MH-C9000 termination the verdict?

[NiOOH]

As Tom and other older members may remember, I was the person who first voiced the opinion that the revised version of C9000 uses maxV (maximum cell voltage) of 1.47 V as the primary charge termination criterion. At first, no one really believed it, but slowly it has become the common opinion. I believed in it too, until tonight.
I was charging four Unioss LSD Hybrio technology cells (not Hybriloops) at 1 Amp. The cells were about 50% discharged. The cells are rather new but have gone through a break-in cycle on the C9000 plus 4-5 fast charges on the same charger and Sanyo MQH-02.Three of four cells hit 1.47 V and terminated as usual. The cell in slot 3 was showing 1.45 V. I knew this is a balanced set, so I decided to watch and see what will happen. The cell stayed at 1.45V for couple more minutes and then... terminated showing 50 mA more input than the rest. It stayed at the same temperature like the others i.e. berely warm to the touch.
Now, the question is what caused this termination. I bet it is not -dV. Neither the charge input, nor the cell temperature suggest that.
So, what is everyone opinion about such events. Has anyone eelse seen that. I would also invite William or any other person at Maha to finally clear that, i.e. WHAT is the primary termination mechanism of today's C9000.

 

[Turak]

Neither method is what I would refer to as the 'PRIMARY' method. It simply has 2 methods....either one may kick in depending on the particular cells internal chemical characteristics at that point in time.

It just depends on which termination method happens to affect that particular battery first.....PresetMaxV or -DeltaV

Because of MAHA lowering the Preset MaxV voltage to 1.47V plus or minus .01V.

MOST......read it again....MOST..... AA cells tend to hit the 1.47V BEFORE they ever hit -DeltaV.

Whether the cell terminates on -DeltaV or PresetMaxV could vary for EVERY single brand you happen to put in the charger....because the internal chemical makeup varies from manufacturer to manufacturer. BUt because of the low PresetMaxV, most tend to terminate on PresetMaxV now.

Now this DOES NOT mean that every cell will always terminate at 1.47V (PresetMaxV).... It IS POSSIBLE that some cells will actually become fully saturated and terminate on -DeltaV before they ever hit 1.47V.

But I tell you...I find it to be a fairly rare occurrence. Most likely to occur with newer batteries that have a very low internal resistance.

Some could argue that the -DeltaV method is the Primary method, even though most batteries now terminate on PresetMaxV. The -DeltaV method is the method that the MH-C9000 was SUPPOSED to use as its so called 'primary' method when the MH-C9000 first came out with the PresetMaxV as a safety backup method.

The problem appeared when the early MH-C9000's (Rev F's) would not terminate reliably when using any charge rates below about .5C. It was an embarrassment for MAHA since they clearly stated in their instructions that you could go down to .3C with no problem. Because of this problem, MAHA started recommending that you use the higher rates with their 'preferred' or 'recommended' rates being .5C to 1C. This suddenly became a problem too....because now, with the higher rates (mostly the 1C rate), when the charger actually terminated using -DeltaV, the batteries came off very hot. Maha's answer to this was to lower the PresetMaxV safety cutoff point to 1.47V.

This had the effect of making the PresetMaxV safety cutoff point now became the 'primary' termination method. Meaning simply that 'most' batteries now terminated on the PresetMaxV instead of ever reaching -DeltaV.

Basically it was a compromise solution because they could not seem to manage to charge the cells at 1C using the -DeltaV method without them coming off the charger hot enough to burn your hand.

 

[Mr Happy]

Everything that Turak said.

There is a post from William in the C9000 thread indicating that the C9000 uses a test matrix to determine the termination point. Measurements considered in the matrix include absolute voltage, change in voltage, absolute temperature and change in temperature.

My interpretation is that no single test is the "primary" test, but a strong enough signal on any one of the tests will tend to tilt the result in favour of that test.

Secondly, while eneloops and some other batteries tend to reach a voltage above 1.47 at the end of charge, not all kinds of battery do. In particular, the Rayovac Hybrid and clones (the Uniross LSD in the UK may be one of these), do not always reach such a voltage. This is supported by the various missed termination instances experienced by Turbo DV8 when charging Hybrids. If Turbo DV8's Rayovac cells routinely hit 1.47 V they would terminate with certainty at that point.

 

[Mr Happy]

Basically it was a compromise solution because they could not seem to manage to charge the cells at 1C using the -DeltaV method without them coming off the charger hot enough to burn your hand.

To add a little insight to this, consider the charging experiment I did on an eneloop in this post. Consider especially the last graph "End of Charge Detail":

Look particularly where I have (incorrectly) marked the "−∆V Signal". This is in fact the 0∆V point, and the voltage has not dropped yet. Even so, look how fast the temperature is rising by then.

Each graticule on the vertical axis is 10 mV. If we suppose termination is to be delayed until a −∆V of −5 mV is seen, we can predict this happening at 2050 to 2100 mAh. Continuing the temperature curve, it is not unreasonable to suppose the temperature would reach 45 to 50°C by that time (the rise is not linear but is still accelerating). If we wait for −10 mV the temperature would get even higher.

So a high temperature on −∆V is an intrinsic property even with a good cell like an eneloop. We could reduce the temperature rise by dialling back to a smaller −∆V or even 0∆V, but this increases the chance of false terminations.

What all this tells us is that charge control in consumer level NiMH chargers is really complicated and some kind of compromise is always going to be involved.

 

[NiOOH]

Everything that Turak said.

There is a post from William in the C9000 thread indicating that the C9000 uses a test matrix to determine the termination point. Measurements considered in the matrix include absolute voltage, change in voltage, absolute temperature and change in temperature.

My interpretation is that no single test is the "primary" test, but a strong enough signal on any one of the tests will tend to tilt the result in favour of that test.

Secondly, while eneloops and some other batteries tend to reach a voltage above 1.47 at the end of charge, not all kinds of battery do. In particular, the Rayovac Hybrid and clones (the Uniross LSD in the UK may be one of these), do not always reach such a voltage. This is supported by the various missed termination instances experienced by Turbo DV8 when charging Hybrids. If Turbo DV8's Rayovac cells routinely hit 1.47 V they would terminate with certainty at that point.

I agree. In any charger that uses multiple termination schemes the charger terminates on whatever comes or could be detected first. I have a couple of Varta Ready 2 use cells that do not reach 1.47 V during charge at 1 Amp and exhibit typical -dV termination. They accept about 10% more charge compared to the other two in the set that terminate on maxV and heat to 50 degrees C. I can even see the voltage drop on screen. It goes down from 1.45 to 1.44 shortly before termination. For these cells there is no doubt in my mind that -dV kicks in.
However, what was seen last night with this Uniross cell is different. The cell reached 1.45 V and terminated roughly at the same charge input as the rest that hit 1.47. It didn't heat either.

Running some analysis based on your data and the voltage profile during charging on this particular cell, it looked like the termination occured at the inflection point. AFAIK, this has never been mentioned as a criterion for termination with C9000.
Has William ever stated clearly what were the modifications done to the termination criteria on the 2007 and later units. Things like "test matrix" sound advanced but does not tell much.

 

[koravei]

I wonder is it possible to reduce the charging current and achieve some safe battery temperature and terminate by hitting -deltaV ?
If the battery temperature is under some safe limit, the charger can charge with the programmed current,
otherwise it must adjust (reduce) the charging current and keep the temperature constant - something like temperature controller in function of the charging current. Please tell me if i am wrong somewhere, is it possible to do such a thing?

 

[Verndog]

Since the voltage cutoff appears to be a safety net and it affects performance on top rated cells, does anyone know why Maha would not at least allow say a 2 stage user defined safety voltage? Say 1.48 and 1.55 volts? When the safety becomes the primary then I think you need to rethink the strategy IMO. Also if this charger has thermal cutoff then why not rely on that as a safety? This is the 1 main issue that kept me from purchasing this charger in the past.

 

[Bumble]

some very valid points there Verndog.. isnt the 2 hr 100mah top-up charge designed to help the undercharging by the c-9000 when charging such cells as eneloops ?? if im in a hurry when it say "done" on the display i just pull them out also waiting for 2hrs when it says "done" also seems a bit contradictory?? . but overall i like the c-9000 never fried a cell yet (touches wood) ....

 

[Mr Happy]

The C9000 charges in pulses and measures the off-charge voltage between pulses to control the end of charge detection. So the 1.47 V max voltage is the "resting" or "off charge" voltage. If we look at the various charging graphs I have included in other threads recently it can be seen that as soon as charging stops the cell voltage drops steeply to about 1.47 V -- this is the voltage the C9000 is testing. If you tried to set the max voltage to 1.55 V it would never be reached and charging would never terminate.

 

[VidPro]

Ya Mr Happy I was just thinking the chart in this thread, shows the 1.47 voltage tagged, but the chart is with the power applied ? isnt it? The 9000 is after the power is removed at least temporarily.
because it is sorta like that, where even with the 200ma topping it might not get 100% charged, but not that far back

Applying the thermal only for dire emergency makes sence, because none of these chargers stuff a thermal probe right on the battery, breezes from HVAC vents, air from windows, different ambient charge locations, and all. It would take a lot more than a probe sticking up wherever to use that information wisely or completly in the algorithm.

 

[Mr Happy]

Ya Mr Happy I was just thinking the chart in this thread, shows the 1.47 voltage tagged, but the chart is with the power applied ? isnt it? The 9000 is after the power is removed at least temporarily.
because it is sorta like that, where even with the 200ma topping it might not get 100% charged, but not that far back

Yes, this is tricky. On that graph I marked 1.47 V with power applied, but since then I have learned that is not exactly the test the C9000 uses. I think I would have to do some further experiments to get a more accurate picture.

 

[Verndog]

The C9000 charges in pulses and measures the off-charge voltage between pulses to control the end of charge detection. So the 1.47 V max voltage is the "resting" or "off charge" voltage. If we look at the various charging graphs I have included in other threads recently it can be seen that as soon as charging stops the cell voltage drops steeply to about 1.47 V -- this is the voltage the C9000 is testing. If you tried to set the max voltage to 1.55 V it would never be reached and charging would never terminate.

Makes sense, but how long are the pulses? Peak voltage will vary brand to brand, cell to cell with many factors as you know. Also for a high tech devise that is a fairly crude cutoff method of termination IMO. Again, if this is added as a safety then it should only be rare occasion it is activated, not the norm. Do I have this correct here? The Maha uses 4 different methods to terminate. 1) minus delta V, 2) Max voltage, 3) Thermal cutoff, 4) some various unknown combination of parameters??

 

[VidPro]

here is Russels Scope stuff, that shows the pulse rate.
http://www.candlepowerforums.com/vb...age-measured&p=3590601&viewfull=1#post3590601

 

[Mr Happy]

Makes sense, but how long are the pulses? Peak voltage will vary brand to brand, cell to cell with many factors as you know. Also for a high tech devise that is a fairly crude cutoff method of termination IMO. Again, if this is added as a safety then it should only be rare occasion it is activated, not the norm. Do I have this correct here? The Maha uses 4 different methods to terminate. 1) minus delta V, 2) Max voltage, 3) Thermal cutoff, 4) some various unknown combination of parameters??

The C9000 seems to use both temperature and voltage measurements to decide when charging should end, with more weight given to voltage. The charger has a minus delta-V detection and this would have been fine, except that Maha screwed up and used the off charge voltage for minus delta-V detection instead of the on charge voltage. You have to use the on charge voltage for minus delta-V because the drop in voltage arises from the decrease in internal resistance when the temperature goes up at the end of charging. If no current is flowing you can't detect the change in internal resistance and therefore you can't detect a clear drop in charging voltage. (The C9000 can detect a sort of voltage drop as the sampled voltage has not fully settled out and still contains some remnant of the on charge voltage. This is why the C9000 can stop on minus delta-V if the Max V isn't reached first, just not reliably so.)

What Maha should have done is to fix this mistake in voltage detection and measure the on charge voltage, but what they did instead is add the max V threshold at 1.47 V to compensate for the other mistake and prevent missed terminations.

There is a similar flaw when the C9000 is discharging. When discharging you want to measure the off load voltage between discharge pulses and set a threshold around 1.1 V (recall that a resting 1.2 V on an NiMH cell is empty, but it make take a while for the voltage to rebound to that level after discharge). However the C9000 measures the discharge voltage while a 1 A discharge pulse is being applied and sets the threshold at 0.9 V. This is OK for strong cells, but for weaker high resistance cells especially AAA cells the discharge ends prematurely.

So we end up with a device where the hardware is good but the firmware is flawed. It would be really neat to have an "open source" firmware program for the C9000 where we could fix the flaws.

 

[VidPro]

When discharging you want to measure the off load voltage between discharge pulses and set a threshold around 1.1 V (recall that a resting 1.2 V on an NiMH cell is empty, but it make take a while for the voltage to rebound to that level after discharge). However the C9000 measures the discharge voltage while a 1 A discharge pulse is being applied and sets the threshold at 0.9 V. This is OK for strong cells, but for weaker high resistance cells especially AAA cells the discharge ends prematurely.

William said they did this On Purpose. they chose the loaded voltage to terminate on, because (and i will paraphrase) When you stop the load on a crappy battery , it bounces right back up in voltage. So it would never stop the discharge.
Add to that (opinion) that they are always at a 1Amp discharge , in these pulses. So take a AAA crappy battery , try and load it with 1Amp load :sick2: it tanks, stop the load, and it sits at 1.3v Repeat till the crows come home.

On the other hand, If i wanted to wait till the crows came home, I could make that decision for myself, if they would let me. The batteries in most need of discharge , wont be very fully discharged on this unit. So the discharge is pretty much, only usefull for testing , testing good stuff.

in that situation, a slower rate discharge (instead of hard PWM) could do it either way, without any problem, so the lesser of the batteries could be discharged.

 

[VidPro]

""except that Maha screwed up and used the off charge voltage for minus delta-V detection instead of the on charge voltage.""

How is that gonna work? in this unit they are using 2Amp :green: pulses, even in the (psudo) break-in mode.
Trying to read a voltage drop when slamming in the power that hard , is like . . . well its like the eneloop you popped just yesterday. the large quantity of power applied easily obscures the voltage changes. When the drop does occur it is visable in the rested. But EITHER way, it is not readily visable on all batteries the same way all the time, especially batteries that have not been cycled in long whiles, and crap batteries. although it becomes more visable the more you beat the heck out of the cell.

My batteries have specifically stated , that they do not like V-drop termination methods
:eeew: they are outside right now picketing my chargers :eeksign:

It may have taken many errors for maha to "cheap fix" some of this stuff with another error, but the "fix" is better than the cure this time .

look at the signal on the scope with the power applied, I wouldnt want to read that with a microcontroller

At lower current, (current control) with clean power, the reading of the voltage drop while power is applied is much easier. I assume that method is used on many of the low rate V-drop terminating style chargers. . . . when they didnt time out because the cell never did v-drop.

As far as allowing the users to make changes , I would love that, a little tweak here and there, as long as I dont have to re-write a whole program. I wouldnt be to thrilled with having to "flash" and decompile and reprogram recompile , but a changing a few parameters might help. Hobby chargers can change a few very useful parameters, with those I only wanted to tweak 2 more things a bit.

 

[Verndog]

What Maha should have done is to fix this mistake in voltage detection and measure the on charge voltage, but what they did instead is add the max V threshold at 1.47 V to compensate for the other mistake and prevent missed terminations.

That there is a pretty big blunder. So, any speculation why after over 4 years they haven't bothered to fix this?

william said they did this On Purpose. they chose the loaded voltage to terminate on, because (and i will paraphrase) When you stop the load on a crappy battery , it bounces right back up in voltage. So it would never stop the discharge. Add to that (my own opinion) that they are always at a 1Amp discharge , in these pulses. So take a AAA crappy battery , try and load it with 1Amp load :sick2: it tanks, stop the load, and it sits at 1.3v Repeat till the crows come home.

Most hobby chargers discharge to V under load as well (although I dont think NiMh cycles are pulsed so there may not be a choice). I would think under discharge measuring V this should be fine (but the end voltage to stop @ should change with the load applied). When we used to run Nimh in RC racing we'd discharge under high 10-30 amp rates, then bench discharge on a balancer (6 cell) @.5A to .9V.

 

[Mr Happy]

also look at the signal on the scope with the power applied, i wouldnt want to have to read that with a microcontroller

I imagine that's why Maha use the off charge voltage. The off charge voltage is smooth and easy to measure, while the on charge voltage has all the noise from the power supply messing up the signal. It's a tough call, but I still think they screwed up.

 

[mccririck]

Is it not better to discharge at a lower rate when you get down towards 0.9V so that there will not be so much rebound? It could discharge at 1A down to 0.95V then kick down to 0.2V to extract as much juice as possible.

 

[Mr Happy]

william said they did this On Purpose. they chose the loaded voltage to terminate on, because (and i will paraphrase) When you stop the load on a crappy battery , it bounces right back up in voltage. So it would never stop the discharge. Add to that (my own opinion) that they are always at a 1Amp discharge , in these pulses. So take a AAA crappy battery , try and load it with 1Amp load :sick2: it tanks, stop the load, and it sits at 1.3v Repeat till the crows come home.

True. If you want to measure and accumulate the mAh capacity accurately, you want to ensure your 1 A pulses are actually 1 A. I suppose here the designers were stuck between a rock and a hard place.