EEVblog #811 - How The Varta 15 Minute Battery Charger Works

[EEVblog]

A look inside the Varta 15 minute 8A NiMH fast charger.
A teardown and look at how it works and the circuit topology.
Can Sanyo Eneloops be charged this fast too?

Datasheets:
http://litel.com.pl/karty-produktow/Varta-AA-2100mah-56706-datasheet.pdf
http://www.din-tek.jp/Upload/Product%20Doc/Datasheet/DTM4410.pdf
http://www.ti.com/lit/ds/symlink/tl494.pdf

[IanB]

Charging fresh, new Eneloops at a 15 minute rate may work, although it probably stresses the batteries and shortens their life. It's something best kept for emergencies and maybe stopping after 10 minutes.

If you try to charge old, weary batteries like this they will pop and hiss. They can't recombine the charge gases fast enough to prevent over-pressuring and popping the seal.

[DanielS]

I had some Energized 15 minutes AA NiMH cells many years ago and they started showing significant degradation after maybe 50 cycles or so.

I ended up modding the charger so it would always be in AAA mode and not cause AA cells to become horribly hot.

[Yansi]

Hmmm. Isn't the silicon coefficient somwhere abotu 2.2mV/K instead of 10?  10 woul'd be too much I think... 

[nuno]

Maybe they're setting the DC-DC output current to a fixed value (I) then "PWM"ing the individual cells and displaying the average current (I x duty-cycle).

[AF6LJ]

One question I have.........
I wonder if the batteries are pulsed in charge mode.
Taking that a step further, could the batteries be receiving a charge at different times.
First one battery then the next and so on.....
This might also make measuring charg current easier since the micro controller knows which battery is being charged.
Time to bust out the scope and make some live measurements.   

[gorf]

What if the lower Fets are only shunts for missing batteries? Then the DC/DC can set the current for charge and discharge.

[retrolefty]

I assume that the datasheet C1 charge recommendation is based on heat rise during typical charging over some ambient temp range.

 So that fact that forced fan air cooling is being used most likely means that one can safely charge at higher 'C' rates, along with temperature sensing that must cut charge rate if some value temp is exceeded. So possibly the batteries are not being abused and will not cut lifetime charge cycle specs. Still charging at 4C is pretty cool.

 I know some of us R/C electric plane fans were charging older nicads at rates to get 15-20min charges, but boy would those cells get hot. 

[TheSteve]

4C charging is still for lightweights. As previously mentioned the RC crowd has been heavily abusing Nicad and NiMH cells for years. Back in the day I routinely charged Sanyo 1200 mAh cells at 10 amps. When we were feeling more adventurous we went as high as 40 amps. The ends of the cells would actually bubble out which increased the capacity of the cell. This came at a major sacrifice to the total # of cycles. The same charger would also discharge cells at up to 40 amps.
A key to this was that it used constant current - no PWM of any kind. If you want major punch from your batteries PWM sucks. Yes it had a decent heatsink on it - it had to!
The same device could also delivery pulses of current in the hundreds of amps at up to 99 volts to revive old cells or give battery packs a little more punch just before a race. Ahh the good day, had a few cells pop and fizzle but nothing like a modern day LiPo fire.

[Warhawk]

Hmmm. Isn't the silicon coefficient somwhere abotu 2.2mV/K instead of 10?  10 woul'd be too much I think... 

I also wanted to point it out. Rule of thumb is -2 mV / K.
A good number where to start.

[barnacle2k]

On reading the video title i thought the day had finally come you did a video on a topic you already covered. (As predicted on the Amp Hour)

Anyway what i wanted to comment on:

Sensefets do not contain a sense resistor.

Current senseFET technology depends on the close matching of transistor cells within the PowerMOS.
A TrenchMOS device comprises many thousands of transistor cells in parallel.
Elements within the device are identical and the DRAIN current is shared equally between them.
The more cells that are in parallel for a given MOSFET chip area, the lower its on-state resistance will be.
It is possible to isolate the SOURCE connections of several cells from those of the majority and bring them out onto a separate SENSE pin.
The PowerMOS can now be thought of as two transistors in parallel with a common GATE and DRAIN but separate SOURCE pins.
When the devices are turned on, the load current will be shared as a ratio of their on-state resistances.

[Helix70]

Although not the same class of device, I was inspired, and here is the inside of my Sanyo NC-MQR06W, with a miserly 560mAx4/1120mAx2/1680mAx1 (AA) charge rate.

I do like the isolation, and high quality soldering.

[hamdi.tn]

i don't understand the use of 1R resistor to link AAA battery terminal to AA battery terminal , of curse it doesn't matter but it will not be the first value to had in mind, why not a 0R. The only reason i see , is that those resistor can measure somehow the charge / discharge current for the AAA battery at least according to the dave's schematics.

i think the current shown on charge mode is just a saved number not really what it is, the fact that they are 8.000 and 4.000 is odd to me, but hey still can measure discharge current by linking the bottom shunt resistor to one particular battery using low side mosfet. one at a time, if done fast enough you will not notice.

[SteveyG]

The "tabs" on the AC adaptor are to meet the UK regulations on the minimum distance from the phase and neutral pins to the edge of the plug

[rgawron]

One can always keep an additional battery in his/her charger, so there's no need to wait until a battery is charged (not even 15 minutes). Charging speed is not an issue then.

[hamdi.tn]

There were several resistors in parallel so the actual resistance was much less than 1 Ohm.

yap am aware of that, still it's made to just link both terminal , it make more sens to use 0R

[Zbig]

One question I have.........
I wonder if the batteries are pulsed in charge mode.
Taking that a step further, could the batteries be receiving a charge at different times.
First one battery then the next and so on.....
This might also make measuring charg current easier since the micro controller knows which battery is being charged.
Time to bust out the scope and make some live measurements.   

Exactly that. The "-dV" detection method implies that anyway: you send a current for a fixed period of time down the cell, then you let go and measure the open circuit voltage. The moment the voltage actually gets lower than after the previous current spike indicates the full charge state fairly precisely for this type of chemistry. So the charger is "multiplexing" the cells. This also means that 8A is a mean current and that the actual instantaneous current is much higher probably.
Not true, as it turns out. IanB has explained it properly later (thanks).


EDIT:
The UI of this device looks awfully like the BC700, BC900 and their later derivatives (like NC-1000, etc.) that are quite popular these sides. That means it's most probably a rebranded unit with its roots pointing at some Chinese OEM somewhere.

[EEVblog]

Anyway what i wanted to comment on:
Sensefets do not contain a sense resistor.

Yes, I showed the equivalent circuit in the video.

[rr100]

I was multitasking during the video and I probably missed most of it ... from the instructions it seems that the charger has just "one speed" (i.e. you can't set lower currents)?!

I don't see any warning about using other cells - yes Dave mentioned that Eneloops might be working just as well but how does it know what cells it has?

[EEVblog]

I was multitasking during the video and I probably missed most of it ... from the instructions it seems that the charger has just "one speed" (i.e. you can't set lower currents)?!
I don't see any warning about using other cells - yes Dave mentioned that Eneloops might be working just as well but how does it know what cells it has?

Correct, and it doesn't know.
Yet no warnings about other brands etc.

[mux]

All questions you have in this video can be answered pretty easily, but probably not very satisfactorily.

All NiMHs (like has been said before already) can be abused a whole lot. They don't have nearly as low of a thermal runaway temperature (and have much less hotspotting) than lithium chemistries. I think the thermal runaway temp is like 175-200C? It's really high. They use these things on satellites for that reason. As a general rule, the failure mode of these cells at high current is dendrite formation (internal short-circuiting), which scales roughly as the square of the current. So 1500 cycles at 2A means 100ish cycles at 8A. Although I've seen RC enthusiasts pump 20A into cells and have them survive 25-50 cycles, so YMMV.

That being said: consumers rarely use their batteries for more than 50 cycles. Especially stuff like eneloops; it gets used maybe 5 or 10 times, then it falls between the pillows of the couch to be forgotten.

As for sensing; they're obviously only doing FET sensing, and very roughly so. FETs aren't good resistive elements for many reasons, especially when used in a switched mode like they are. The only reasonably accurate current sensing that is going on is on charging. Current sensing is very expensive so the trick in these low-cost devices is always to just do as little as possible of it. So that's why you're not getting the mAh estimate after a discharge, only after a charge.
 
Just some perspective on this machine: at full power, presumably it's drawing about 40W from the wall to charge 2 2100mAh 1.2V (i.e. ~2.5Wh) cells up to 70%. In that time it consumes 10Wh to get a total of 3.5Wh into batteries. That's pretty shitty, no wonder it's go to vent all that excess energy!

[Zbig]

So that's why you're not getting the mAh estimate after a discharge, only after a charge.

You do, actually. That's how the "Test" function works.

[gorf]

You don't need to have independent current measurement during discharge. Use the same current measurement that you used during charge. Regulate the DC/DC so that the voltage on the shunt resistor drop is +ve during charge and -ve during discharge. ie the dc/dc will output 1.5ish V/cell out during charge and 0.7ish V/cell during discharge. Current flows both ways through a FET when it is on.  Probe the Fet gates to check which ones are on during charge and discharge.

[mikerj]

If you have four cells in series being charged, then you're going to need somewhat over 6v (w.r.t ground) to switch the top MOSFET on, so presumably there is some kind of gate driver circuit rather than the gates being driven direct from the micro?

Do the voltage sense connections go into diff-amps or just to potential dividers with the diff operation performed in software?

I also assume the series resistors for the AAAs only indirectly limit the current, e.g. the micro regulates the current to ensure some maximum cell voltage isn't exceeded, and the resistors increase the apparent cell voltage of AAAs during a charge cycle.

[AF6LJ]

I have a little different take on what is going on in the charger.
Only one cell is being charged at any given time.
They are multiplexed.