Saturating a transistor using another transistor

Hi everyone.
I'm trying to make a driver circuit for an LED strip. The strip is rated for 12v and I've got a 12v switching power supply for it. What I'm trying to achieve with this driver is to minimize the loss across the switching component. Because as you know LEDs are very sensitive to voltage/current and a little voltage drop means a lot of brightness is lost.
What I'm using for the switching are 2SD882(NPN) transistors. The datasheet says it has a Vce(sat) of ~0.4v@1A which is pretty good for me. I'm not using a MOSFET because they're more expensive.
Problem is I'm using an ESP8266 to saturate the transistors and the pins can only output 12mA max. I need about 30mA base current to properly saturate the transistors and minimize Vce. So I thought to myself why not use a regular PNP transistor to provide the current to the base of the main NPN transistor and saturate it?
After googling I found out that there's this thing called a Darlington pair that essentially does just that.
But what I don't understand is why I have to connect the collectors/emitters together.
Internet says I should do this to form a darlington:

I think I should just do as follows. Because the only thing that Q1 needs to saturate is enough base current and voltage which Q2 provides (right?) :

I'm hating the Darlington setup because:
1- It drops more than 1 Volt.
2- I don't understand it.
3- I've tried my own setup(2nd pic) on a breadboard and it's working perfectly. Q1 is saturated and voltage drop is 0.25v.

Note: I also found out about another kind of darlington called Sziklai Darlington which is more similar to what I had in mind but even in this one the emitter of the PNP is connected to the collector of the NPN which I don't understand why.

My question is, is there anything wrong with my setup? And if there is why does it work?

My question is, is there anything wrong with my setup?

Yes, forget bipolar transistors, use logic level FETs.

1 Darlingtons have about 1.2-1.4v drop due to two diode drops. They are used for high gain but not when final power can’t afford the drop.

  1. You basically have a Sziklai pair which only has one diode drop at the second transistor... but the way you have it wired, your final drop is only at CE. No real issue with your wiring, saturating is easy.

If you need an explanation assuming about 0.3 Q2 voltage drop at CE, you have approximately 30mA at the base of Q1. With beta of about 50 you get 1.5A through Q1. I’ve left out a lot of detail but it’s a quick check. Your beta varies from device to device and affected by current and heat.

  1. MOSFET is what you should use for this. Much more efficient and you get better isolation between the two sources.

Thanks wolframore.

About using MOSFET, as I mentioned, the logic ones are very expensive where I live (they cost ~100x the price of a BJT). They're also hard to come by. And normal ones are going to need driving circuitry to fully open which makes them no less difficult to use than BJTS.

Where do you live?

wolframore:
Where do you live?

Iran.
Also I just checked and IRLZ44 is 30 times more expensive than 2SD882. So I might have exaggerated the price difference in my previous post a little bit :smiley:
But still, given I need three for RGB it's a considerable difference in price. Also the online store I get my stuff from doesn't have logic MOSFETs so I have to order them from another store which adds shipping costs.

I'm happy with the breadboard results with 2SD882. I'm gonna make a prototype and see if things keep working because my breadboards are crap and drop like 1V@500mA in a ~5cm trace. So results are always inaccurate on them.

Iran.

Ah!
Then I apologise for my unhelpful answer, I assumed FETs would be as easy to get as bi-polar transistors. You have had some pretty good answers and I have nothing helpful to add. Good luck with your project.

PerryBebbington:
Ah!
Then I apologise for my unhelpful answer, I assumed FETs would be as easy to get as bi-polar transistors. You have had some pretty good answers and I have nothing helpful to add. Good luck with your project.

Thanks.

Did you consider the cost of electricity lost for driving the base of a bipolar transistor? Maybe you'll find the MOSFET is chapter after all.

Diagram#2 is a good solution for the parts at hand, but with R2=100ohm you only have 20-25mA base current and probably not the <=0.4volt saturation you were hoping for.
Can reduce R2 to as low as 22ohm (100mA base current) if your 3.3volt supply can deliver that.

Another option is to use an NPN drive transistor with collector to 3.3volt* and emitter to R2.
Then you don’t need the 1k base resistor (direct connection to ESP pin).
There will be at least 1.4volt base drive drop that way, so calculate R2 for about 1.8volt.

*With this NPN drive configuration you can also connect the collector of the small transistor to raw (5volt).
Leo…

Most logic level MOSFETs (all those in TO-220 packages at least) are not easy to use with the ESP8266 - they need 4.5-5V on the gate to be fully open, at 3.3V they’re only partly open. That’s a serious issue. The only logic level ones that work well with 3.3V gate (or even lower) are in SOT-23 package (i.e. tiny SMD parts).

If you can get your hands on regular n-channel MOSFETs at a reasonable price, do consider using them instead. You can use a small NPN transistor to drive the gate, 12V opens them well. PWM won’t work too well, though.

Darlingtons are available in 3-pin packages, for precisely the reason that you often need more than a 10 or 20-fold gain in a switch driven from a logic circuit.

In a 3-pin package you have to put up with the voltage loss due to the shared collectors, but in the old days much of the circuitry controlled was 24V or more, and a lost volt is not a problem.

In 5V or 12V circuits that extra volt loss becomes important, darlingtons are not used anymore commercially as MOSFETs are far far better in all departments. Commercially there are literally 10000 or more MOSFET devices on the market, and only a 100 or less legacy darlingtons.

Your emitter follower circuit is fine BTW.

There are some more modern BJTs with really good saturation characteristics, the so-called superbeta devices, which will saturate fairly well with only 2% of collector current base-drive, such as the ZTX851 / ZTX951 family.
Often they will out-perform a darlington upto a couple of amps, although they are in a very small package so dissipation is limited.

A fact to deal with is that just a MOSFET requires the gate to be at a higher voltage than the voltage being switched for the MOSFET to be fully on (google "MOSFET boot strap"), otherwise it has resistance issues. One might google "LED driver" to see what is available and possible circuits.

zoomkat:
A fact to deal with is that just a MOSFET requires the gate to be at a higher voltage than the voltage being switched for the MOSFET to be fully on (google "MOSFET boot strap"), otherwise it has resistance issues. One might google "LED driver" to see what is available and possible circuits.

Huh?
Then how come I can switch 2-3A, 12V loads with 3.3V on the gate, and no resistance issues in the MOSFET? Those SOT23 parts can't dissipate much before overheating!

"Huh?"

What mosfet are you using? is it designed for "logic level" control?

Yes, logic level control.

3.3V to the gate from an ESP8266 to switch a 12V load. Standard low side switching circuit.

zoomkat:
A fact to deal with is that just a MOSFET requires the gate to be at a higher voltage than the voltage being switched for the MOSFET to be fully on (google "MOSFET boot strap"), otherwise it has resistance issues. One might google "LED driver" to see what is available and possible circuits.

Get your facts straight.

wvmarle:
Most logic level MOSFETs (all those in TO-220 packages at least) are not easy to use with the ESP8266 - they need 4.5-5V on the gate to be fully open, at 3.3V they're only partly open. That's a serious issue. The only logic level ones that work well with 3.3V gate (or even lower) are in SOT-23 package (i.e. tiny SMD parts).

If you can get your hands on regular n-channel MOSFETs at a reasonable price, do consider using them instead. You can use a small NPN transistor to drive the gate, 12V opens them well. PWM won't work too well, though.

IRLU120NPBF - 10A 100V,gate trigger voltage 2V 0.265Ohm at gate voltage 2-4V.
0.054 Ohm when driven by 10V
At massive £0.58 each, better parameters than proposed 2SD882 that cost £0.488
IRLU120 - 0.47V drop at 2A(0.1V drop when gate voltage 10V or higher)
2SD882(NPN) 0.7V drop at 2A

Look up AO3400A in sot23 package. Rds of 48 mOhms Max at 2.5v Vgs. So at 2 amps it drops 0.096V which dissipates less than 1/4 W... it could do that all day with decent pad design giving about 15-20C C while dissipating only 0.2 watts... all friggin day and night long. Typical Rds is about 24 mOhms but used worse case scenario. It could do double that easily... or more.

Sasquatchv:
Get your facts straight.
IRLU120NPBF - 10A 100V,gate trigger voltage 2V 0.265Ohm at gate voltage 2-4V.

Well, I looked up the data sheet for that MOSFET to get some facts straight.

Also the data sheet gives that on resistance you mention for 4V on the gate, not less than that. At 3V it plateaus at 2A maximum current, that's obviously not completely on, making it quite useless when using with a 3.3V part.

And indeed that's a pretty high on resistance compared to to ~20 mOhm that I'm used to - at 6A with just 3V at the gate, and a much lower price. That's a pretty high price for a MOSFET, especially for one with properties as poor as this.

So as yet I still haven't seen a TO220 that's fully open with 3V at the gate, which would make it usable with a 3.3V signal.

The typical graphs for the IRLU120N clearly show the plateau at about 3.5V, so its definitely not suitable for 3.3V operation. I'm not sure why they bother adding an Rds(on) rating for Vgs=4V, there is no common use of 4V supply in circuitry.