About darkness activated circuits...

[Illum]

I'm planning to build a fixed astronomy light for someone who doesn't like to toggle things on and off manually when he doesn't have to, The red LEDs are to come on when its dark outside and have it turn itself off to conserve energy when the day breaks.

I came up with this simple elementary circuit that most ppl I'm sure are familiar with



1 general purpose NPN signal transistor (I used an 2N4401 because thats what I happened to have around, I'm pretty sure 2N3904 will work as well) and a voltage divider with the conventional R2 being a varister (CdS photocell)

The operation principle is simple, when there's light out the photocell increases resistance, which cuts off the power biased to the base, which interrupts the flow between collector and emitter. As light dims, collective resistance decreases, base is biased, and negative contact is rebuilt.

Unfortunately this is where I start encountering problems. The LEDs are red luxeons and will encompass a perimeter of 10ft by 10ft worth of workspace. Its power supply is a boostpuck that will allow me to run it on 12VDC with dimmable function. 2N4401 is in a TO-92 pkg, which the maximum allowable current may not be able to exceed 100ma. the Boostpuck is well capable of drawing up to an amp to power 8-9 LEDs.

I plan to wire the assembly as follows:


The Boostpuck will draw nearly an amp if dimmer set to max, given the surplus of 2N4401s on hand, I was wondering if this could work [paralleling transistors all on a single voltage divider, up to 9 in parallel yielding a 9x100 = 900ma allowance in current flow]


I know there are power transistors that can do this more efficiently, but trying to heatsink them might be a bigger challenge than I intend the project to be.

Any thoughts?

 

[Benson]

If you're concerned about heatsinking a power transistor, your plan for dissipating the same heat from a pile of signal transistors is just having more of them, and... plenty of airflow?? I dunno, seems easier to ensure decent heatsinking from one big one with a nice tab on the top.

Anyway, you might be better off setting them up as a Darlington pair (more like Darlington dozen?) -- I started to run some numbers for this as an optimization, but since HFE is about 200, you're looking at 0.5mA IB, so 5mA total with your parallel arrangement. But you're running (12V-0.6V)/100kOhm = 0.1mA (even if the photocell was open-circuit), so you're off a factor of 50 to start with -- your design won't work as it stands. That or I grabbed the wrong datasheet, am making an elementary mistake, or whatever...

Not sure what turn-on resistance you're aiming for, but taking your voltage divider and working from the assumption IB=0, that'd be 5k, so I'll work from that (obviously recalculate if your photocell is different...). Also, don't take any of my results as gospel -- there's lots of quick-and-dirty math with unspecified assumptions and rounding to 1 figure, just to get the idea across.

So, for your current design:
Given 5mA base current (for your current design) and 0.6V/5k=0.1mA, you'll need 11.4V/5.1mA=2.2k, but then it would start to turn on at only 100 ohms. Note that the resistance ratio here is about 50 -- even if you tune it to a more suitable range, it'll be very slow switching on and off.

OTOH, if you add one to drive the rest in a Darlington arrangement:
Now the 10 are pulling 5mA from the input transistor, which would be operating with HFE about 80, so only drawing IB=0.06mA at full turn-on. Now 100k provides about 110uA, of which 60 goes through the emitter and the remaining 50 goes down the photocell, so it turns fully on at 0.6V/50uA=12k, and starts turning on at 0.6V/110uA=5k. The ratio is reduced from 50 to 2.4, at the cost of one additional transistor.

 

[Light Sabre]

With such a slow changing input (darkness/photocell) and that the exact turn on point of each transistor being slightly different that you would get a cascade failure. One transistor would turn on first and try to handle the whole load and go . Then it would happen to the next one, and the next one and so on. I would say stay out of the linear region of a transistor and go digital with an IC with a schmidt trigger input would be better. It would me more stable and more power efficient. I'm sure that someone else could come up with something more inline with what you're thinking and work just fine. Maybe have multiple transistor/photocell circuits that run a few LED's each not exceeding the 100ma/transistor. I use the ~75% rule, so limit it to no more than 80ma/transistor.

 

[Illum]

If you're concerned about heatsinking a power transistor, your plan for dissipating the same heat from a pile of signal transistors is just having more of them, and... plenty of airflow?? I dunno, seems easier to ensure decent heatsinking from one big one with a nice tab on the top.

Anyway, you might be better off setting them up as a Darlington pair (more like Darlington dozen?) -- I started to run some numbers for this as an optimization, but since HFE is about 200, you're looking at 0.5mA IB, so 5mA total with your parallel arrangement. But you're running (12V-0.6V)/100kOhm = 0.1mA (even if the photocell was open-circuit), so you're off a factor of 50 to start with -- your design won't work as it stands. That or I grabbed the wrong datasheet, am making an elementary mistake, or whatever...

I'm not predominantly concerned with heatsinking at this time, but it will become something I will need to review when it does get built.

I'm somewhat familiar with this darlington pair, I have seen it used in other places, but I havent a clue how to calculate the resulting amplification or how to go about determining how many transistors is necessary to allow 1A pass through. IIRC the first may be a low power type, but the second will need to be high power as the maximum collector current for the pair is the same as the collector current for TR2. Is it safe to assume I can use an 2N4401 as the first and find something good for an amp like 2N4898 and go from there? Granted 1A might be pushing it a bit but the resulting current gain should compensate for it right?

as an example, say the first transistor happened to be a BC182 [Ic(max)=100ma, Hfe(min)=100, TO-92] and I couple it with something like a TIP29A [Ic(max)=1A, Hfe(min)=40, TO-220], the resultant gain would be 100x40 = 4000 yes? how do i convert that to current?
:thanks: for the reminder Light Sabre, that 75% rule is valuable, I learned of this working with resistors, but as it appears I failed to translate that over

How can I arrange the darlington pair to help with my other project?
http://www.candlepowerforums.com/vb/showthread.php?p=3258076

 

[Dave_H]

I'd skip the multiple 2N4401's and save them for some other project...
The Darlington disadvantage (as noted on the Wiki page) is higher
collector-emitter voltage drop. At 1A you'd be wasting another 0.5W
to ~0.75W. TIP121 might work OK though but its guaranteed
minimum current gain is 1000.

Again, no circuit at my fingerprints but a simple power FET circuit
using IRFZ30 or similar at 1A would need no heatsink. Trick would
be to get it biased correctly, with some hysteresis to avoid getting stuck
in the linear range, which could heat up the FET and possibly damage
it.

If I had a bit more time I'd jump right in to test out something, but
consider this a nudge in the right direction.

Dave

 

[Illum]

turns out he didn't need it, the friend just bought a case of solar yard lights and modded the lamps to red

 

[Dave_H]

I've been using solar garden lights as indoor portable night-lights
for several years. The most practical I find are ones with easy
access to the battery (without having to take it apart each time)
and with ON/OFF switch (not that one couldn't be added). This
way you can stick in a charged battery, without having to worry
about the light running down prematurely. In fact you can stick
in non-rechargeable cells, just don't try to charge them during
the day. I've been using up discarded alkaline cells typically
showing 1.0-1.3v this way.

Dave

 

[Illum]

Solar yard lights are certainly one way of doing things...I prefer to do the soldering myself [even if it meant lifting the schematic directly from the solar light] for one thing it looks more original and for another I can build it to fit a certain dimension such as an enclosure, and I won't end up with a bunch of extra trash

Too bad not many schematics are posted for these things...

 

[jason 77]

Too bad not many schematics are posted for these things...

I don't know, I googled dark sensing circuit and got the following links

http://www.technologystudent.com/elec1/lgtdrk1.htm

http://www.evilmadscientist.com/article.php/nightlight

http://www.educypedia.be/electronics/circuitssensorslight.htm

The last one has a bunch of different types of circuits... or am I missing something?

 

[Illum]

There are plenty of dark activated circuits flowing about on the internet, what I was referring to is the exact circuit in the solar garden lights that includes everything from the charging component to the switching:thumbsup:

that last link proved interesting though, theres a heck of alot of info within those pages, I've bookmarked it awhile ago but I have yet to finish reading it:duh2: