Optimal Charge Rate: new technique to find

[PeAK]

The start of this post came long ago when from this thread's post about missing terminations and hot batteries. Like a capacitor, the slower the charge rate (i.e. current), the slower the voltage changes. The "negative delta-V" based chargers rely on the slope information to decide when to stop charging. So after a bit of experimentation, I think I can explain why larger batteries charged at smaller currents have issues related to stopping charge...it is all tied up in their like behaviour to a capacitor.


Summary:
For 1000 mA pulsed current chargers (i.e. 0.5C) using "negative delta-V" termination, the rate of change of the battery voltage during the off/measuring portion (see second post) of the pulse should be at least 1mV/sec over twenty consecutive measurements and/or twenty seconds.

Background:
In another thread discussing/summarizing some of the issues with NiMh chargers/batteries, the charging profile (i.e. change of voltage with time during charging) of 2 different NiMh batteries was found to be quite different given the same charging current of 1600mA.

This lead me making some measurements of batteries that did not terminate and those that terminated well.

Method:

1. Partially discharge a fully charge battery for 1 minute at about 200mA. For a charger with a decent full charge detection circuit, the charger should terminate with about 3 minutes, max, with vibrant batteries.
2. Place battery in pulse charger and monitor the voltage with a digital multimeter that updates at a rate of at least 3 readings/second (most DMM).
3. The charger used has a period of 1 second pulse with 50% duty cycle at 2amps. This averages out to a current of about 1 amp The DMM will give 3 readings for every pulse of the charger. Note the lowest voltage and record.

Observations:

1. A 1600 mA-hr Failsafe NiMh battery was charged at 1000mA with the following charge profile:
   
   

   
   
   ​
2. In the charging profile, we see the battery voltage spike during the 2A current pulse. During the off interval when the current is zero, the voltage settles to a steady state value. Note this value and also the next settled value. This is shown as "delta V" in the graph. Record "delta_V" for at least 20 consecutive pulses.
3. 1600mA-hr Failsafe AA results:
   
   Notes: voltage increases monotonically
   delta_V: 1mV difference
   Termination: Termination in 3 minutes​
4. 2650mA-hr Energizer AA results:
   Notes: voltage differences not monotonic.
   Sometimes negative.
   delta_V: 0.25mV difference
   Termination: No Termination in 1/2 hour
5. 900mA-hr Sanyo AAA results @ 600mA
   Notes: voltage differences are sometimes negative
   delta_V: 0.5mV difference
   Termination: No Termination in 1/2 hour
6. 900mA-hr Sanyo AAA results @1000 mA
   Notes: voltage differences are lways increasing
   delta_V: > 1mV/sec
   Termination: In 3 minutes

Discussion:

The experiments did not have the availability of a programmable charger/discharger. A 4 ohm battery was used for discharge purposes by shorting the battery with it for 1 minute. The charge rates for AA batteries is 1000mA and for AAA, it is 600mA. The AAA battery was charged at 1000mA by jumpering the battery to the AA connections.

When the batteries did not terminate (both Sanyo AAA @600mA and Energizer AA @ 1000mA), the voltages would increase by 4mV over about 8 to 16 seconds seconds and then suddently drop by 4mV. This negative drop did not occur for good batteries that would increase by this amount over 3 seconds (i.e. 1mV/sec). I think the charger averages several consecurtive measurements (say 20) to determinte the state of the battery. In the middle of the charge , the voltage profile is fairly flat and slowly rising (near zero slope), however, measuring over a longer interval with several measurements can help the charger distinguish this from the "zero slope" region for "negative delta-V" type chargers. It is necessary for that charger to recognize this end of charge "zero slope" region so that it can get ready for the next stage which is to find the "negative slope" termination signal. So the termination signal can be thought of as a two stage process where the "zero slope" signal is first found and generated and then allows the charger to enter a new state where the "negative slope" and final signal is generated to stop the charge. The problem with batteries that miss termination is that a false zero slope condition is maintained, indefinitely, by small rising voltages over about ten cycles, followed by a sudden, drop. The average seems to a "zero slope".


Referring to the graph, the non-terminating behaviour for the Energizer would have the graph stretched out to 16 intervals before reaching 4mV. The pulse would then drop from the 1.504 steady value to a value of 1.500 vollts and this pattern would repeat itself.

When the charge rate of the 900mA-hr battery was increased to 1000mA (i.e. 1.11C), the voltage rate was increased to about 1mV/sec and the non-monotonocity dissapeared and the battery terminated in 3 minutes.

Conclusion:

1. The 1000mA charge rate delivered over 1 second gives a charge of Q=I*t=1A * 1 sec = 1 Coulomb
2. For a 1mV/sec voltage charge rate, the effective capacitance is given by:
   Delta_V_rate = I/Ceff
   This implies that
   Ceff=I/Delta_V_rate
   For the Failsafe 1600mA-hr battery:
   Ceff=1 amp/1mV/sec =1000 Farads
3. For larger capacity batteries, the Ceff scales rougly by the capacity, but not always, as in the casee of the Energizer battery. Ceff_energizer=4000 Farads.
   To achieve, the Delta_V_rate of 1mV/sec, the charge rate would need to be increased up 4 amps!!!

The interesting thing to note from the earlier charge profiles of the eneloop batteries, is that the slope near termination of a higher value. This means that Ceff is small and that smaller charging currents can be used to achieved a 1mV/sec minimal slope. This may the solution to achieving effective termination with NiMh batteries.

Based upon my few tests, can the slope of the voltage be used and related to the standard recommendation of charging at somewhere between 0.5C and 1.0C ? Yes...with several caveats:

1. Not all NiMh batteries and their chemistries are the same as my well terminating batteries
2. I cannot determine the discharge capacity of my batteries with my charger

So assuming the stated 1600mA-hr rating to accurate and the slope data to apply equally to other NiMH batteries, we have the following data:

[B]
%1-hour Charge-Rate    Slope        Notes
[/B]
0.625 CR              1.0 mV/sec     Failsafe used as reference
0.5   CR              0.8 mV/sec     based on min recommended rate
1.0   CR              1.6 mV/sec     based on max recommended rate

PeAK

 

[PeAK]

.
.
.
The interesting thing to note from the earlier charge profiles of the eneloop batteries, is that the voltage rate near termination is sharp. This means that Ceff is small and that smaller charging currents can be used to achieved charging ramps of 1mV/sec. This may the solution to achieving effective termination with NiMh batteries. The other alternative would be to use lower capacity batteries less than 2x the charge rate per hour. In the case of 1000mA charger, batteries less than 2000mA-hr.

PeAK

I thought I followup on the above recommendations and elaborate on the term "sharp": By looking at the eneloop battery as it neared the end of the charging cycle. The following figure came from this post:

In the middle portion of the charge, the slope chages by about 30mV between C=500mAh (t=19 minutes) and C=1500mAh (t=56 minutes). The charge rate is a bit under 1mV/minute (?? in light of previous post). As we get near the negative delta-V portion of the curve with the voltage plateau, the slope increases to about 10mV/min. At the plateau, the slope is 0 mV/min and this dramatic change in the slope makes it easier for termination algorithims to end the charge. The other benefit of the eneloop is the sharp but small temperature rise (5 degees celsisu) during this short slope transition region.

Now contrast this with another "No-name" battery from the same thread:

Over the same middle portion of the charge (C=500->1500 or t=19min->56min), the voltage changes by about 30mV or just under 1mV/min like the eneloop. The difference is near the zero-slope plateau region where the slope "does not increase" and gradually drops to zero over the course of about 20 minutes from the peak slope of about 1mV/min (vs 10mV/min for the eneloop). A plot of the slope change trend is give below (ignore the units for now....needs to be sorted out with Mr. Happy):

For the eneloop the time between the peak slope and the zero slope is much smaller as can be seen from the plot below:

In the above plot the transition from the peak positive slope ( 10mV/min) to zero slope happens between C=1800 to 2000 mAh (i.e. time=68 min to 75 minutes). The time interval is only about 7 minutes or about a third shorter than the 20 minute interval for the "no-name" battery.


Higher Charging Rates...some thoughts:
If the transition interval, above, is based upon the accumulated charge, increasing the charge current by 2 would halved this transition interval. The other postive effect is that slope would double. This argues well for using higher charge rates to lead to less ambiguous termination conditions. Above the 1C max rate, typically seen and recommended, reduce number of charge cycles and other degradation may come into play. However, the degradation and heating resulting from a missed termination may be more catastrophic to the longevity of a NiMh battery.

This might explain the unambiguous charge termination and cooler batteries charged using the Maha C808M despite the higher charging rate of 2Amps (1C for 2000mA-hr cells). Users have reported success with this charger using low capacity NiCd which means that the charge rates are well above 1C. The solution to the missed termination issues with early generation Maha C9000 chargers could potentially be solved by setting the charge current to a fixed value of 2000 mA regardless of the stated capacity of the cell being charged.

This thread also raises the issue of what happens with "C" and "D" size batteries as recommending at least a 0.5C charge rate with 8,000mA-hr D size battery would mean a charge current of 4 amps!!! My feeling is that answer lies with engineering the sharper slope transition region pioneered in the Sanyo eneloop.

PeAK

 

[PeAK]

.
.
.
Most consumers erroneously hold to the idea that slow charging will result in better performance from their cells, so if a charger was available that charged at these "normal" rates, it would be shunned by the general public. It is also more expensive to produce a charger that charges at higher charge rates.
.
.
.
Tom

.
.
.
...However I know that with advanced charging and charge termination algorithms, you can charge very fast and still keep the cells cool. My Schulze charger charged some 1200 mAh cells at 5 amps and the cells remained cool.

Tom

Seems like I'm having a discussion with myself but this is just a place holder for points made in other posts related to the comments by Tom above and previous posts in this thread.

1. Many of the missed terminations seem to only occur when a battery is refreshed (i.e. charged and then discharged/charged ?
2. Maha bought out the 800s charger with reduced current charge levels (1A) compared to the 801D (2A) due to the concerns of users accustomed to lower charging currents and NiCads.
   
   Outdated paradigms and this thread argues for higher charge rates and the 800s may be more problematic with older cells with strange/low voltage ramps near the end of the charge phase near the zero slope plateau. I take the 808M.

First Set of Eneloops

Purchased my first set of eneloops and will be conducting experiments to validate some of the data and conjectures above

Early Results:

1. Early testing indicates a voltage ramp of 0.3mv/sec at the 1.4 volt level on a new set of batteries. The batteries were on the charger for about 20 minutes and the batteries were removed from the charger. The batteries were fairly warm.
   
   After a 10 minute discharge into a 4 ohm load, the battery attained a much sharper voltage change rate of 1mV/sec. Termination occured within 3 minutes at 1.507V. It seems as if it is not prudent to charge LSD batteries prior to using them. A 10 minute discharge at 0.3C primes them for charging with proper termiation.
2. Repeating experiment on set number 2. Results to follow. It seems as if this second set generated very similar results. The noteworthy thing is when the battery voltage was in the mid 1400mV (i.e. 1450), the last digit on the voltmeter acted like a counter with a one millivolt increase per second.
3. More measurements were made the next morning. Six charged eneloops measured 1.433V...amazing repeatability. The battereis were replaced into the charger to see how (and if) they would terminate charge. The voltage ramp was about 1mV/sec and within 3 minutes the battery, charged from the night before, terminated at 1.525 V. At the point at which it terminated, the voltage did not increase monotonically. (i.e. increase at 0.5mV/sec for six seconds and then suddenly drop 3mV) for a few suspicious cycles. The overall larger trend was monotonic, though.

NiCad Experiments:
The charge I use charges both chemistries and probably uses "zero delta-V" termination scheme. The most remarkable thing about the charging profile of a 600mA-hr (came with a solar outdoor light set) is that clean and unabiguous termination. The voltage ramp seems to maintain a 1mV/sec ramp right up to the termination of the pulse charge algorithim (signalled by green LED). At that point or just before it, the voltage (or slope) has either reduced to a point that a single measurement provides enough information to terminate.

The more ambiguous termination occurs with the slower ramping NiMH cells which can drop to 0.2mV/sec and stay there a while along with 1mV gains that are lost by sudden drops. All this seems to confuse some of the chargers. The algorithims are not known but it seems as if a longer history is used for slower changes and a shorter history for faster voltage ramps.



PeAK