Design help - Reference voltage with increased current

[jfigge]

Hi guys, I'm trying to build a circuit using a 12v supply and pull off a 5v rail with 200mA.  The 5v circuit is fixed so doesn't need constant current or anything too complex, but trying to understand how not to burn up the BJT?  Below is what I have and it works, but it doesn't take too long before the 2n22222 gets too hot to handle.  I have a TIP102 (A darlington in a To-220 Package) that can handle the heat, but I'm sure there's a better design to avoid the inefficiencies?

Thanks for any help,
Jason


12v------------------------------(c)-BJT-(e)--------------
           |                      |               (b)         |            |
           |                      |                 |          |            |
           5v-ref-----(+)OpAmp---R1--+          R2       LOAD (5v @ 200mA)
                          (-)     |                            |            |
                           |   Gnd                           |            |
                           |                                    |            |
                            ---------------------------------------
                       

My best guess at what is going on is that there's both 200ma and the excess supply voltage (~7v?) across the BJT and that's over 1W?

[Conrad Hoffman]

You can't beat ohm's law in a linear circuit. If the load is known you can sometimes reduce the power in the BJT or FET by putting a resistor in parallel with it, but the load always has to be present and stay within a certain range. As you say, if you can reduce the input voltage a bit, that will help.

[chuckb]

You have a series pass regulator now. For this application it will have to dissipate 1.4 watts in the BJT. An LM7805 will do all of this in a single package. But you still need to dissipate the 1.4 watts. A TO-220 package with increase temperaue 67 deg C for every watt it has to dissipate. So in this application the BJT or LM7805 will operate about 100 deg C above ambient temperature.

Another method is a non-isolated switching regulator. They are more complex and usually require small input and output filter capacitors, but they are available in a prepackaged unit.
Something like the CUI P7805-500 would work for you and only dissipate 0.1 watt. No more burned fingers!

The switching regulator will generate more electrical noise in your circuit. If this is an audio amplifier you may hear the switching noise. If it's digital logic the noise may not be noticed.

[jfigge]

Hm, thank you for the quick responses.  First impression is that at least I have the circuit somewhat right!  This is for an e-load project and I 'think' i'll need ~10+v for the next opAmp that'll drive a power mosfet down the road.  I'll give the switching idea a go, too.  Thanks for the suggestion

[magic]

Why would a MOSFET-based electronic load consume 200mA from a 5V supply rail? I wonder if you may be considering some other inefficient solutions in that circuit

Anyway, yes, a buck regulator is the efficient solution, and for a hot and inefficient solution, an integrated voltage regulator like 7805 is a no-brainer unless you have exotic requirements it cannot meet.

[jfigge]

The 5v rail is just to drive the sensors and processor to control the circuit.  The actual load will be put through power mosfets that are not logic level- at least that's the goal

[magic]

I don't know how large the project will be, maybe it needs to have 100 regulated channels managed by Raspberry Pi.

But a simple MCU, small LCD display and a few opamps will consume closer to 20mA than 200mA. In such case, old 7805 may be good enough.

[jfigge]

Hm.  Still learning so forgive me if it's not making sense   The project isn't that big, but is meant to be a learning experience as I work though all of the different pieces and parts I might need.  The biggest draw comes from the relay.  I see other e-loads rely on the mosfet being off to cut block the input load, but I thought a hard disconnect might be a better fail-safe? Before the relay kicks in the current draw is ~70mA, with the relay energized its ~170ma. I rounded to 200ma to allow for whatever else I might need to include?

So far I have...
* Arduino Nano as the brain (will drop to an ATTiny85 if I can get I2C going in software)
* ADS1115 - 4 channel 16-bit ADC.  Senses the 5v rail, load, and temperature of the heatsink
* LM335Z temperature sensor
* MCP4725 DAC for driving the power mosfet
* LCD 20x4 display.
* TL431 voltage reference
* LM324N OpAmp

Based on this discussion I should probably move the relay to the 12v rail (assuming it's acceptable).  This would reduce the 5v rail to 70ma and ~.49W

[Terry Bites]

I've used a teleprinter to decode the cuneform. Why would you need to run relay from a reference. Is it a voltage regulator you are wanting instead?

You are correct. In any set up like this the transistor has to offload (12-5)/0.2  or 1.5 W give or take.
A tin can 2N2222 can barely mange 500mW So you do need a beefy transistor and very likely a heatsink.

Adding a very basic current limit is straightforward. The current through sense resistor R6 creates a voltage drop, this turns on Q2 which robs Q1 of base drive.Imax is very roughly 0.7/R6.
You need to limit the output current from the opamp with R1 else the current limit transistor will be shorting it out.
You can offload circuit self protection using an LM317L: www.radiolocman.com/shem/schematics.html?di=644555

Note that emitter follower circuits like this have a tendecy to burst in to oscillation.
The precsion of the reference is being lost in the opamp and feedback resistors.
You need a good precsion opamp and low temco feedback resistors to get the set point and drift where you want it. Those will not be cheap.

Super cheap. Ditch the opamp.Still dissipates 1.5W. Works great and is easy to tweak.

Another way if you have a 3 terminal series ref.
Buffer a 5V reference. The reference's own precsion feedback resistors and opamp take care of the output voltage set point.
You can set a current limit (R2) to stop your transistor melting. R3 sets a minimum load current.
It still has to shed the same amount of heat.

[Terry Bites]

I've been thinking. A cheap MOSFET may be a better choice of pass element. I've added some small mods a tweak and improved protection.
No precsion parts needed but a good opamp like an OP77 is needed for good regulation over temperature.
.