SLA battery charging voltages

[fsr]

Hi,

I'm on the process of improving an emergency light that likes to kill batteries too soon, so i did a search on the charging voltages, and found this:  http://batteryuniversity.com/learn/article/charging_the_lead_acid_battery

So, i was planning to make a 3 stage charger. The light uses a 6v 4AH battery. Basically i'm making a CC/CV charger (200mA/6.9v) that switches to 6.75v when the charging current goes below 120 mA.

But then i got the new battery (a yuasa np4-6), and the datasheet recommends charging with CV float charge at 6.825v, or 7.26v for cyclic use.
Most curious of this, is that some documents say that this battery is lead-calcium, which normally requires higher voltages.

What do you think? Should i charge at 6.9v then switch to 6.75, or go with 6.825 then 6.75, or just 6.825v permanently, or another combination? What would be the best to achieve the longest battery service life?

Regards

[NiHaoMike]

If you're using a microcontroller (easiest way I can think of to implement what you're trying to do), have it use the 6.825V setting most of the time, but switch to 7.26V and hold for 15 minutes or so every month or so.

[Ian.M]

Also, if you are using a MCU, its probably worth implementing temperature compensation of the float voltage, unless the charger is only going to be used in an air-conditioned environment with a narrow room temperature range.   Quite a few MCUs have an on-chip temperature sensor, which may be good enough if you locate the MCU carefully to avoid heating by near-by components, and to minimise its own power dissipation, otherwise a temperature sensor in contact with the battery case, is the preferred option. 

[Red Squirrel]

The nice thing about lead acid is that they can be floated indefinitely at a constant voltage charge.  You could make a constant voltage of 6.75v (2.25v per cell, a 6v battery has 3 cells) and just leave it.  It will charge slower after a power outage though.  So you could still do the multi stage charge cycle if you want it to charge faster.  But when it's done you can just leave it at 6.75.

And yeah can also do temp compensation.  The 2.25v per cell recommendation is based on 20-25C if I recall.   

If you do constant voltage without worrying about current it also means less components as current is more complex to measure.  Need shunt, op amp bunch of filtering etc.

This is basically how the telcos do it, they have large flooded acid cells, 24 in series, and float them at around 54v.  All the equipment runs off it in parallel. 

[tautech]

Of concern also is the charging rate which for SLA is generally C/10 so with a 200mA supply you're well inside that.

If it's to be charged continuously be sure to set the lower charge voltage of 6.825V accurately, err just on the lower side preferably.

[fsr]

Thanks everyone!

Well, that's easier than i thought! I had previously drawn a circuit with one lm317, one lm358 and some transistors, but if i only need to supply one constant voltage, then i can get rid of the opamps, and the current limit could be one transistor with a current sense resistor that sinks the ADJ terminal to GND (the LM317 has a 2.2A typical current limit, and the transformer isn't going to be that big). The temp compensation is going to be a vbe multiplier under the voltage set resistors to get about -9.2 mV/degree C.

At first i was going to use the transformer originally included. It's 9v 200 mA, even if that means that the charging current would need to be only 120 mA, but when i tested the transformer, it had 9v rms with no load, and only 8.1v rms with a bridge rectifier, capacitor and a 100 mA load, so i think that i'm going to buy another transformer for this, or maybe some off-line 12v module. Well, at least i can use the LEDs and the plastic box...

[fsr]

I still don't have the MOSFET, but it's likely that the circuit is going to be like in the attached image.

Unfortunately the LED board has only two wires, so the "brightness control" is not the best thing in the world, but at least the battery is going to last longer than directly connected  to them. The LED board is weird: one string of 27 parallel leds in series with another string of 24 parallel leds.
I suppose they chosen the Vf of the leds so that the battery voltage is under the point where the current "skyrockets", as the board connected directly to the battery seems to peak at about 500 mA.

[floobydust]

I tried a similar circuit and it did not work for me. This is in a hack project for a quick and dirty float charger, good to below -20C outdoors.
The self-heating of the temperature-sense transistor and diode always gave the wrong charging voltage.
Because R6 is low value, the divider-string current is quite a bit; look at Q2 power dissipation and its temp rise.
I ended up using an emitter follower and increased up to 1k ohm, I think it was -4mV/C tempco with the E-B junction and diode I got.

edit: your filter capacitor C3 is tiny at 33uF and will have a short life due to the ripple current.

[fsr]

Thanks, i didn't thinked too much about the divider's current. There's no real need for the 10 mA minimum load current to flow there, i can raise R6's value, and just sink the rest elsewere. Some Vbe multiplier recalculation will be in order, also.

[fsr]

I finished the project, it's hanging on the wall since probably more than a week now.
I'll attach the final schematic. The AC to 12v module and the LM317 were hotter than i thinked while the battery is discharged, but not nearly high enough to be a problem. The Vbe multiplier was located on the low side of the board, so that it doesn't receives any heat by convection.