More voltage shunt reference.

You're shunting all the cut-in current around the base, no gain to speak of since voltage-to-current of the diodes is logarithmic. This is NOT the way to go. At full charge your transistor dissipates all the power available from the solar panel. Use a 10W panel and your transistor will dissipate 10W, it's that simple.

Whereas for this method:

George, this is your business, but I am curious as to why you want to shunt regulate rather than series regulate. The next question is why linear regulate rather than bang-bang regulate for either series or shunt?

A TL431 would make things much easier ...

I added a few series diodes labelled TC to add some negative temperature compensation, I drew two but I haven't stopped to think how many you'd actually need.

piglet

A TL431 would make things much easier ...

I added a few series diodes labelled TC to add some negative temperature compensation, I drew two but I haven't stopped to think how many you'd actually need.

piglet

Awesome, thanks piglet. Dang I forgot to order some TL431's. I should understand that IC more... kinda like the shunt version of the lm317.. (well less power)

George H.

Hmm well I guess I started with the shunt idea... 'cause it's got no voltage drop unless it's on. I also like that I can add an LED indicator to it. I guess I haven't thought much about the bang-bang approach, 'cause then it'd have to be more complicated with an opamp or comparator... I know not a big deal.

George H.

Thanks for your explanation.

Many wall-wart switchers use a TL431 to drive the optocoupler, so you may find one in your junk box. extracting and testing it will probably cost more than buying new, but may be faster.

The TL431 already contains an opamp (with one input tied to a bandgap ref). You can easily make it bang-bang with a little positive feedback. Will try to sketch out a way to do that later this weekend.

piglet

Hi George, Here is an example bang-bang charger. Voltage drop across a schottky diode is lower than a series pass transistor so I opted for a very simple shunt version. The resistor values shown should give approximate charge cutoff at 14V and resume at 13V.

Adding temperature compensation is left as an exercise for the reader.

piglet

Hmm OK feedback into the voltage divider.... some tap that changes between on and off.... can I ask one state to turn on (or off) an led too? (LED in line that feeds pass element... maybe not a darlington, but just a single npn. ) With hysterics on needs no temp co. George H.

Thanks piglet. I didn't look. I need to try and scribble my own thing first. I want an led too. :^) When it's on (tl431) the cathode is only a V_ce drop above the anode? That's a lot of voltage head room... Hey thanks again!

George H.

Not quite that low, the internals still needs biasing so the cathode never drops anywhere near as low as 2-3 vbe drops. You would need to study datasheets or take measurements to find out.

Another handy trick to introduce feedback is to raise the anode slightly above ground with a low value resistor and inject feedback there.

The LM4040 is also good to have in your repertoire, it is to tl431 what pnp is to npn. i.e. TL431 has reference to anode, LM4040 has reference to cathode.

piglet

Oh 2.5 V, @1mA... I could spice it... Years ago I used the lm4040 to show when a pass transistor was saturating. I've mostly forgotten the circuit. :^) And none in parts box. I'll add some to order.

GH

Given how Wikipedia describes the TL431 as "the de facto industry standard error amplifier, it seems like an excellent component to "add to my repertoire," as Eric says. Today my scavenger's heart experienced great joy after a TL431 was discovered on a Cisco switch power supply pulled from my bone pile. :)

Danke,