I need some help in analog land!

[justDIY]

Having come into a large hoard of VRLA (valve regulated lead acid) batteries, I wanted to build a charger to keep them in tip-top shape.

Searching far and wide, I was presented with plans simple to very complex.   I settled on a plan designed by some guy for use in recharging gel-cells that powered his guitar amp.  The original is found here.



The design is based somewhat on refrence designs in the datasheet, but there aren't any that are an exact match.  This charger offers a constant voltage, set by R2/R1 as usual, but the feedback voltage is also modified by the transistor, Q1, which monitors the rate of charge.   Ideal RoC for absorbed glass mat (what most gell cells are these days) is around 1/10 C ... so in my case, 700 mA for a 7aH battery.  The transistor I chose for my circuit offers a 650 to 850 mv saturation voltage, so I chose 700mV as sort of a middle ground.   My design differs mainly in the filtering capacitors ... from the research I've done, the occasional transient will do the battery good.   I figure 200uF should do a good job of filtering out anything major, and the 1uF ceramic on the output is per OEM recommendation to prevent any oscilation.

The circuit is working well on a breadboard, and is topping-off my pile of VRLAs right now, two at a time (batteries in parallel).

What I would like to add before I print this design on copper, are some LEDs to indicate the charger state.   I know I could put an op-amp and a microcontroller on R4, do a quick ADC and once I'm in digital land, the sky is the limit... but I'm wondering if there's an easy way to add some indication.    Ideally I would like a three state indicator:  red = bulk charging (current near 1/10 C), yellow = trickle charging (current less than 1/50 C), green = float charging (very little current).   The catch here, of course, is I would like this to be flexible... probably based off R4 or R2 somehow?   Those are the only parts that will change if I'm building this for a 6v 21ah array or a 24v 14ah array or something in between.

Any thoughts?!

here's my current pcb layout ... it measures 2.5" by 1.5", 17 holes, 4 active components, 7 passive, 3 mechanical, 17 drill-holes, single sided

[justDIY]

A few more details:

Input voltage:
(I know my original post said I want to charge a 24 volt array... well, I haven't run across a capable surplus transformer yet)

For the 12v batteries:
24 VAC -  I ended up with a good deal on a 60 watt 24 volt transformer

For 6v batteries
12 VAC - Another good deal on a 36 watt 12 volt transformer

Regulator IC:

Probably the LM338 LM1084 - they're cheap at bgmicro and pin-compatible with the LM317, but handle up to 5 amps

[Rob]

Ideally I would like a three state indicator:  red = bulk charging (current near 1/10 C), yellow = trickle charging (current less than 1/50 C), green = float charging (very little current).   The catch here, of course, is I would like this to be flexible... probably based off R4 or R2 somehow?   Those are the only parts that will change if I'm building this for a 6v 21ah array or a 24v 14ah array or something in between.

I think you have the right idea with the comparator. You can set it up to measure the voltage across the resistor R4.

If you used a single red/green bicolor LED, you could hook the output of one comparator to the green side, and the other to the red side. Turn on green when the voltage across R4 drops below some setpoint (set by a trim pot I suppose), and turn on red when V_R4 rises above some setpoint that is below the first setpoint.

that gets you
 V_R4 < V_set1 green
 V_set1 < V_R4 <V_set2 yellow
 V_R4 > V_set2 red

I think I got that right. Since the input impedance is real high, the comparators won't load the charger. You can get chips with multiple comparators on them, or just use a bog-standard op-amp. Comparator mode only needs a positive supply (yay). So the parts count is 2 trimmers, one op-amp IC and the indicator LED.

How's that?

[cpemma]

... from the research I've done, the occasional transient will do the battery good.   I figure 200uF should do a good job of filtering out anything major, and the 1uF ceramic on the output is per OEM recommendation to prevent any oscillation.

What you'll get with small capacitors is a sort of pulse charging, the regulator ingoing voltage will drop somewhat  below its drop-out level every half-cycle so the output will be a flat-topped (clipped) wave. It won't (or shouldn't) go any higher than the regulator setting.

In effect, this shape but with the peaks cropped. It does mean your LEDs could flicker continuously.

[justDIY]

ok, I see your point ... the capacitors don't have sufficient holdup time to carry the load during the low points in the wave, so the vreg drops out, and charging stops until the wave comes back up.  So if I wanted to prevent this, I'll have to figure out how big a cap I'll need to carry the load.

Happen to have the forumla for that calculation handy?  Maybe Rob has a wizard waiting in the wings?

I plan to use an vldo regulator as the current source for the LEDs and to power the vref dividers ... I'll setup comparator outputs as open collectors.  This way I can get away with a smallish cap to carry the leds and logic during the down time, not that I'd really mind the a 60hz flicker either... I mean, it is just a battery charger.

[justDIY]

I think you have the right idea with the comparator. You can set it up to measure the voltage across the resistor R4.

If you used a single red/green bicolor LED, you could hook the output of one comparator to the green side, and the other to the red side. Turn on green when the voltage across R4 drops below some setpoint (set by a trim pot I suppose), and turn on red when V_R4 rises above some setpoint that is below the first setpoint.

that gets you
 V_R4 < V_set1 green
 V_set1 < V_R4 <V_set2 yellow
 V_R4 > V_set2 red

I have my breadboard setup with 3 trimmers, 2 of which are vref dividers for the comparators, the 3rd (VR3) emulates the sense resistor.

I have the red vref (VR1) set to turn the red led on for voltages over 600mV ... indicating a high rate of charge
I have the green vref (VR2) set to turn the green led on for voltages under 250mV ... indicating a low rate of charge
However, when the charge is intermediate, between these two points, neither LED is lit.

How do I get yellow?   If I use a third comparator, I could wire it to both red and green, and set it to come on when the voltage is below 600mV, but then red will still be on after the voltage drops below 250mv.

fyi the comparator i'm using is the LM339, a quad comparator with open collector outputs.

here's the schematic of my current setup

[justDIY]

well, thinking about it a little more, I think I have a solution

this is the current setup:
comparator one goes low (turns the red led on) when the voltage is => 600mV and
comparator two goes low (turns the green led on) when the voltage is <= 250mV...

so if I connect an n-mosfet to both these comparator outputs, using diode AND gates, and use something like a 10k pull-up to Vcc on the fet gate, the fet will be low as long as both comparators are open.   as soon as the sense voltage exceeds 600mV or drops below 250mV, a comparator will go low, the fet gate will be pulled to ground (weak pull up would be shorted wasting what, 0.5ma of current?).   this fet would then sink both the red and the green LEDs, resulting in yellow.  Since the LEDs cathodes would essentially be connected together on the fet's drain, now I have a problem of when either comparator goes low, it will also be sinking both LEDs... argh ANALOG!

I'm thinking the phrase "for every problem, there's a microcontroller solution." is sounding good right now.

a simple analog -> digital conversion refrenced on a 700mV (or maybe 800 for a little fudge factor) volt scale would give me a pretty accurate reading of the sense voltage, and then I could build as many 'digital comparators' as I wanted using If Then Else constructs.  The microcontroller could do other things like watchdog the battery temperature... or...

The uC could also be used to program the charging voltage, for example;

1) Read the battery terminal voltage and calculate the state of charge.
2) If the SoC is below 1.5v/cell, the battery may be damaged, alert user and require manual intervention. 
2) If SoC is low, go on to step 3, otherwise (SoC high) float charge indeffinately at 2.2-2.3 v/cell.
3) Low SoC means a cycle charge is required and the battery should be charged at a higher voltage (2.4-2.5v/cell).   
4) After 12 hours, interrupt charging and allow 1 hr for battery to normalize, and resample terminal voltage ... if it's still low, continue cycle charging for 12 more hours (repeat process for up to 48 hours), if it's now normal, contine float charging indeffinately.

[cpemma]

...not that I'd really mind the a 60hz flicker either... I mean, it is just a battery charger.

To clarify, I meant the 'trickle charge' LED could be turned on whenever the 'high rate' current fell enough due to the small caps being part-flattened to trigger the comparator. And 120Hz flicker, so to the eye on steady if dim.

Cap calc: C1 (uF) = [ ( IL * t )  / Vrip ] X 10^6

t = 1/120 = 0.0083 for 60Hz mains full-wave rectified
IL = load current
Vrip = permitted ripple voltage.

So a fairly standard 25V secondary should give you around 33V peak DC, I'm guessing 29V input will still be enough overhead to charge the cells, so at 750mA, C1=2000uF should be OK.

[Rob]

How do I get yellow?

red + green = yellow. Swap those two setpoints.

[justDIY]

ok duh

i was overthinking the whole ordeal, as usual

so I reversed the setpoints, and hooked up a proper multichip LED, and voila

thanks

although I might still build that uC powered unit, seems like it would be kind of handy.