Improve this solar controller?

Also, I did check the 20w panel, I knew it had a diode, but it turns out to be a bypass diode, not a night drain prevention diode! So I removed it as it is only needed with multiple panels and fitted a stocky diode right on the panel.

Steve

The TL431 shouldn't even get warm to the touch. I suspect you have something miswired

Check both the TL431 pin out, and the PNP

MikeMl said:

The TL431 shouldn't even get warm to the touch. I suspect you have something miswired

Click to expand...

Assuming it's the circuit from post #15, then the TL431 is directly across the supply with just a Vbe drop of the transistor in series with it.

Could probably do with a resistor from the TL to the base?.

It definitely isn't wired wrong.
It gets hot when the tip30 and the 47r 5w resistor is drawing half an amp.

Steve

Nigel Goodwin said:

Assuming it's the circuit from post #15, then the TL431 is directly across the supply with just a Vbe drop of the transistor in series with it.

Could probably do with a resistor from the TL to the base?.

Click to expand...

EccentricDyslexic said:

It definitely isn't wired wrong.
It gets hot when the tip30 and the 47r 5w resistor is drawing half an amp.

Steve

Click to expand...

Yes, at a panel current of 0.5A, the TL431 will get quite hot to the touch. I considered its power dissipation with the orginal panel current of 300mA when I ran the simulation. At higher panel currents, the problem is that the Hfe of the TIP30 at Ic=500mA is only 30. Making Q1 a Darlington, or using a PNP with a higher Hfe would reduce the power dissipation in the TL431 at higher currents.

If the Emitter current of Q1 is 500mA (the entire panel current because the battery is fully-charged), the current out of the base of Q1 is Ic/Hfe or about 500/30 = 17mA. At the same time, the current through R3 is Vbe/R3 = 0.7/680 = 1mA, so the total current into the cathode of the TL431 is ~18mA.

The voltage from the TL431 cathode to the TL431 anode is the battery voltage - Vbe = (14.2-0.7) , so the power dissipation in the TL431 is I*E = 18mA*(14.2-0.7) = 243mW. The thermal resistance of the TO-92 package from junction to ambient is 140degC/W, so the temperature rise of the TO-92 package would be 140*0.243 = 34degC, which will not burn it up. It is still well within its maximum operating temperature. The circuit as-is is fine up to the original 300mA Panel current, begins getting a bit marginal at 1/2A.

I am not in love with Nigel's suggestion, because putting a resistor between Q1 base and TL431 cathode will create a voltage divider with R3, thereby reducing the shunt regulator's loop-gain by the divider ratio. This would flatten out the nice sharp knee that this circuit has.

To reduce the power dissipation in the TL431 at shunt currents of >0.3A, I would do one of the following:

1. Replace the TIP30/32 with a TIP142 PNP Darlington. It has a Hfe of 1000. That reduces the TL431 cathode current to 2.5mA, which reduces the dissipation to (14.2-1.4)*2.5m = 32mW, which makes the temperature rise 140*o.032 = 4.8degC.

2. Make a homebrew Darlington by adding small PNP ahead of the TIP30/32 base. Results similar to above.

3. Use a PFET in place of the TIP30/32. Now the TL431 cathode current is determined only by the threshold voltage of the PFET divided by R3, so about 3/680 = 4.4mA, so the dissipation is (14.2-3)*4.4m = 50mW, so the temp rise is 140*0.05 = 7degC.

The tip142 seems to be a npn tip 147 you mean? Seems cheap so will buy a couple, for the 20w panels I have. Will stick with the tip30 for the 5 watt panels.

Thanks Mike,

Steve

EccentricDyslexic said:

The tip142 seems to be a npn tip 147 you mean? Seems cheap so will buy a couple, for the 20w panels I have. Will stick with the tip30 for the 5 watt panels.

Thanks Mike,

Steve

Click to expand...

Yep, here is the data sheet. TIP145,146,147 are the PNPs. For the 20W panels, you might have to diddle the power resistor you added from the collector to ground. Goal is to have about 80% of the battery voltage (14.2V?) across the resistor and about 20% across the transistor at the max panel current being shunted.

I went with a 20w 10 ohm for the 20w panels, sound ok?

steve

I = P/E = 20W /14.2V = 1.41A

E=IR = 1.41*10 = 14.1V, so with a 10Ω resistor there is no margin for the PNP to regulate... It looks to me you need a 8Ω resistor. See, this is why I think you are better off with no resistor, and just put a bigger heatsink on the PNP , or use the 2A car taillight bulb.

Hi Mike, finally got round to using this with a TIP147. For the 5w panel it works fine but the TL431 still gets hot, but it hasnt failed yet. I have built another for my 20w panel(with 8ohm 20w wire wound resister), now that gets the TL431 very hot. How can i make this a reliable circuit and take the pressure off the TL431?

Cheers

Steve

EccentricDyslexic said:

Hi Mike, finally got round to using this with a TIP147. For the 5w panel it works fine but the TL431 still gets hot, but it hasnt failed yet. I have built another for my 20w panel(with 8ohm 20w wire wound resister), now that gets the TL431 very hot. How can i make this a reliable circuit and take the pressure off the TL431?

Click to expand...

Steve,

I have lost track of what you are doing. Please post a schematic of your current hook up.
Show the added diodes and load resistors.

This is it, from post 42. I have done away with the large resister now as you suggested, and am wanting to stop the voltage going higher than 13.5v, so the battery (an 8ah SLA) doest over charge.

Cheers

Steve

EccentricDyslexic said:

Hi Mike, finally got round to using this with a TIP147. For the 5w panel it works fine but the TL431 still gets hot, but it hasnt failed yet. I have built another for my 20w panel(with 8ohm 20w wire wound resister), now that gets the TL431 very hot. How can i make this a reliable circuit and take the pressure off the TL431?

Click to expand...

As I said way back in post #63, add a resistor between the base of Q1 and the TL341 - it's getting hot because you're running LOT'S of current through it.

Mike said he didn't like that idea, well I don't like NOT doing it

I don't see as his loop gain fears have any merit (it's as much a switch as a linear system) - stick a suitable resistor in and see what happens.

EccentricDyslexic said:

This is it, from post 42. I have done away with the large resister now as you suggested, and am wanting to stop the voltage going higher than 13.5v, so the battery (an 8ah SLA) doest over charge...

Click to expand...

Is the panel 5W or 20W?

The panel has an internal isolation diode?

R4 is deleted?

Nothing at the big X?

Q1 is a TIP32?

What did you do to adjust the voltage to 13.6V?

This time it is a 20w panel. Q1 is a TIP 147, R4 deleted. Yes panel has an isolation diode. Nothing at big X. I have a 15k resister inline with a 10k variable resister for R2 this is where I adjust the trigger voltage.

Cheers

Steve

EccentricDyslexic said:

I built this charge controller,

**broken link removed**

the problem is the leds light drawing 30ma all of the time, which makes the circuit useless. Can it be improved with a better comparator?

Cheers

Steve

Click to expand...

What do you want the LEDs to do? Turn on when dark or turn on with full charge Voltage or turn on when battery fully charged with no current to battery? Or turn on with a switch ? Or ?

Instead of a shunt charger, you can turn the TIP into a series regulator and current sense with a small R and indicate when Vbat is above 12.5 and near zero charge current or even flash when done.

Ok, one more slight change to reduce the heating in the TL431. Make R3 1.5K. With the high gain in the TIP147, by raising the value of R3, the peak current that flows through the TL431 is reduced, making for less heating.

I modified the sim to show a gradual turn on of the panel (sunrise), a steady output at 1.46A (20W), and then a gradual turn off (sunset). I also changed R2 to limit the battery voltage at ~13.7V. Remember that the time scale is arbitrary.

I plot the battery voltage V(plus) green trace. It levels out at 13.75V at midday (100s). Nigel, notice how much the battery voltage drops after the sun goes down ;

V(b) dk. blue trace shows the voltage at the base of the PNP Darlington. Note that [V(b)-V(plus)] is the Vbe of U2.

The red trace shows the power dissipation in the TL431. 19 mW will not even make it feel warm!

I(c1) lt. blue trace shows that the entire panel output current goes into the simulated battery (as long as V(plus) is < 13.75V), and then stops going into the battery even though the sun is still shinning.

The purple trace is the emitter current of the shunt Darlington transistor. U2. A bit after 100s, the full output (1.46A) of the panel is being shunted to prevent the battery voltage going any higher.

The gold trace shows the power dissipated by U2. Note that the full panel output of 20W is dumped into U2, and it has to be adequately heat sinked...

I'm not sure if this has been covered in this thread, but it's good to understand the way current and voltage is usually specified for many solar panels.

Output voltage is often spec'd as open circuit voltage, meaning when the load current is zero.

Output current is often spec'd as short circuit current, meaning when the load voltage is zero. (Due to the short on the output)

The point is, that you won't get max voltage and current at the same time. So if you are simulating the behavior of your charge controller with a bench supply, you may be working it harder than what it might see in real world conditions. Of course, beefing up the design to meet higher than normal conditions is never a bad thing.

1.5K works a treat! Thanks Mike!

Steve

Having to rebuild this circuit as the U2 had blown a hole in itself. This time going to fit a larger heat sink as I suspect summer days shouting the whole panel output killed it, I won’t have been excessive current as it’s protected by a 2amp fuse.

Steve