Improve this solar controller?

[MikeMl]

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

... Note that the full panel output of 20W is dumped into U2, and it has to be adequately heat sinked...

2A*13.6V = 27.2W requires a good size heat sink.

First, here is the Thermal Resistance for a TIP147:



Here is an online Temperature Rise Calculator:



To limit the junction temp to ~100deg C, you require a heatsink with less than 2degC rise per Watt. Just for an example, here is one that has 1.8deg/W with natural (convective) air flow if mounted vertically. How big was your heatsink?

 

[EccentricDyslexic]

Thanks mike, i fundamentaly misunderstood... mine was a quarter of that size:-(

I do have an spare intel cpu heatsink....

Thanks for the help really appreciate it!

Steve

 

[EccentricDyslexic]

Or...I can still wire in a Head light bulb can’t i? The price of these decent size heat sinks with good temp rise values!

Steve

 

[MikeMl]

Or...I can still wire in a Head light bulb can’t i? The price of these decent size heat sinks with good temp rise values!

You can add a lamp as shown by RL. However, you want one with a hot resistance of less than ~7Ω if the panel current is 1.5A @ 13.7V. If RL>8Ω, the Darlington saturates, and the regulator stops regulating. If Rl is much less than ~6Ω, then the transistor still does most of the dissipation. There has to be a minimum voltage across the Darlington of a couple of volts.



I simulate the panel as a 1.5A current source and vary RL from 0 to 9Ω (x-axis). V(bat) shows constant voltage of 13.75V while RL <8.2Ω. The light blue trace V(C)*I(RL) is the power dissipated in the lamp in Watts. The violet trace is the power dissipated in the Darlington.
Ideally, about 75% of the power would be dissipated in the lamp, and about 25% in the Darlington (it still needs some heatsinking).

This indicates that you should aim to get a 12V lamp rated at 12 to 15W, like this one, or this other one.
An automotive headlamp is too low a resistance, a parking light is too high a resistance.... Also look at an automotive #93 lamp.

 

[EccentricDyslexic]

Hi, i'm back...again! Still trying to get this thing reliable.
It failed this time due to an over heating tl431.
The circuit is as the last posted image immediately above.
RL is a 20w8ohm resistor. The tip147 doesnt get hot, but the resistor does naturally.
I've found a new tl431 and stuck it onto a heat sink. even so, it still gets very hot and smokes at 2amp/15v through the entire circuit.
Any help to take some heat off the tl431 would be greatly appreciated.
Cheers
Steve

 

[Nigel Goodwin]

You need a resistor in series with the base of the TIP, as it stands the TL431 is drawing current right across the supply, via the BE junction of the darlington.

 

[EccentricDyslexic]

1k?

 

[Nigel Goodwin]

Depends on how much current you want through the TL341, the voltage on the TL341, and the supply voltage.

 

[EccentricDyslexic]

Just enough to stop it getting hot. The circuit kick in at 13.7v and is sinking 2amps by about 15v

 

[Tony Stewart]

Thanks! I notice you are using a TIP32c not a TIP30c, are they interchangeable? i only have tip30c handy.

Cheers!

Steve

I believe TIP30 = PNP lower gain 1A than the TIP32C but but both will saturate about same

the design shunts the excess current to LEDs and seems to overheat the Open collector NPN which cannot saturate the base but could saturate if the load R as indicated in comment 2 is 10k but then should never work as a shunt Load with 10k series NPN-collector of Tl431 to PNP base

I may not have the datasheet for your PNP parts but they do not appear to be a Darlington transistor.

Normally a series cutoff PFET is used or series PNP rather than this poor design.

It would much better to have your power consumption and temp rise specs chosen carefully. Or a CMOS LDO

 

[EccentricDyslexic]

I dont use the tip30 nor 32, the circuit evolved from the original and the current one I’m using is in the last pic posted in post 84.

Cheers
Steve

 

[Tony Stewart]

You built a charge controller in 2015, and the reason this thread is still unfinished is because you failed to define what the circuit should do with values, tolerances or a link to compare. Please Start with specs next time, not a long meandering story with no measurable goals for all variables; in/function/out