Charging NiMh battery pack with BQ2004

[Anard]

Hi.
I'm trying to charge a 18V/3Ah NiMh battery (15 cells x 1.2V) with the BQ2004E component.
So I have studied the BQ2004 datasheet and their switch-mode developpement system to adapt it to my specific case.
I tied TM1&TM2 to GND for selecting a C/4 charge. My Rsns is a 300mΩ/5W (Icharge = 0.225/0.3 = 0.75A = 3000/4).
I tied INH and VSEL to Vcc for choosing a PVD termination. I selected the number of cells with N = (780/120 + 1)*2 = 15 cells.
A 10kΩ NTC thermistor is placed on the TEMP connector.
So here is my circuit.

It seems to work, when I connect my DC Alim (30V/6A) it goes in "Fast-charge pending" mode for any seconds, then goes in "Fast-charge" mode. But the "Charge complete" mode comes very early (about 5-10 minutes whatever it's initial charge). I've no more that 18.6V at the terminals of battery after charging.
I'm very newbie and don't know how to find where could be the problem.
Can you see anything wrong on my circuit and could you help me to make tests to highlight the issue ?

Thanks very much for your help

 

[BobK]

Did you calculate RB1 and RB2 correctly? They should divide the battery voltage by a factor of 7.5 since it is expecting Vbat to be the voltage across 2 cells.

I don't see anything in the datasheet about what kind of thermistor to use and what values for RT1 and RT2. How did you come up with these?

Bob

 

[Anard]

Hello and thanks a lot for your answer.
For RB1 and RB2, I use normalised resistors, so 2 serial resistors for RB1 (680+100kΩ) and a 120kΩ for RB2 (I note them RB1, RB2 and RB3 on my circuit). => 780/120 = 6.5 and they give this formula : RB1/RB2 = (N/2) - 1, so for 15 cells, RB1/RB2 = 6.5. You don't agree with this result ?

For the thermistor, I don't understand the choice of resistor RT0, RT1, RT2 and RT3, so I just follow the technical document from TI DV2004. There is a clean shematic on page 5. I just rename their R2 in RT0 and I use a normalised resistor for RT3 (3.9kΩ in place of their 3.92kΩ). My thermistor is this one. It isn't one of the references listed in the DV2004 on page 2, but it's a 10kΩ NTC as recommanded. I place it between 2 cells with a 10cm wire, to prevent overheating due to the board's temperature, which is placed against the battery.

Please excuse my poor english…

 

[Anard]

One precision for the choice of RB1/RB2 : the datasheet note that

This resistor-divider network input impedance to end-to-end should be at least 200kΩ and less than 1MΩ.

In my case : 780+120 = 900kΩ.
Maybe it's too much, do you think I have to consider the low-pass filter made with RB0 (510kΩ) and CB (1µF), and how could I calculate the right network if I have to consider this filter ?

 

[BobK]

For RB1 and RB2, I use normalised resistors, so 2 serial resistors for RB1 (680+100kΩ) and a 120kΩ for RB2 (I note them RB1, RB2 and RB3 on my circuit). => 780/120 = 6.5 and they give this formula : RB1/RB2 = (N/2) - 1, so for 15 cells, RB1/RB2 = 6.5. You don't agree with this result ?

]
Assuming you mean 680K and not 680Ω, your calculation appears to be correct.

The datasheet says that it holds off for a while before checking the termination conditions. It sounds like it terminates as soon as it starts testing, which means one of the conditions is true right away.

I would suggest monitoring the voltages on BAT and TS to see why it is terminating. Do you have a storage scope?

Bob

 

[Anard]

Thanks for your advice. Yes it's 680kΩ, my parentheses were misplaced, I'd have wrote (680+100)kΩ

I had seen this mention in the datasheet about the hold off period, but I hadn't considered that this could last several minutes.

Anyway, I had to advance the work on the system itself, so I have charged the battery by an external charger. Now, it's afresh emptied, so I'm going to make new tests on my embedded charger.
First I've disconnected the (battery+charger) from the system to test it without other parameters.
Before connecting it to the DC Power, I've tried to mesure some resistors.
My multimeter doesn't read the real values on RB1,RB2 and RB3. I don't know why...
But as you said, I'll try to make some mesurements on Vbat and Vts with a DCPower connected.
For this, I've just one multimeter and one little numeric scope, I hope it'll be sufficient.

 

[Anard]

Right. Empty and when disconnected from load, the battery pack mesured about 16V. With load, it dropped fastly to 13V.
I connected the power and mesured :

• Vcc = 4.97V
• Vtco = 1.3V
• Vts = 1.83V
• Vbat = 1.32V : it was mesured with the scope but it doesn't give exaclty the same value as the multimeter...

Vbat slowly increased up to 1.60V. After 2 minutes Vts descreased quickly to 1.72V before stabilizing.
After 35 minutes, the end of charge appeared, but I didn't saw if it were a Vmax, a descrease of Vbat or Vts. After charge :

• Vbat = 1.53V
• Vts = 1.79V
• Vbattery = 19.83V

Everything seems normal execpt the charge time. For 3000mAh with a C/4 charge, I was expecting about 4h ... Also, 19.83V seems few. I wait about 1.45V per cell, so 21.75V max (I note that with commercial external charger, I have between 20.8V and 21.6V when fully charged).

 

[BobK]

Those numbers cannot be right. If the battery voltage is 19.83 then Vbat should be 18.83 / 7.5 = 2.57. Are you sure about the resistances in your divider?

Bob

 

[Anard]

For part, I think it's due to be the difference between my multimeter and my scope (Vbat was mesured with the scope). With the same instrument (multimeter), I have now : Vbattery = 19.60V, Vbat = 2.26V, both mesured in relation with Vsns (-Batt). But according to your equation, i'd have 2.48V.
Impossible to mesure correctly the resistors but the references (color code) are OK. Maybe I should disconnect the chip and the battery from the board in order to mesure it ?

EDIT : I'm searching but values or my resistors are senseless. I doubt about my multimeter.
Anyway, even if there's a short juice anywhere in my voltage divider, why did it cut the fast charge ? The conditions for fast charge are Vedv < Vcell < Vmcv
0.4Vcc < Vcell < 0.8Vcc
1.97 < Vcell < 3.94
The voltage I mesured was Vcell (Vbat-Vsns). So conditions are OK for fast charge. If the chip stopped the charge, there was a PVD (Vbat descreases), a -ΔT or a peak temperature. Am I wrong ?

 

[BobK]

Actually, I now remember that the voltage divider has resistance in the hundreds of K. The multimeter is probably 1Meg input resistance, so you will not get an accurate reading.

The Thevenin resistance is 1.04K, So if the meter is 1Meg you are getting a reading that is about 10/11 of the real reading. This would put it at 2.48V, which is correct!

Bob

 

[Anard]

Thanks. I confess I don't understand theses calculations, but if you say it's OK, I trust you

So.
I made a new test. Assuming that mesurements were correct, I deduce that end of charge is triggered by PVD, -ΔT or peak temperature. So as it's explained in DV2004 document, I replaced my 10kΩ NTC thermistor by a fixed 10kΩ resistor. The charge would theoretically stop only if the chip detects a Peak Voltage (Vcell falls by 6mV).
The Vtemp voltage (Vts-Vsns) were 1.79V, as in the end of previous test. It seems to be a nice condition for fast charging according to the datasheet :

Vhtf<Vtemp<Vltf, where:
Vltf = 0.4Vcc±30mV
Vhtf = [(1/3Vltf) + (2/3Vtco)]±30mV

In my case : Vltf = 1.972V, Vhtf = 1.551V

I mesured Vcell (Vbat-Vsns) with my meter and follow Vsns (Vsns - GND) with the scope.
After connecting the power, Vcell was about 2.38V and grews slowly to 2.39V (I have not a 6mV precision...) during hold-off period. During this time, I clearly hear the noise from switching signal. Here is the Vsns signal, the value on the right-bottom corner is peak voltage :



After 30s to 1m30, it goes in fast charge, the switching noise disappears. Then Vcell rises to 2.40-2.43V during a short time (less than one minute).


After which it goes in end of charge, Vsns is about 0V and Vcell goes back around 2.40V.
So I excluded an error on temperature detection. Even with the scope, I can't visualize the 6mV drop of Vcell but it seems to appear, else I can't understand the end of charge status.

However, I don't believe that the battery is fully charged. A 30min C/4 charge isn't normal.
After my first tries (Feb 22), I connected the battery to this external charger which is given for detecting -dV (8mV). It's nearly the same conditions, with a little more current (0.9A in place of 0.75A) but it's top-off charge charge came after about 3 hours…

 

[Anard]

Is it normal that when Power is off, voltage on Vbat pin is lower (0.5 - 0.8V) ? I feel it strange cause it's directly connected to battery…

 

[Anard]

I'm thinking to another difference between my board and the DV2004 on which I based my PCB.
Their toroid inductor L1 is very big and I prefered something smaller (and lighter) but I don't know what is the concrete difference between these two references and what is its influence (except smoothing current). I choose the same inductance value (100µH).

You can see their choice on the first page of the document.
Mine is this one, here is a shoot of my home-made board :

 

[Anard]

Hi.
I haven't solved my premature charge interruption and don't know more where to search.
If you have some ways to solve it, I'm still interested…

I also noticed that when pluged without battery, the charger does the same : it indicates first "waiting for conditions" for some seconds, then "charging" for about 1 to 5 minutes, whereupon it indicates "end of charge".
According to datasheet, it should turn off the 2 leds when no battery is connected. Maybe it could be a way to search.

 

[Anard]

?
When you select 12 cells and connect a 10-cells battery, or when no battery is connected ?

 

[BobK]

I'm thinking to another difference between my board and the DV2004 on which I based my PCB.
Their toroid inductor L1 is very big and I prefered something smaller (and lighter) but I don't know what is the concrete difference between these two references and what is its influence (except smoothing current). I choose the same inductance value (100µH).

You can see their choice on the first page of the document.
Mine is this one, here is a shoot of my home-made board :

I think you might be on to something. If the inductor is far smaller and lighter, it probably has a lower max current. If it is not sufficient for the charging current, it might be saturating, causing a large ripple in the voltage, which might trigger the deltaV condition.

Bob

 

[Anard]

Thanks for your answer. But my choice was given for 1,7Amax (here is the datasheet and my ref is the ELC10D101E).
Knowing that my fast charge current is configured as 0,75A. Do you mean it could saturate on transient currents ? And if so, how could I choose the right inductor (how to calculate the minimum current it have to admit).
So sorry for these newbie questions (and my SO poor english )...

 

[Anard]

Hello.
I just relpaced my inductor by this one, gived for 4.4A max.
I'm very surprised to see that charge stopped now after less than 10 minutes with empty battery pack...

 

[Anard]

Hi.
It's the first time I use self in a board. Do you think it could be polarized ? Datasheet of my first inductor (1.7A) dissociates poles "S" and "F"…

 

[farukh037]

when disabling the voltage termination mode by connecting VSEL pin to GND or VSS, IC take complete time (650 min) to charge the battery pack. otherwise by leaving VSEL as floating (negative delta voltage termination) IC terminates the fast charge within 10-30 mins.