PWM Circuit for AGM Battery Charger 2

[Jacques2014]

Hi, For different reasons, I am designing a custom battery charger for an 12Vdc 100amps AGM battery. I use a PWM control circuit, but the driving Mosfet tend to heat a lot. The actual circuit simply switch the input DC source (solar panels) using a power mosfet directly in the battery trough diodes. I tough that adding a coil between the mosfet output and the battery may help to lower the heat dissipated but the mosfet. I need help figuring out how to compute the proper inductance value in relation with my switch frequency. So any help will be welcome. Maybe I am wrong, and the coil will not help at all. Your input will be welcome. Input DC voltage vary from 12 to 22Vdc approximately while the battery will peak around 15Vdc, I try to keep current constant using the PWM. The circuit is working, but heat a lot. Peak current may raise to 10amps when running in bulk charge.

Bye
Jacques

 

[ScottyUK]

Bigger mosfets have lower on-state resistance. Big low-voltage mosfets are fairly cheap and they parallel easily, so you can reduce the on-state losses quite easily. The downside is that the demands upon the driver increase as the gate capacitance increase with the larger or greater number of devices. Can you post a circuit diagram of what you already have?

 

[Jacques2014]

I tough I did post an attachment, I did attached a JPG of the driver circuit section with my original post. Maybe it did not work, I will repost it tomorrow, I am no longer at my office computer. I remember seeing a message indicating that the JPG was uploaded correctly, I do not know were it is now...

Bye
Jacques

 

[OperaHouse]

The problem is likely the gate driver. The FET should be driven hard >10V If a simple PWM controller, use a low frequency 100-500hz. Never allow short pulses which only cause heating. Run between 5-95%, then jump to all on or off. I have built several of these using a UNO for my solar power. I run a manual power point 17V rather than track as it only changes with temp and it is only used in the summer. At low frequencies a power transformer works reasonably well as a buck inductor. Very little to be gained with a 12V panel
array. If you have more than one panel put them in series.

 

[Jacques2014]

Here the circuit, I hope the attachment will work this time.

Bye
Jacques

 

[Jacques2014]

The File again, this time I think it will work, sorry for the previous mistake.

Bye
Jacques

 

[hacksaw]

Just curious as to the manufacturers recommended charging rate for your battery.

Also a bit puzzled by your current sense resistors of 3 X 0.033 Ohm in parallel, but a listed voltage range of 0 - 3.3 volts. Is this your current design drawing?

 

[OperaHouse]

That 100K gate resistor guarantees a sloooooooooooooow turn off since you have to discharge all the gate capacitance through it. That is the source of your heating. If you were driving that with a standard UNO 490hz PWM the maximum resistance would be 1000 ohms and you would still need to limit short pulses in software. What is the rest of your circuit.

 

[DanEE]

And equally important for AGM batteries is to build into the charging circuit temperature compensation (preferably monitoring actual battery temperatures) to ensure the battery float voltage stays on the recommended curve.

While batteries seem like simple devices, the chemistry of AGMs especially considering the gas recombination process is quite complex and demands the resultant float voltage to be correct for the battery temperature.

If resultant float is off in one direction, positive electrode (plays a more significant role in the gas recombination than the negative and, there is a heating effect and delta voltage effect on terminal voltage from the gas recombination itself). If off the curve, positive electrode corrosion tends to occur; if off in the other direction, sulfation build up is likely.. This aspect was always question number 1 when approaching the battery manufacturer for a warranty replacement.

All the battery manufacturer's in my experience specify float versus temp for their AGM stationary batteries (measured on the positive terminal) to 2 significant digits and of course all modern commercial microprocessor controlled chargers have the selectable or customizable profiles in their design to match the battery manufacturers requirement.

One last reason for this with AGMS is the possibility of battery thermal runaway if the batteries are in an uncontrolled temperature environment where battery temps can get up to 95 deg F or beyond. AGMS as they get hotter start drawing more charging current for a given charger voltage and without remote battery temperature sensing by the charger, a thermal run away can happen..

I've seen AGM thermal runaway happen 3 times at commercial customer sites, in one case destroying about $25,000 dollars of commercial batteries where the customer refused to replace their aging non compensated chargers with newer technology. We ended up placing temperature indicating stickers

http://www.wahltempplates.com

on the battery systems on all our accounts and cautioned the customers about this, in that if the system were damaged, it would be at their expense.

 

[OperaHouse]

Too late now but I would never get an AGM battery. Unless one has an over abundance of panels, most solar systems seldom get beyond bulk charge. I hate batteries. Actual use turns out to be about 8 cents per KWH, about the same as grid power.

 

[IRstuff]

Why is MOSFET heating a surprise? You're dumping 15A across a maximum of 7V across the MOSFET, which works out to 105W. A DC-DC converter would seem to make more sense to me.

TTFN
faq731-376

Need help writing a question or understanding a reply? forum1529

Of course I can. I can do anything. I can do absolutely anything. I'm an expert!

 

[OperaHouse]

Did you grab the wrong datasheet? 74A PFET in PWM mode. Though with 3400pF gate capacitance and 100K it would be kinda analoggy. And on a good day you could blow the gate out with 23V. Take that gate drive resistor and move it down to form a voltage divider.

 

[Jacques2014]

Hi, a few information's in regard to everyone post:

The recommended bulk charge rate is around 10% of the battery capacity of 100amps so around 10amps. I design the circuit to handle 15 amps, but I limit the current to not more than 10amp via the PWM.

The 0-3.3V in relation with the sense resistor come from an ADM4073T.

I did switch R8 from 100K to 1K but forget to correct the schematics.

My PWM frequency is 1Khz.

I do not use UNO, the PWM is drive by an Atmel 89C5130 MCU.

I did implement a temperature sensor to control AGM charge.

Battery temperature will stay low, they are surrounded by water not going over 10c.

Correct me if I am wrong, but Q1 (IRF4905) has a 20mohm Rds, so at 15amp, it should drop around 0.3watt, not 105. It's not a linear regulator that in fact have to drop 7 volts.

The parts that heat a lot are the sense resistors, Q1 & D20 that are on the same heatsink. I can easily raise the resistors power, but it's a bit more complicated to dissipate the heat produce by Q1 and D20.

Q1 should only produce 0.3Watts
D20 have been replace by a MBR2515L with 450mv drop voltage so it should dissipate around 7 watts at max power.

I wish I can find a diode with a lower drop voltage for D20, this should reduce the heat.

Sorry if the schematic was not up to date with all the modifications I did during the test period, and I did not post the full schematics, the MCU is not relevant to the heat problem except for the PWM frequency.

I will correct the software to remove short pulse as suggested.

Bye
Jacques

 

[IRstuff]

" It's not a linear regulator that in fact have to drop 7 volts."

If the input is 22V and the output is 15V, then yes, you're dropping 7V. If your calculation is correct, then there should be no reason for the transistor to get hot. Since it is getting hot, then the calculation may not what you think it is.

TTFN
faq731-376

Need help writing a question or understanding a reply? forum1529

Of course I can. I can do anything. I can do absolutely anything. I'm an expert!

 

[OperaHouse]

Use a FET backwards instead if a diode. Frankly I don't use diodes till I parallel up high voltage strings. Still have to worry about excessive gate voltage. If gate resistor 1K, why are you getting hot. Everybody runs at about 100hz.

IR, solar panels are current sources. In simple PWM they just short them to the battery

 

[DHambley]

You asked about how to pick an inductor but I don't see any inductor in that schematic. Do you have a certain ripple current in mind? If not, I'd suggest 30% peak-to-peak of the DC to start with.

Darrell Hambley P.E.
SENTEK Engineering, LLC

 

[LionelHutz]

I would have expected a circuit more like a buck regulator. You could likely use a N-channel FET with an IR driver IC.

But, if you expect 15V output with 12-22V input then maybe take a look at buck-boost type circuits.

 

[IRstuff]

" solar panels are current sources. In simple PWM they just short them to the battery"

OK, but, nevertheless, they are not instantaneously compliant; there's always a response time that's dependent on series resistance and junction capacitance, particularly on a large panel. For however many nanoseconds or microseconds it takes to actually discharge the junction capacitance, the MOSFET is dumping a truckload of power.

TTFN
faq731-376

Need help writing a question or understanding a reply? forum1529

Of course I can. I can do anything. I can do absolutely anything. I'm an expert!

 

[PeterKsiazek]

Typically in this application you would actually have a DC/DC converter which extracts the power from the solar panels and charges the batteries. The DC/DC would provide MPPT (maximum power point tracking) to maximize the power extraction from the solar panels and also perform SAFE charging of the battery. Using a pass transistor working with some sort of hysteretic, on-off, control is a bad a idea for many reasons:

1. High peak currents (EMI issues, increased losses)
2. High ripple currents in the battery (increased losses, heating, etc)
3. Lack of safe charging and control
4. Cannot charge the battery when your input voltage is low

A simple buck converter will solve 1-3 but to deal with 4 you will need a more complicated topology. Further information about power levels, isolation requirements, etc is needed as well. To do this right is not a simple task.

 

[OperaHouse]

I must have missed that day in class. My camp is all solar, lights entertainment, refrigeration and hot water. Some old UPS from the town recycling and some hot melt glue......I'm good to go. Everything runs on a $10 micro. No power is every wasted it always has a place to go. Most RE people waste more power than the average housewife in Sheboygan.