Temperature compensation

[MikeMl]

I'm building a battery charger for lead acid batteries, and need to compensate the reference voltage to match the corrected charging voltage vs temperature.

Ideally, I need a nominal 12.00V reference which is reduced by 10mV per degC at temperatures above 25degC, and is increased by 10mV per degree at temperatures below 25degC, over the range from -20 to +40degC.

Ideas? (Extra points if it can be done with Si diode(s) or diode-connected transistor and an LM317, 78L05, or TL431--- trying to avoid yet another Digi-Key Order).

 

[Roff]

Mike, the way you described the data, it sounds like you need the slope to change at 25C. Where did you get that compensation curve data?
Here is a table that I found.
**broken link removed** is a graph.

 

[crutschow]

Actually I believe Mike is describing a monotonic decrease of voltage versus temperature.

The voltage across a forward biased silicon diode or diode-connected transistor decreases by about 2.3mV/C, thus if you amplify the voltage across the diode by 4.4 with an op amp you will get a voltage of approximately 2.8V that varies by about -10mV/C. You can sum an accurate DC voltage into the op amp input to get your desired +12V output at 25C.

Edit: If you let the +2.8V output of the op amp be the common point of the adjustment circuit of an LM317, and set it to give an output of 12V (9.2V from the regulator), then you will have a high current output of 12V with a -10mV/C slope. You could use a pot adjustment of the LM317 to give the exact output you need.

 

[MrAl]

Hello,

You could try simply connecting four properly biased si diodes in series to get the approximate slope, then add to another constant voltage source (diodes in series with it) to get the actual voltage, and that will end up with a slope of about -10mv/degC. You'll need a constant voltage reference of around 11 volts or so depending on how you intend to use your LA battery.

In fact, if you connect four si diodes in series with the lower resistor of a LM317 (lower resistor will be around 1700 ohms with upper 220) you'll get near -9mv/degC that way too, with little effort. If you are after more accuracy in the temp co, you can bias another diode with the output of the LM317 and use a couple resistors (like 50 and 25 ohms) to divide that fifth diodes voltage down by about 1/3, and connect the other four to the divider tap. That will be darn close to -10mv/degC and you can adjust the temp co a little by changing the 50 and 25 ohm resistors for slightly different divide ratio (a 100 ohm pot might help here if you are that picky). Total parts: five diodes like 1N4148, two low value resistors like 50 and 25, one bias resistor like 270 ohms, another resistor around 1700 ohms and another at 220 ohms, and of course the LM317.

We also dont know how much current you are after, ie what capacity LA's are we dealing with here.

There's a cheap little three terminal current regulator that has been used to create a 10mv/degC reference for temperature measurments too. LM385 or something like that.

My LA battery slope is -5mv per deg C so im not sure where you got -10mv/degC, but i guess the batteries could be different from different manufacturers. It's good to get the data sheet and double check.

BTW you need at least 13.8v to charge a 12v LA battery, not 12v because 12v wont allow enough charge current.

One little note of caution here though...
Diodes dont make the best temperature compensation devices. The completed circuit should be well tested for the expected temperature range. If it doesnt comply you may want to go with a temperature sensor for more accurate compensation.

 

[MikeMl]

Thanks for the suggestions. Here is a better description of what I'm trying to do:

This is for a Ford E350 V10 Triton-powered Motorhome I recently bought. It seems that the parasitic current drain to keep the ECU alive with engine off is ~31mA, which will kill the starting battery while parked for about 1 month. If you disconnect the battery while parked, the ECU looses its engine constants, and it reverts to the "limp-home" mode, where the engine runs like crap. If you need to get the state-mandated emissions inspection, you have to drive it 100mi just so the engine begins running well enough for it to pass the inspection... (What a @!#$$%^%%$ STUPID DESIGN)!!!

The solution will be a charger which will be plugged-in to the AC for the several months at a time while the motorhome is parked. I am planning to operate from a 24VAC 2A transformer, which will be in my utility shed, and the wire running to the motorhome will only have 24VAC on it for safety. That way, if I accidentally hit it with a snow shovel, I wont electrocute myself.

The charger will rectify the 24VAC and make the FLOAT voltage for 23.75hours, and will bump up to the CHARGE voltage for 15min per day, simulating driving the motorhome once per day. The CHARGE voltage is similar to what the alternator voltage regulator puts out, however, since the engine compartment wont come up to normal operating temperatures, ao the temperature compensation will follow the electro-chemistry based temperature correction.

The data for temperature dependence of the FLOAT and CHARGE voltages came from Dr Mark's web site, here . The attached sim shows the required temperature correction coefficients for a 6cell battery in the behavioral voltage sources, and the plots show the temperature dependence.

 

[ericgibbs]

hi Mike,
The LM335Z has a 10mV/Cdeg change in output voltage, I would consider using this as the 'measuring' device for the straight line slope from -20C thru +40C.

With reference to the Charging voltage it does seem higher than the data I have seen for 12V SLA's,, typically a charge is at 13.8V/14V and float at 13V/13.4V.

 

[MikeMl]

I'm looking at the LM335. Thanks.

The intent of the 15min per day CHARGE voltage is to briefly get the battery into the mild gassing range, primarily to stir the acid to prevent acid stratification. In car that is driven daily, the electrolyte is sloshed due to movement, but in my case, unless we have an earthquake, there will be no movement, so a slight bubbling will replace the mechanical agitation. The batter(ies) in question have caps, so I can periodically add distilled H2O if the electrolyte level drops...

 

[ericgibbs]

I'm looking at the LM335. Thanks.

The intent of the 15min per day CHARGE voltage is to briefly get the battery into the mild gassing range, primarily to stir the acid to prevent acid stratification. In car that is driven daily, the electrolyte is sloshed due to movement, but in my case, unless we have an earthquake, there will be no movement, so a slight bubbling will replace the mechanical agitation. The batter(ies) in question have caps, so I can periodically add distilled H2O if the electrolyte level drops...

I hadn't considered the lack of the 'slosh' factor...

If you can get a better fit zener diode, the BAT84 could be dropped.

BTW:
Did see this zip on Yahoo

 

[MrAl]

Hello again,


Yes it is the LM335 not LM385 as i remembered the part number wrong. The LM334 might work too with a little effort.
Of course the temp co is positive not negative, so you'd have to put it into a subtraction circuit to get the proper slope.
I think this is a better idea overall because it will be more accurate over that wide temperature range you are looking for (outdoors).

 

[ericgibbs]

hi Mike,
I guess you have already dropped the zener from the previous circuit and opted for a TL431.

Another option.

 

[bryan1]

Hi MikeMI,
The simplest solution I see if the ECU needs a current to stay alive is stick say a 5 watt PV panel on the roof, install a separate battery to power the ECU when the ignition is off and a simple charge controller circuit for the new RE install. When the engine does start a simple smart will turn off the RE supply and get you motoring. No need for complex circuits to monitor a AC voltage only a charge controller for the PV panel is needed.

In my opinion too easy

Regards Bryan

 

[RCinFLA]

Thanks for the suggestions. Here is a better description of what I'm trying to do:

This is for a Ford E350 V10 Triton-powered Motorhome I recently bought. It seems that the parasitic current drain to keep the ECU alive with engine off is ~31mA, which will kill the starting battery while parked for about 1 month. If you disconnect the battery while parked, the ECU looses its engine constants, and it reverts to the "limp-home" mode, where the engine runs like crap. If you need to get the state-mandated emissions inspection, you have to drive it 100mi just so the engine begins running well enough for it to pass the inspection... (What a @!#$$%^%%$ STUPID DESIGN)!!!.

Did you check the leakage over time? The car manf.'s are usually pretty picky about OFF current drain.

I thought I had a similar issue on my Ford and pulled every fuse out looking for leakage. It was the engine controller but I found that after about 5-8 minutes the leakage current dropped off to very low level.

12 v lead acid battery temp comp is between -30 to -33 mV / deg C during charging and about -12 mV/ deg C during float.

 

[MikeMl]

Did you check the leakage over time? The car manf.'s are usually pretty picky about OFF current drain.

I thought I had a similar issue on my Ford and pulled every fuse out looking for leakage. It was the engine controller but I found that after about 5-8 minutes the leakage current dropped off to very low level.

I was aware that some vehicles are supposed to revert to a lower parasitic current drain after the ECU goes into a "sleep" mode; however, this one does not. I was very careful to leave it set up for parasitic current monitoring for about 1/2 hour during which it never dropped below ~31mA. This a 2003 Ford E350. Looking around on some RV websites, it seems that this is well documented issue. The Starting Battery in this one was stone dead when I bought it. The previous owner was a widow, and she didn't have a clue. I have put a new battery in it, but don't want it to go dead sitting around.

12 v lead acid battery temp comp is between -30 to -33 mV / deg C during charging and about -12 mV/ deg C during float.

That is what I got off Dr Mark's PowerStream Web site.

 

[MikeMl]

Hi MikeMI,
.... No need for complex circuits to monitor a AC voltage only a charge controller for the PV panel is neede

Regards Bryan

Actually, there is a reason for a "smart charger". I have years of experience with this. The solar charge controller would need the same temperature compensation (complexity) as the AC powered one. If all you do is temperature-compensated float charging, that would keep the battery from self-discharging and prevent the parasitic load from slowly discharging it. However, that does nothing to solve the acid stratification problem.

This points to a two-state charging algorithm, likely incorporating a PIC.

 

[bryan1]

Hi Mike,
Siliconchip did a project fot a night light that used a pic for mttp charging and released the code too. Off memory it used a 16f88 chip and the code is in asm. It did also incoroprate temp variations for charging too Silicon Chip Online - A Solar-Powered Lighting System thats the link to the project but don't go buying the rest of it. If your interested I'll scan it and send it to you in pm.

It would be easy to modify the code to suit your application too..

Regards Bryan