Duracell Power Gauge CEF21 mod for improved charging of AAA cells

[Mr Happy]

Here is what the CEF21 looks like inside. This is the Duracell Power Gauge charger (Indicator Charger in Europe).

Top view:

Upper left is the mains power supply. Transformer TNY266B brings low voltage power over to the rest of the circuitry. Lower left there are four parallel power supply channels using four coils (two hidden under the metal shield). These supply 5 V DC to the USB power output lower left when they are not being used to charge batteries. The white socket lower center is where the cable from the LED bar graph display plugs in. Lower right would appear to be a test connector.

Bottom view:

Erm...

This ain't no simple charger.

I've made these pictures as large as forum guidelines allow. I have higher resolution versions I could link to.

The square chip upper right is an ATMEGA. There is some measure of intelligent control going on in this charger.

The four big resistors seem to be connected between the AAA and AA contacts of each channel. If I read the color code correctly, it is orange-black-silver-red making them 0R3, 2% tolerance.

My goal is to modify the charger so that AAA cells get charged with the same 400 mA that AA cells get. As manufactured, the charger uses 175 mA to charge AAA cells and it seems to be unreliable at terminating. I think it would terminate better at 400 mA.

Edit: Please see post #50 below for details of how to make the modification.

 

[Mr Happy]

Re: Duracell CEF21 disassembled (modifications, anyone?)

Here are close up views of the connections around the battery contacts.

Lower side:

Upper side:

I'm beginning to think the resistors are just connected in series in the AAA charging circuit as current limiting components. And therefore if I shorted them out with direct links, the AAA cells would get the full charging current?

 

[PeAK]

Re: Duracell CEF21 disassembled (modifications, anyone?)

Seems to be either one extra (maybe two) connection beyond the resistor to this AAA node.

For the charging bank nearest the test connector, a trace runs along the top view to the middle of the board. On the bottom view, the square pad has a smaller round pad that seems to go nowhere.

I would hazard a guess that the current comes from the AA terminal to the AAA terminal through the resistor. Not sure about the value of the resistor as a 175mA would generate very little heat/watts across a 0.3 ohm resistor...???

 

[Mr Happy]

Re: Duracell CEF21 disassembled (modifications, anyone?)

I've been puzzled by those round pads that seem to go nowhere too. But if I hold the board up and shine light through it, they really do seem unconnected.

I was looking for the connections to the AA pads. In each case there seems to be a thick current supply trace leading to the AA connector, and one thin (voltage sensing?) trace leading away.

I think there is a similar situation with the AAA connections: a current supply through the resistor from the AA pad and a sensing trace leading away. I'm trying to follow where those traces go.

Maybe the circuit senses the voltage across those resistors? When an AA cell is being charged the AAA pad will be at equal potential and the voltage difference will be zero. When an AAA cell is being charged the charging current will pass through the resistor producing a voltage drop of 175 mA x 0.3 ohms = 52.5 mV. The circuit may throttle the charging current so that the sensed voltage is less than or equal to ~50 mV?

If those are current sense resistors then I can increase the current to the full value by shorting out the resistors.

The tight tolerance value of 2% would add weight to the idea that they are sense resistors.

 

[PeAK]

Re: Duracell CEF21 disassembled (modifications, anyone?)

....When an AAA cell is being charged the charging current will pass through the resistor producing a voltage drop of 175 mA x 0.3 ohms = 52.5 mV. The circuit may throttle the charging current so that the sensed voltage is less than or equal to ~50 mV?

If those are current sense resistors then I can increase the current to the full value by shorting out the resistors.

The tight tolerance value of 2% would add weight to the idea that they are sense resistors.

It might be an idea to monitor the voltage across "the resistor" when charging a AA battery and also when charging an AAA battery to validate some of the above assumptions about the current traversing it.

 

[Mr Happy]

Re: Duracell CEF21 disassembled (modifications, anyone?)

It might be an idea to monitor the voltage across "the resistor" when charging a AA battery and also when charging an AAA battery to validate some of the above assumptions about the current traversing it.

Good plan.

 

[Mr Happy]

Re: Duracell CEF21 disassembled (modifications, anyone?)

My measurements are consistent with the current sense resistor hypothesis.

I placed a voltmeter between the AA and AAA connectors (+ to AA, - to AAA). When I turned on the charger with an AAA cell connected the voltage difference rose rapidly towards ~80 mV and then dropped down to settle around 50 mV, remaining fairly steady at that value with a small up and down cycle imposed.

On doing the same experiment with an AA cell connected the voltage difference remained close to zero, hovering around -2 mV (why negative?).

For my next test I will short the AAA connector to the AA connector and then charge an 800 mAh AAA cell. If my theory is right it will take about 2 hours to charge and will then terminate on cue.

 

[Magic Matt]

Re: Duracell CEF21 disassembled (modifications, anyone?)

I've been puzzled by those round pads that seem to go nowhere too. But if I hold the board up and shine light through it, they really do seem unconnected.

I'm lead to believe it's some sort of anti-reverse-engineering ploy. I have loads of them in my VCR, and when I spoke to an engineer about it I was told it was to make "building circuits that bypass copy protection, or even just copying the design, much harder".

Given there are fake batteries and chargers out there, I can see some sense in it. It would certainly make it easier to prove a design had just been copied if all the dummy tracks were duplicated!

 

[PeAK]

Re: Duracell CEF21 disassembled (modifications, anyone?)

....there seems to be a thick current supply trace leading to the AA connector, and one thin (voltage sensing?) trace leading away.

I think there is a similar situation with the AAA connections: a current supply through the resistor from the AA pad and a sensing trace leading away. I'm trying to follow where those traces go.

Maybe the circuit senses the voltage across those resistors? ...

Nice Sunday morning project...

There is a style of current source derived from a floating reference voltage that is used in LM78xx type voltage regulators. A voltage is impressed across a resistor that attaches across the output and a sense/feedback node to derive a current.

For the above to hold true in the "fuel gauge" charger, this voltage would correspond to about 93mV and a resistor of about 0.223 ohms would be needed to derived the 400mA AA charging current. The 0.3 ohm resistor would be needed to be added in series to the 0.233 ohm resistor to make up a 0.533 ohm resistor to derive 175mA.

For the bottom charging bank (nearest the test connnector), I see a trace going up from the AA pad and another one going up from the AAA pad that I think you refer to as the "sense" traces. For these to serve this function, they would have to be traced back to the same point on the IC chip. On the bottom view, they seem to do converge toward the middle of the board but then the resolution does not hold up. A high res shot would be useful here.

Question: On the top side around the coils glued on the top side are two components marked Z3 and Z4 on the board. This would indicate that they are zener diodes but could they be those 0.223 ohm resistors I mention. A picture would be useful.

Experiment: You mentioned that putting a short across the resistor might be one way to boost the current. A half step would be to put a 1 ohm in parallel (held in place by your hand) with the 0.3 ohm resistor and measure the corresponding current going into the battery. If the above ideas hold, then the current should increase to about 200mA. Note, somebody here suggested a way to use an folded aluminum foil (8 folds) used to form a sandwhich of foil-paper-foil that can be used break the connections of a battery while sitting in a charger to provide two attach points (foil) for inserting an ammeter. See pic below:

 

[Mr Happy]

Re: Duracell CEF21 disassembled (modifications, anyone?)

Here are high resolution images of each side of the circuit board to help with following the traces:

 

[Mr Happy]

Re: Duracell CEF21 disassembled (modifications, anyone?)

I'm running the charger disassembled at the moment and using jump leads to connect from the charger to the battery. Last night I tried the experiment of connecting the AAA terminal to the AA terminal but it was unsuccessful. The AAA cell took about 5 hours to charge as normal. I'm not sure what went wrong; maybe my shorting lead was too long and too high resistance.

I am now charging an AAA cell directly from the AA connector as a reference test to see if that takes the expected two hours to complete.

Current measurements are difficult. While my meter sees steady voltages, the current through the cell appears to be pulsed and the meter doesn't respond fast enough to catch the values. (As I write this I think I might have a different meter with a max hold setting. Maybe that will do better.)

I don't think the charging circuit is using a linear regulator. It think it is using a buck configuration with current sensing. For instance, the transformer leading from the mains power section appears to be feeding a set of eight power diodes (mostly buried in the "stuff" encapsulating the coils) but two of these diodes can be seen in the top view as D8 and D9. In the bottom view they are the two rows of eight thick round pads just north of the transformer.

Above these pads are four configurations of big coil, big capacitor, transistors, chips and other components (e.g. Q4, Q8, R5, C12 on the bottom view). Each of these configurations are what I think are the buck circuits. Just above R15 a thick trace leads back to near R5 and Q8. This I presume is an output feeding one charging channel.

 

[Mr Happy]

Re: Duracell CEF21 disassembled (modifications, anyone?)

Well, my attempted modifications are failing

I have tried two different methods of temporarily connecting the AA post to the AAA post. I have not got the resistance down to zero, but my second attempt got the voltage differential down to about 5 mV. Even so, this is not getting the charging current to rise much above 175 mA when connecting a cell directly to the AAA post. I have verified that when I connect an AAA cell directly to the AA post the charging current is 400 mA. Somehow the circuit seems to be detecting which charging post is in use regardless of my attempt to electrically join the AA and AAA posts. This is very puzzling.

I have been trying to avoid any destructive modifications or use of a soldering iron until I could verify that a planned modification would work, however I am now running out of ideas. What the circuit is doing such that it seems almost entirely insensitive to the resistance between the AA and AAA posts is a mystery.

On the other hand, I have further evidence for the Uniross Hybrio cells I call "hybriloops" being the same as eneloops. I noted that a Rayovac Hybrid reached a maximum voltage of 1.48 V at charge saturation, whereas the Hybrio reached 1.54 V. This latter voltage matches exactly what I observe from Eneloops and not from any other cell I have tested.

I have also noted that AAA cells of either Rayovac or Eneloop varieties do not give a very strong -dV signal. In both cases I saw the voltage reach a plateau and remain steady without decreasing as the cell temperature at the same time began to rise indicating full charge. I think that temperature sensing is more important for termination with AAA cells than with AA cells.

Related to temperature sensing the CEF21 has a temperature sensor tucked inside each negative post right where it touches the battery and far away from heat generating components in the power supply. This would certainly help with generating a good temperature signal.

 

[PeAK]

Re: Duracell CEF21 disassembled (modifications, anyone?)

Well, my attempted modifications are failing....the circuit is doing such that it seems almost entirely insensitive to the resistance between the AA and AAA posts is a mystery.../QUOTE]

Here's a reason to be happier:

I traced the two signals in your high res photo and found that the 50mV signal is used to drive one of the four comparators found in the good old LM339 (quad comparator). The comparator is configured to provide positive feedback using a 2.2 megaohm resistor to provide some hysteresis. This is why the current did not decrease with the short circuit you used. Take a look at the marked up board below:

​

The "AAA" trace in the top left marked with yellow ultimately makes it way thru R13 to pin5 (positive input) of the LM339. Similarly the "AAA" trace marked in red makes it way through R12 to pin4(negative input).

When an AAA battery is placed in the charger, the 50mV signal causes pin2(marked in orange) to drive the output "LO" [EDIT:"HI"] and signal the charging circuit to lower the current output. To have the charger put out a 400mA current, I suspect it would be as simple as configuring the outputs to put out a logic "HI" [EDIT:"LOW" (pins 1, 2, 13 and 14). These are shown as the purple traces in the above image. For the most part,these traces are on the other side of the board. This can be achieved on the input side or on the output side making a few selective cuts and joins. Try it out on one channel by using an Exacto knife to cut the orange trace (near R13 marking) and jumper a wire on the left hand side to pin 12 (ground pin) of the LM339. The left/bottom bank should now provide a 400mA charge current to AAA batteries.

Note: Edits in BLUE

 

[Mr Happy]

Re: Duracell CEF21 disassembled (modifications, anyone?)

Thank you. I noticed the LM339 yesterday, tried to follow traces and was forming the same idea that on/off control is involved. Therefore almost any voltage differential between the AA and AAA posts trips the comparator and flips the circuit into low current mode?

You have succeeded better than I did in following where the traces go. It's not easy when they keep jumping from one side of the board to the other. I shall examine your diagram with interest.

 

[Mr Happy]

Re: Duracell CEF21 disassembled (modifications, anyone?)

The "AA" trace in the top left marked with yellow ultimately makes it way thru R13 to pin5 (positive input) of the LM339. Similarly the "AAA" trace marked in red makes it way through R12 to pin4(negative input).

When an AAA battery is placed in the charger, the 50mV signal causes pin2(marked in orange) to drive the output "HI" and signal the charging circuit to lower the current output. To have the charger put out a 400mA current, I suspect it would be as simple as configuring the outputs to put out a logic "LOW" (pins 1, 2, 13 and 14). This can be achieved on the input side or on the output side making a few selective cuts and joins. Try it out on one channel by using an Exacto knife to cut the orange trace (near R13 marking) and jumper a wire on the left hand side to pin 12 (ground pin) of the LM339.

If I'm not mistaken, the pin and trace you have marked in orange on the image is actually pin 3, V+? Pin 2 would be the next pin north, next to the first 3 of C33?

 

[PeAK]

Re: Duracell CEF21 disassembled (modifications, anyone?)

If I'm not mistaken, the pin and trace you have marked in orange on the image is actually pin 3, V+? Pin 2 would be the next pin north, next to the first 3 of C33?

Mr Happy...you're absolutely right. I realized this after looking at the other 3 comparators this afternoon and decided to draw out the circuit. The outpus all come out of the top of the LM339 as positioned in your bottom view. You will see four traces, the second from left is V+, the other three are outputs with the other output coming out directly from pin#1.These outputs go to the ATMEL chip and I guess programme the current level. In the diagram below, I show this current source with the logic connection. Hope this helps:

Detect circuit for AAA/AA current selection
(Drawing courtesy of OpenOffice Draw)

​

Note: Diagram has comparator flipped...see post #26 and #27 for update
In the diagram, you see a red wire for when the AA battery is in the circuit. This puts both inputs at around 1.2V but the 2.2M will cause the negative input to be slightly higher than the positive input and the output will be a "LOW" logic. When the "green wire" is connected, the circuits works as described earlier. There is no hysteresis/"positive feedback" in the circuit and I will strike out those comments in the earlier post.

EDIT: The output traces can be cut either on the non-component side (top view) by cutting the respective purple traces and connecting the side going to the ATMEL chip to the ground plane. Just as easily, they can be cut on the component side and connected to the GND node. The leftmost charging bank output trace going to the ATMEL chip is the rightmost purple trace. These traces on the non-component side are shown in purple below:

 

[Mr Happy]

Re: Duracell CEF21 disassembled (modifications, anyone?)

Thank you, nice diagrams. I shall reflect on what to do next. :thinking:

 

[PeAK]

Re: Duracell CEF21 disassembled (modifications, anyone?)

Thank you, nice diagrams. I shall reflect on what to do next. :thinking:

You're most welcome but most of the thanks goes to Sun Microsystems that provided the OpenOffice Draw program. It is the first time that I've really tried it out and it is much better than the equivalent "drawing functionionality" provided within Microsoft Word...back to the CEF21 farm.

Provided you are comfortable with surface mount components, the easiest modifcation would be to remove the 10K resistor going to the negative input: Pretty easy to taste a CEF21 charger pumping a sweet 400mA into AAA batteries.

 

[Mr Happy]

Re: Duracell CEF21 disassembled (modifications, anyone?)

While I await my next move, there is another neat thing I discovered over the weekend. After I charged a Hybriloop up to 1.54 V I later attempted to put it back on charge. I found that the charger flashed the error indicator and refused to accept it. It evidently has a high voltage test. It would be appealing to think this is to prevent the overcharging of a fully charged cell, but I think perhaps it is more likely to reject an alkaline cell that might be placed in the slot.

 

[PeAK]

Re: Duracell CEF21 disassembled (modifications, anyone?)

While I await my next move...

The following change to the circuit is the simplest to test in circuit but takes a bit more room than "10K removal mod" as component side clearances to the casing may be tight. It is also the most elegant if you change your mind and wish to UNDO the mod.

(drawing courtesy of OpenOffice program)
[EDIT: error found January 22/2010:See post #26/#27 below]

​

Tack a 270 Kohm resistor across R39 and then try a AAA in the leftmost bank. The current should be 400mA. The change pumps additional current through the bottom 10K resistor (toward the AAA battery) to generate a 60mV signal across it toward the comparator. This compensates for the 50mV drop of the 0.3 ohm resistor so that the net result is the positive [edit: negative] input terminal ending up at about 10mV higher than the negative [edit : positive] input terminal. This generates a logic HI [edit: "low" and sets the ATMEL/"current source" to 400mA. The only change is about 6uA of additional current flowing down from the 270K resistor and through the bottom 10K resistor that is then sunk by the AAA battery. The named of the output called HI_CURRENT.

EDIT: Corrected diagram follows all references to negative should be changed to posistive and vice-versa