NiCD battery charger circuit that drives me nuts

[_Vendetta_]

Hello EEVbloggers,

I am trying to design a rather simple NiCD battery charger, with just some voltage peak detection to cut off the charge at (4.5V).
Therefore, i tried to implement it using the TL431 voltage reference. And the circuit is as follows:



The problem is that the circuit behavior is fine on simulation. The circuit charges the battery at 100mA (C/15), However, in reality, unlike the simulation it does not stop the charge at 4.5V. it remains at 83mA.

Your suggestions are highly appreciated !

Thank you.

[CaptDon]

That circuit will never 'stop' the charge, it will only taper the charge down to satisfy the
current the batteries will draw at 4.5vdc which is going to be very high if you have 3
batteries in series who would be fully charged at 4.05vdc. If you want the circuit to
'cut off' it will need some sort of hysteresis to cut back the charge voltage to maybe
3.65vdc until the battery voltage drifts down to that level and then re-activate back to
the higher voltage. Your circuit is too simple, it needs more parts/features.
 

[_Vendetta_]

Thank you for your reply. Well technically, i will be satisfied with simply tempering down the charge current to 30mA when it is fully charged.
For the voltage of the battery, when it is fully charged it shows a reading of 4.15V. I would assume when charging it. The cut off occurs at 4.5 (1.5V per cell) due to internal resistance.

[CaptDon]

I think your voltages are a bit off. A fully charged NiCd after resting for a few days is
around 1.2 to 1.25 volts. I would never try to push more than 1.4 volts per cell. Even
at 1.4vdc the cells will heat and then at constant current will sink back to 1.35vdc. This
is an indication they are fully charged and converting the excess charge to heat. You
don't want to be on that part of the curve if you want any life out of your batteries.
That voltage sinking knee is how some chargers know the batteries are fully charged
and revert to a float charge of just a few mills. I feel your target of 4.5vdc is much to
high. 4.05 to 4.15 would be a better limit and even that would be high if the batteries
are hot.

[tooki]

Eh, for normal quick charging, 1.45V is right around where you expect the peak to be, according to the book on battery charging that I was reading last night! But you’d only do that when using actual peak detection, which isn’t really what’s being done here.

OP, if you’re not going to implement a proper charger, why not just make it an fixed constant-current trickle charger at C/50-C/30 or so?

[iMo]

I've been using similar charger with LM317+TL431. Works perfectly CI/CV.

[floobydust]

TL431, depending on manufacturer, can be unstable with 0.1uF C5 across it and oscillate. I'd remove it and see if things change. It's not a situation like the usual 0.1uF decoupling cap you put in for an IC's power. There are capacitance/stability curves on the datasheet and 0.1uF seems to be in the middle of the unstable region for TI, On-semi, ST.

[_Vendetta_]

I think your voltages are a bit off. A fully charged NiCd after resting for a few days is
around 1.2 to 1.25 volts. I would never try to push more than 1.4 volts per cell. Even
at 1.4vdc the cells will heat and then at constant current will sink back to 1.35vdc. This
is an indication they are fully charged and converting the excess charge to heat. You
don't want to be on that part of the curve if you want any life out of your batteries.
That voltage sinking knee is how some chargers know the batteries are fully charged
and revert to a float charge of just a few mills. I feel your target of 4.5vdc is much to
high. 4.05 to 4.15 would be a better limit and even that would be high if the batteries
are hot.

Well i have a couple of those sitting around fully charged. They all show values between 4 and 4.2V ! I have been charging it one for one week using this charger. I had no heat what so ever.

[_Vendetta_]

I've been using similar charger with LM317+TL431. Works perfectly CI/CV.

Would you mind sharing more details about your circuit ! thanks .

[Benta]

The biggest problem you have with this approach is, that the charge voltage of NiMH/NiCd cells will drop if you pass the peak voltage point. This will bring you into an area that's destructive for the cell. Even worse: charge voltage is temperature dependent.

That's why d²V/dt² algorithms were created.

This old data sheet gives a lot of info:
https://cdn-reichelt.de/documents/datenblatt/A200/U2402B%23TEM.pdf

[_Vendetta_]

TL431, depending on manufacturer, can be unstable with 0.1uF C5 across it and oscillate. I'd remove it and see if things change. It's not a situation like the usual 0.1uF decoupling cap you put in for an IC's power. There are capacitance/stability curves on the datasheet and 0.1uF seems to be in the middle of the unstable region for TI, On-semi, ST.

Nicely remarked ! thank you for grabbing my attention. In fact i tested the circuit without the decoupling capacitor. i am using one 1uF for the battery though.

[_Vendetta_]

The biggest problem you have with this approach is, that the charge voltage of NiMH/NiCd cells will drop if you pass the peak voltage point. This will bring you into an area that's destructive for the cell. Even worse: charge voltage is temperature dependent.

That's why d²V/dt² algorithms were created.

This old data sheet gives a lot of info:
https://cdn-reichelt.de/documents/datenblatt/A200/U2402B%23TEM.pdf

Thanks for the reply. I am aware of these methods. I am willing to terminate the charge before that peak and switch to trickle charging. ( C/15 intially then trickle it with C/50 ).