# SLA Battery float charger w/ LM317T

Thread Starter

#### Wendy

Joined Mar 24, 2008
21,929
A long while back I showed a fence charger I had designed. Wookie mentioned the float charge circuit was flawed.

Here is what that section looked like.

Here is the same schematic revamped.

The idea is the battery is kept fully charged, but is used to provide a large rush of current if needed. The actual discharge is quite low (30ms pulse every second). I got the idea from working on gap discharge units, where large amounts of currents are needed for extremely short durations.

I basically picked R13 out of the air (100ma max) and would like some feedback on it and the rest of the concept. Thanks.
.

#### Attachments

• 3.2 KB Views: 876
• 3.3 KB Views: 1,258
Last edited:

#### marshallf3

Joined Jul 26, 2010
2,358
Float chargers can be tricky, do it wrong and the battery won't last long. Do you have any specs on the actual battery you'll be using? They usually recommend float charge voltage and current.

Thread Starter

#### Wendy

Joined Mar 24, 2008
21,929
The voltage is easy, you adjust the regulator to the voltage of a fully charged battery via R3. R13 limits the current, usually it is going to be a fraction of the max. If you shorted the output of R13 the max out you would get is 100ma, generally it will be a lot less.

I'm using an off the shelf SLA unit similar to what you use for a UPS.

#### SgtWookie

Joined Jul 17, 2007
22,210
Hi Bill,
I've modified your schematic a bit.
V1 and D1-D4 represent the plug supply.

R1 ensures a 10mA+ minimum regulator current for guaranteed regulation.
R3 sets the basic output voltage of the charger.
R2 is the "fine tune" for output voltage.
R4 sets the maximum charge current. When Vbe of Q1 reaches ~0.62v, it begins to conduct enough current to lower the LM317's output voltage to decrease the output current.
To calculate the value needed for R4, use 0.62/Iout, where 0 < Iout <= 1.5A

RSbat and RPbat are very rough approximations of battery parasitics.
The battery was simulated as BAT; simply a capacitor.

Note that the charge current is fairly constant right up to when the battery is nearly charged.

With the circuit you proposed, the charge current is non-linear.

#### Attachments

• 57.1 KB Views: 321
Last edited:
Thread Starter

#### Wendy

Joined Mar 24, 2008
21,929
OK, thanks Wook. The unregulated wall wart is pretty much a fixed part of the design, but I don't see it matters much either way.

From what I'm seeing your treating the battery as a 1000µF capacitor. Don't think I've seen that approach too often.

My thought was to charge the battery with a modern car charger, then use the circuitry to keep it topped off. If I ever make a gap welder I'd use the same technique, it was how Hughes did it (using 8 or more matched germanium transistors). I suspect MOSFETs could do it better.

If the battery was charging Q1 would start to conduct, dramatically dropping the overall voltage. You have a max charge current of 250ma?

Last edited:

#### SgtWookie

Joined Jul 17, 2007
22,210
OK, thanks Wook. The unregulated wall wart is pretty much a fixed part of the design, but I don't see it matters much either way.
Not really. Minimum specs for the wall wart would be 16V @ 160mA. If >150mA, you could use a 16v wall wart, as the actual output voltage will be higher with a lower load.

From what I'm seeing your treating the battery as a 1000µF capacitor. Don't think I've seen that approach too often.
Actually, it's pretty common to use a cap to simulate a battery being charged. It doesn't work well to simulate a battery being discharged; the model gets quite a bit more complex. I just used a 1mF cap to keep the simulation run time short.

My thought was to charge the battery with a modern car charger, then use the circuitry to keep it topped off.
ACK! Don't do that, as a car battery charger has far too much output current for an SLA battery. You'll likely overheat the battery. If it's a gel-cell, you will create gas pockets in the gel, which will permanently reduce battery capacity.

Follow the manufacturer's recommendations provided in the datasheet for your particular battery. Note that the datasheet should provide temperature compensation information; it'll likely be around -3mV per cell per °C deviation from 25°C, or -18mV/°C for a 6-cell 12v battery.

I posted an Excel spreadsheet in the "Tips and Tricks" thread in the General Electronics Chat forum, here: http://forum.allaboutcircuits.com/showpost.php?p=262143&postcount=38
The spreadsheet gives you %charge vs battery core temp, and suggested cycle(equalization)/float voltages.

If I ever make a gap welder I'd use the same technique, it was how Hughes did it (using 8 or more matched germanium transistors). I suspect MOSFETs could do it better.
Gap welder? Do you mean spot welder? I don't know what a gap welder is.

We had a Hughes spot welder at a previous employer. I was surprised that it was still working after a great many years of nearly constant use. Didn't know they used Germanium transistors in them. No point in using Germaniums nowadays unless you're dealing with very low-power RF signals.

John posted an interesting spot-welder circuit that uses a high-power SCR instead of transistors/MOSFETs.
Link: http://forum.allaboutcircuits.com/showthread.php?t=24385
Just charge the input caps to provide the welding energy needed, then disconnect the charge source and trigger the SCR. Clean and simple.

If the battery was charging Q1 would start to conduct, dramatically dropping the overall voltage. You have a max charge current of 250ma?
There's an initial spike because in the simulation, the battery starts at 0v. In the real world, a lead-acid battery that was discharged to 0v for any appreciable length of time would be a goner. Discharging a battery below 85% results in decreased service life. The deeper the discharge, the shorter the service life. You really should never discharge any lead-acid battery below 70%, as the cost of battery replacement will be excessive.

The circuit is about as simple as possible while providing a reasonably constant current and positive cut-off.

A very worthwhile enhancement would be adding temperature compensation; this would increase battery life, particularly if the battery is being operated in a non-temperature-controlled environment.

Another worthwhile enhancement would be to incorporate an "equalization charge" phase, where the charge voltage was raised by perhaps 8% for 20-30 minutes every 24-48 hours. This will remove plate sulphation and stir the electrolyte.

You should never attempt to equalize a gel-cell battery, because of the gas pockets that will result, and there is no practical way to remove the gas pockets. A centrifuge might work, but the plates would be subjected to a great deal of stress, and would likely collapse before the gas pockets migrated out.

Last edited:

#### SgtWookie

Joined Jul 17, 2007
22,210
OK, here's a start at providing temperature compensation for the terminal float voltage.

R3 in the prior version is replaced by Q2, R3, and R5. Q2 must be thermally coupled to the battery positive terminal. This is a very simple and cheap (1 resistor, 1 transistor) change to get a greatly improved charger.

Values for R3 & R5 were selected to give a reasonably close match for the temp compensation required. As the battery temp goes up, the float voltage is decreased.

Decreasing the resistance of R3 and R5 while keeping the ratio of resistances proportional will decrease the temp compensation effect. Total resistance of R3+R5 should not be reduced below about 2k Ohms.

Changing the ratio of R3 to R5 will change the voltage drop across Q2, thus affecting the output voltage.

Keep in mind that this is a simulation, using ideal components. Your mileage with actual components may vary - considerably. However, it is a good starting place.

#### Attachments

• 62.1 KB Views: 238
Thread Starter

#### Wendy

Joined Mar 24, 2008
21,929
This is a modern charger is programmable for smaller types of batteries. A lot of different types, not one of the older one size fits all types. Judging from the weight it doesn't have a transformer either. Gell cell is in its menu.

Hughes called them a gap welder, my old shop called it a gap welder, my new shop (which also has the same make/model) calls them a gap welder. Looked up the listing, HUGHES PARALLEL GAP WELDER;MODEL #MCW-550-CO2. I suspect this is a piece of clean room equipment. Long since obsolete, they are still in service, so Hughes must have done something right. You adjust the gap between the electrodes, set the voltage (thumb wheel switch, 10mv resolution) and the duration (10ms resolution). It used resistors on a rotary switch, and a lead acid battery (probably a gel cell) kept on floating charge similar to my scheme. There is a touch sensor, you bring the electrodes down to the work and bzzzt, instant weld. We all look to our backgrounds for projects I would guess. PITA to replace all the transistors, but they were simply wired in with wire.

I'll keep the temperature compensation in mind. I hadn't really heard about a lot of what your talking about concerning batteries. This will likely spend days or weeks outside, protected from direct weather but open to the outside air and temperatures. My thermometer at my house has been running 104°F to 108° for the last two weeks. Fortunately it cools off to 85° at night.

BTW, the chips are in the mail.

#### marshallf3

Joined Jul 26, 2010
2,358
My thermometer at my house has been running 104°F to 108° for the last two weeks.
Stop gap welding in your kitchen.

#### SgtWookie

Joined Jul 17, 2007
22,210
Stop gap welding in your kitchen.
It's his LM358's - they're sweating bullets.

#### k7elp60

Joined Nov 4, 2008
561
Bill,
Here is a 12V SLA charger I developed about a year ago. R1 limits the charge current and there is a circuit that turns on an LED when the charge current drops to the capacity/100. Most manufacturers say at the float charge voltage(13.5 to 13.8) when the charge current drops to C/100 the battery is charged. The 1N5817 on the output is there in case the line power is lost the battery will not damage the LM317 or be discharged by the circuitry. The 1N4002 on the input is to prevent incorrect polarity from the wall wart.
There is a 3 terminal 12V charger IC that is designed for 12V lead acid batteries. It is the PB137. It's internal current limit is greater than 1A, so for smaller SLA it may not be suitable.
Ned

#### Attachments

• 69.7 KB Views: 248

#### SgtWookie

Joined Jul 17, 2007
22,210
Here's an update; basically just a change of R3/R5 values. Guess I was talking through my hat earlier about not going below 2k total; 1100 and 82 Ohms make for a really close match for a 6-cell lead-acid battery in the simulation.

One important change to note is the new location of the ground reference; this avoids problems if other areas of the circuit past the battery are grounded.

#### Attachments

• 59.1 KB Views: 319
Last edited:

#### SgtWookie

Joined Jul 17, 2007
22,210
K7elp60's schematic is quite similar to what I'm proposing, except it is not temperature compensated, and the output voltage adjust is much more coarse.

The Schottky diode on the output is a good idea. It'll have between 0.15v and 0.28v drop across itself, depending upon charge current and temperature.

[eta] The PB137 is really only suitable for batteries that will be in a temperature-controlled environment. With your current high outdoor temps, the battery internal temperature will require a lower float voltage than the PB137 provides.

In the winter, 13.7v will not be high enough to keep the battery at the proper float voltage level.

Last edited:

#### sage.radachowsky

Joined May 11, 2010
241
Do you think that supercaps could supply the pulse current required?

You could stack up some 100F caps in series -- 5 of the 2.5V ones will get you up to 12V and then you have 20F at 12V. I think they're about $20 each these days, so that's$100. Expensive but no worries about battery chemistry or degradation. Just some ~5K resistors in parallel to balance them out.

#### iONic

Joined Nov 16, 2007
1,650
Let's bring this topic back to life.
I tend not to do a lot of driving, and when I do, most of it is for very short distances. I rack up an astounding 4k - 5k miles a year. This brings me back to Bill's/Sgt's float charger circuit with temperature compensation. I really want to call it a battery maintainer as my intended use is to just keep the battery at 13.xV with a small compensation current to override the natural losses of a hermit car battery. I live in the north-east and have the gamut of temperature ranges (under the hood in the some from 100 deg F down to -20 deg F in winter. So just keeping a constant 13.x voltage applied will not be as effective as adding compensation for temperature differences.

Questions regarding the last posted circuit by Sgt., Post #12.

Are the 20M RSbat bat, the 1uF cap and the 100K RPbat part of the circuit design or rather there for the purposes of the simulation?

Do you really think that with just a maintenance voltage and very low current that there is a significant difference between the batteries core temperature and the ambient temperature, provided a fairly well enclosed system?

I am thinking of using the LT1083 but got a good deal on the 1085 for the regulator.
It's a tad more efficient than the LM317 and I wouldn't think a heat sink wouldn't be necessary.

Thanks

#### SgtWookie

Joined Jul 17, 2007
22,210
I later updated the schematic some more, so I'll re-post it.

See the attached; there are comments in the schematic.

Are the 20M RSbat bat, the 1uF cap and the 100K RPbat part of the circuit design or rather there for the purposes of the simulation?
They are there to represent a battery. I've since created a much better model of a battery, but a cap and a couple of resistors works OK for the purpose of this simulation.

Do you really think that with just a maintenance voltage and very low current that there is a significant difference between the batteries core temperature and the ambient temperature, provided a fairly well enclosed system?
The core temp of a lead-acid battery is fairly stable. Once you heat it up, it takes a good while for it to cool down. The battery temperature will wind up averaging about the same temp as the environment it's in, plus whatever power is being dissipated inside the core via charging, discharging, chemical activity, etc. But, the battery is a large mass, and it's in a plastic case. It will take a good bit of time for the heat exchange to take place.

Note that for best results, the 2N3904 transistor should be thermally coupled to the battery positive terminal.

I am thinking of using the LT1083 but got a good deal on the 1085 for the regulator.
It's a tad more efficient than the LM317 and I wouldn't think a heat sink wouldn't be necessary.
Don't kid yourself. Those are still linear regulators, and they will dissipate power as heat just like an LM317 will when passing current with a voltage drop across itself.

The LT1083-LT1084-LT1085 family has a lower dropout voltage than the LM317 does. HOWEVER, if the differential between input and output voltage is the same for the current passed, they will dissipate exactly the same amount of power.

Also, this design was LM317-specific; it does not include the mandatory 10uF tantalum capacitor that the LT108x family requires. If you are going to use the LT1085, then you will need to use that cap. You will also need a heat sink.

If you want to get away from needing a heat sink, then you will need to go to a switching regulator.

#### Attachments

• 126.7 KB Views: 133
Last edited:

#### iONic

Joined Nov 16, 2007
1,650
Well, since I am planning to try out the BQ2031 I will then stick with the LM317, that is if I can find my bag of Adjustable Voltage regulators.

#### #12

Joined Nov 30, 2010
18,173
Just throwing this at the wall without reading everything in the thread...

If you use a 317L it will limit itself to 100 ma. Well, that's what the datasheet says. I have seen them do 130 ma. I have seen them used as regulators in a retail garage door opener, so I know they can do this continuously for years. If/when they get hot from current limiting, the overheat safety kicks in too.

If this design won't hit the wattage limit on the 317L under normal operating conditions, you could remove some complication by using the 317L as a self-current-regulator.

#### iONic

Joined Nov 16, 2007
1,650
Just throwing this at the wall without reading everything in the thread...

If you use a 317L it will limit itself to 100 ma. Well, that's what the datasheet says. I have seen them do 130 ma. I have seen them used as regulators in a retail garage door opener, so I know they can do this continuously for years. If/when they get hot from current limiting, the overheat safety kicks in too.

If this design won't hit the wattage limit on the 317L under normal operating conditions, you could remove some complication by using the 317L as a self-current-regulator.
Am I right to doubt the capability of the LM317L in Bill's design above as it is temperature compensated and therefore the voltage requirements can be low or high, requiring a voltage source approaching 20V. With 20V in and possibly as little as 12V out I would think there might be some heat issues.
Wouldn't want the built-in heat safety mechanism of the tiny T0-92 pkg to work like an oscillator!...although I was considering a "Pulsed" maintenance charging routine.

#### #12

Joined Nov 30, 2010
18,173
Looking this up, I'm finding 200 to 300 ma capability, not the 100 that was stuck in my mind. Probably from a guaranteed current delivery spec on a 78L05 or something. Looks bad for use in this circuit as a current limiter.

The heat thing goes like this, 160C/W J-air and 180C thermal shutdown.
from 25C ambient, temperature gain limit is 155C or 155/160 of a watt in a TO-92 package. When the thermal limit is reached, the chip doesn't go into PWM (LOL). It tends to scrunch down the current so the junction temperature stops rising. The junction holds at 180C on a package that is rated at Tj = 125C max??? This is apparently good enough for a retail product because the manufacturer doesn't mind if it fails and you have to buy a new machine.

This looks like a bad idea that I posted because the current limit is way higher than I thought and living in overheat limiting is just a bad idea if you want the machine to be reliable.

Similar threads