AA fast chargers - Delta V?

[SilverFox]

Hello Mr Happy,

Excellent work.

I think it would be interesting to see what happens with the same cell charged at around 400 mA...

Also, you may find that you do get a -dV termination when charging Eneloop cells at 1C.

Tom

 

[altis]

Perhaps the Energizer 15-minute charger uses the inflection point to reduce its charge current - as could be suggested by this post:

https://www.candlepowerforums.com/posts/2306504#post1719120

 

[Mr Happy]

Well, curiosity got the better of me again, so following on from my previous experiment with a cheap brand of NiMH cell I did the same experiment with an Eneloop. This is a cell with a previously tested actual capacity of about 1900 mAh.

The test protocol was the same as before: charge at 1600 mA and record voltage, charge, and temperature for later analysis.

So here is the result of experiment 1, using the automatic charging mode of the MH-C9000:

The first point of interest is that the cell didn't come close to the 0 ΔV point before the C9000 stopped charging. This was apparently due to the maximum voltage safety termination that the C9000 applies at 1.47 V. The second point of interest is that the cell remained quite cool at a maximum 34°C throughout the charge. This contrasts significantly with the previous cheap cell which got steadily hotter throughout the charge.

Charging stopped after 72 minutes with a supplied charge of 1754 mAh. A subsequent discharge at 500 mA give a measurement of 1763 mAh, so the charging efficiency was essentially 100%. Note how this compares with the 80% measured for the cheap cell.

Since the C9000 kindly stopped charging before the inflection point or the −ΔV point was reached, it became necessary to perform experiment 2. This was the same as experiment 1, charging at 1600 mA, but with manual charge termination instead of automatic:

(I have removed the experimental data points from the chart for clarity and have just shown the trend lines.)

This time the voltage reached a maximum of 1.54 V and the cell temperature started climbing rapidly. I pulled the plug when the temperature reached 40°C (poor Eneloop).

With this new test it is possible to look at the slope of the voltage curve and find the inflection point, which occurred at just a little over 1800 mAh:

Finally, we can examine the end of charge conditions in more detail:

The C9000 would have terminated at 1.47 V and 1720 mAh, working out at 1720/1900 = 90% of a full charge and an end point temperature of 34°C. The inflection point algorithm would have terminated just past 1800 mAh, this resulting in 95% of a full charge and much the same temperature of 34°C. Lastly the −ΔV signal would have been detected at about 1960 mAh and 38°C. Presumably this would have resulted in a 100% complete charge. I'm discharging the cell at the moment to see what charge it actually retained. (Edit: the discharge test at 500 mA showed 1834 mAh.)

I think the conclusion from testing the Eneloop is that it clearly demonstrates a much better level of performance than the cheap cell. It works at a higher voltage (the high charging voltage is reflected in higher discharge voltages), it accepts charge more efficiently, and it remains much cooler during charging.

Secondly, there is the interesting result that using the C9000 to charge Eneloops gives an early, low temperature charge termination that approximates what the inflection test would do. It means that to get a 100% charge you have to leave the cells on the charger for two hours of top-off charging (100 mA x 2 hours = 200 mAh, 1700 mAh + 200 mAh = 1900 mAh). On the other hand, the early termination will increase the cycle life of the cells.

 

[shadowjk]

I would have thought that a higher voltage during charging meant higher internal resistance?
After looking at voltages of AA I've charged at 1000mA rate on BC900, I found the 1.47V cutoff a bit odd...

 

[Mr Happy]

I would have thought that a higher voltage during charging meant higher internal resistance?

If you were comparing the same cell at different charging rates this would likely be true. However when comparing two cells of different provenance, the difference in voltage is almost entirely due to differences in cell construction and chemistry. Not all NiMH cells are the same, and there are very observable differences in operating voltage between different brands and cell types.

After looking at voltages of AA I've charged at 1000mA rate on BC900, I found the 1.47V cutoff a bit odd...

You are right about this. A limit of 1.47 V is fine for cells of older vintage like the first one I tested, where the zero dV/dt point usually occurs at somewhat lower voltages. But for newer and higher quality cells, the zero dV/dt point seems to occur at voltages above 1.47 V. Among cells I've tested this is true of the Eneloop, the Uniross Hybrio and the Duracell 1700. It's not all bad though, since the cutoff occurs pretty close to a full charge and it certainly avoids the cells getting hot at all.

 

[wptski]

So where are the inflexion point chargers? Nothing mentioned in their paper about percentage of charge. Looks be very close to the dT/dt method of termination.

 

[Mr Happy]

So where are the inflexion point chargers? Nothing mentioned in their paper about percentage of charge. Looks be very close to the dT/dt method of termination.

Yes, this is a good question.

I think it may come down to engineering conservatism and perceived market value. To make a consumer charger using the inflexion point method would require extra design, prototyping, testing and risk analysis. This would have to be weighed against the potential market benefit from such a design -- would it reduce the price the charger could be sold at, or would the average consumer recognize a benefit worth paying more for?

Now for a specialist niche product such as the Maha C9000, maybe there is a possible market. We can perhaps hope that someone out there reads this thread and considers some of the questions raised here.

 

[Mr Happy]

I think it would be interesting to see what happens with the same cell charged at around 400 mA...

I agree it would be interesting to do, but at that current it would take 4-5 hours for charging to complete. I just can't see me watching it and writing down measurements for that length of time, unfortunately. Maybe if I ever have access to automatic data logging equipment...

 

[NiOOH]

Excellent work Mr Happy. This exactly matches my results with recent Maha C9000 chargers. The C9000 terminates on maxV of 1.47 V/cell. Towards the end of the top-off stage the voltage may climb to 1.48-1.49 V on some cells.
I wrote about this some months ago, but noone here believed it.

 

[SilverFox]

Hello NiOOH,

I remember you bringing this up, and I think I gave you a bad time about it. It looks like your observations were correct.

I have seen termination voltages above and below 1.47 volts, but there does seem to be a lot of charges that end right at 1.47 volts.

Tom

 

[MrAl]

Hi there,


A few quick points:

First, that article Tom linked is very interesting and informative. I did find
that their terminology is a little strange though when it came to what they
were calling "Inflexion Point", which is really the calculus "Inflection Point".
If you ignore the rather strange spelling, they are both the same.
And just to clear things up a bit, the inflection point detection when it
comes to charging cells like NiMH is very much the same as zero voltage
change (or zero slope) detection. It's the point where the slope either
changes sign or is just about to change sign, or viewed another way,
it's the point where the voltage has been rising and is now about to start
decreasing. The difference between true zero voltage and inflection point
detection methods is that the true zero voltage detection does not require
the voltage to actually start to fall, whereas the inflection point might, or
might pick up the voltage starting to rise again, which could also trigger
an inflection point detection. It largely depends on how the manufacture
pre-filters the signal first though. There are many ways to do this and this
changes the overall response to some degree too. It also depends greatly
on the manufacturers algorithm, which can be better or worse.

Second, the difference in pre-filtering also affects the minus delta
detection too, and doing this wrong could lead to overcharging also,
which is then blamed on the minus delta technique itself. The minus
delta technique isnt as bad as it's made out to be on some web sites,
and the proof is in the monitoring of the voltage *and* temperature
of several cells of different manufacturers and of different cycle ages.
What happens is that as the voltage rises, the temperature does too,
and since it's the temperature (times the time) that does the most damage
the shorter the time the cell is subjected to a higher than normal
temperature the longer the cell will last, but for short temperature
vs time increases not that much extra damage will be done to the cell.
This means that a small time addition where the temperature is higher,
although we want to avoid this, wont do that much extra damage to a cell.
I does matter however how the relatively noisy signal is filtered, and
that makes a big difference, and every voltage measurement has to be
pre-filtered even if that filtering is indirect (part of the algorithm).

Another thing (#3) to think about is that when comparing zero voltage or
inflection point detection to minus delta V detection one has to also take
into consideration the effect on cell capacity after a typical charge cycle.
Many cells capacities are defined by using the minus delta V technique,
so avoiding that method voids the cell capacity rating. Of course it's up
to the individual here to make the choice between highest cell capacity
and shorter charge time. For example, with my Li-ion cells i almost
always undercharge them just a little (10 to 20 percent) so i get longer
cell life.

 

[Mr Happy]

Hi MrAl,

Thanks for your comments.

Actually "inflexion" is an alternative spelling of "inflection" and both are the same word. In the past "inflexion" was the preferred spelling (and language purists would argue it still is), however in modern times "inflection" has become the norm.

If you examine my chart titled 'End of Charge Detail', you can see the inflection point where the purple arrow is, and the zero voltage slope where the red arrow is. The inflection point comes before the zero slope, and also before the temperature has started on its rapid rise.

An interesting point for me is that the particular cell I tested does not seem able to hold a charge any greater than 1830 mAh, so the inflection point essentially marks the point of 100% full charge. Continuing to the zero slope point does not add any more charge to the cell, and in fact is just overcharging it while heating it up. One might as well stop at the inflection point and consider the charge complete.

 

[wptski]

If the inflexion/inflection point is a 100% charge, it contradicts everything else that many of us ever read. I agree that this point occurs before the zero voltage and even before dT/dt.

This document was dated in 1994, so that means that they are refering to early Ni-MH cells. There were a few chargers that used dT/dt termination but they never worked well, so it would seem like using temperature input would be tricky.

 

[Mr Happy]

If the inflexion/inflection point is a 100% charge, it contradicts everything else that many of us ever read.

I'm only saying that it appears to be so for the Eneloop cell on test for this experiment. Eneloops are a new technology and they seem to perform better than previous generations of NiMH cell.

Here is the evidence for my statement:

When the C9000 terminated automatically at 1.47 V the charge input to the cell was 1754 mAh and a discharge test at 500 mA gave 1763 mAh, so the charge acceptance up to this point was 100%. In the second test when I terminated charging manually, the charge input to the cell at the −ΔV signal was about 1970 mAh and a discharge test at 500 mA gave 1834 mAh. This is an average charge acceptance over the whole period of 93%. If we assume the charge acceptance had again been 100% up to the 1700 mAh point, then the charge acceptance over the last part from 1700 mAh to 1970 mAh was 50% at best.

My observation is that the 100% cell capacity is about 1830 mAh, and the inflection point occurred at 1810 mAh. Making the presumption that the charge acceptance remained high up to this point, the cell was essentially fully charged at that time and further high rate charging was of little value.

I believe that if you want your cells to be fully 100% charged to the max, then the best approach is to stop the high rate charging before the rapid temperature rise begins and move to a lower top-off rate to complete the charge. The inflection point does look like a good predictor of the rapid temperature rise since it captured this point for a cheap Chinese cell of unknown manufacture and also for a modern high performing Eneloop.

 

[eluminator]

I guess the C. Crane NiMH and NiCad charger uses "inflection". The Saitek manual says it uses Negative Delta V and Delta V Square over Delta t Square. Whatever it uses, it works. You can charge anything from one AAA cell to 4 D cells and it always gets it right. You can charge NiMH or NiCad. The only restriction is that all cells charged together should be the same. I think the negative delta V is only used to switch from top-off to trickle.

I've had a couple of these chargers for 5 or 6 years and it's the only charger I use. I like the fact that it analyzes the cells, and I always do that before I charge them. The LCD display shows what's going on when charging or discharging on analyzing. Actually it's a good idea to test the cells before charging in case the cells have different voltages. Because it charges cells in parallel, the voltage of each cell should be similar.

This charger has "soft start" so it will charge cells no matter how much they are discharged. They claim it also prevents the generation of damaging heat. The charger also has "negative pulse" which they claim prevents crystallisation and reduces gas bubbles which increases charging efficienty.

The charger has four phases. Soft start, fast charge, top off, and trickle. None of these phases are timed and I don't think they depend on absolute voltage either. They are all based on feedback from the battery. When I charge different size cells or cells with a different initial charge, the duration of each phase (except trickle) is different and the amount of charging current during the various phases is different also. You can't fool this charger. No matter what I put in it, it handles the situation. Before it starts charging, it analyzes the cells for ten seconds and then starts the charge.

Here's the manual for the Saitek. It's no longer available as far as I can tell but the C. Crane is functionally equivalent except it doesn't charge 9 volt batteries.
http://www.hypercon.net/~blisscomm/Pictures/CPF/Saitek_CCrane_charger/Saitek_SmartCharger.pdf

 

[NiOOH]

Hello NiOOH,

I remember you bringing this up, and I think I gave you a bad time about it. It looks like your observations were correct.

I have seen termination voltages above and below 1.47 volts, but there does seem to be a lot of charges that end right at 1.47 volts.

Tom

Hello Tom.
Yes, I remember it also. In fact, it is possible to observe cut-off voltages belov 1.47 V. I have a set of cheap, low capacity NiMH cells that came with a multiband receiver. The celles are labeled at 1300 mAh. When charged at 1000 mA the voltage peaks at 1.45-1.46 V. In this case the -dV kicks off and the charger terminates at voltages lower than 1.47 V. On all other cases the termination occurs shortly after the display shows 1.47 V. During the top-off, the voltage could rise above 1.47, but I haven't seen terminating the fast charge above 1.47.

 

[Mr Happy]

Yes, I remember it also. In fact, it is possible to observe cut-off voltages belov 1.47 V. I have a set of cheap, low capacity NiMH cells that came with a multiband receiver. The celles are labeled at 1300 mAh. When charged at 1000 mA the voltage peaks at 1.45-1.46 V. In this case the -dV kicks off and the charger terminates at voltages lower than 1.47 V. On all other cases the termination occurs shortly after the display shows 1.47 V. During the top-off, the voltage could rise above 1.47, but I haven't seen terminating the fast charge above 1.47.

This is exactly the same as I observe. The cheap Chinese cells in my test above terminated on the -dV signal below 1.47 volts. However if you watch the charger with cells like the Eneloops, Done appears in the display as soon as the voltage rolls over 1.47. Following this the voltage may creep up to 1.48 or 1.49 during top off, but this is always after Done has appeared and never before.

I have version number 0G0D01. It is possible that the first firmware version of the C9000 is different in this regard to the later one.

 

[bob_ninja]

Wow, very nice work Mr. Al
Thank you very much

I remember the termination problems in early C9K versions and debates about difficulties in identifying a reliable end of charge signal, especially for older cells. I am only speculating here: Could this max voltage 1.47V end of charge trigger be the latest firmware update resulting from early problems? Perhaps they added this max voltage end of charge signal as a safer alternative to other methods to stop those runaway charges that kept pumping many Ah into cells. Just wondering.

The interesting outcome of stopping at 1.47V is that the old problem of cooking cells if now gone and C9K is actually cooler than BC900. Funny reversal.

After the inflection point temperature starts to rise much faster, so it is probably the best point to stop. However, there is only about 100 mAh between it and the 1.47V "signal", or 5.5% difference. The important point is that the remaining 200 mAh delivered by top off charge is done at a much lower rate (100 mA instead of 1600 mA in your example). Thus it still squeezes those last few mAh without heat buildup. I'd say that is a damn well way to ensure virtually 100% full charge without much complexity.

Also, I am not sure what is the point of trying to squeeze in every last mAh anyway as SD is highest at 100% SOC, even for LSDs. The car batteries are charged only up to 80% SOC for longevity. In this instance, the original 1.47V charge termination resulted in 94% SOC which is pretty close.

Once again I am impressed by this charger. Great stuff.

 

[Mr Happy]

Wow, very nice work Mr. Al

But...but...it was me that did the testing... :mecry:

Thank you very much

You are welcome. It was fascinating to look at the detail of exactly what happens with this charger.

 

[wptski]

Somewhere in the early C9000 threads is a post of mine with a temperature graph of two cells. One is charging inserted in the unit and the other is connected outside the unit. If I remember correctly, the cell outside the unit runs >20F cooler, so the built in PS is generating much of the cell's heat. So how to you enter that into the equation? Some slots run warmer than others too.