesis One Data Logger V 2_6.vi), and I know you mentioned that was a feature of this version, so something isn't quite working right. I can try a reinstall at an appropriate time.

B

Originally Posted by Mike Dabrowski 2000

You may want to download the new V 2.6 of the datalogger, as it has the battery V and slope/temp graphs lined up so you can see the relationships better.

[Mike Dabrowski 2000]

This is what V 2.6 graph screen should look like