LED current regulation in several parallel strings of LEDs

led allowed current

hello,

is this a good way to regulate current in several parallel strings of LEDs?



....i mean will all LED strings have the same current in them if they are the same brand of LED?

Re: LED current regulation

The concept is OK. Your matching is going to depend on the NPN matching as well as the matching of of the emiiter degeneration resistors. So depending on how tight you need the current matching to be, you may or may not meet spec.

Also, you have to be careful driving 3 white LED's in series. I have seen the Vf voltage of white LED's be as high as 4V. If that is the case you don't have enough head room. Around 3.2 is a typical value, which is putting your NPN's very close to saturation so be careful. If just one of the NPN's saturate you are going to screw up the other 3 strings.

LED current regulation

thankyou haff99.......they are not matched as such...they are the same parts but not matched.

I just wopndered if the feedback given by the 4R7's is a waste of time?

I mean, will they improve regulation in the strings that much?

By the way, this circuit gets used in a product which sells worldwide.....but i can't understand how it guarantees equal current in each LED?

Re: LED current regulation

The emitter degeneration resistors are critical. Adding emitter degeneration makes the matching of the NPN's mirrors much better(NPN current mirrors is basically what you have).If you do the math you'll see that emitter degeneration inmproves the current matching by decreasing the effect the NPN mismatch. your mismatch would be much greater if you got rid of those resistors.

Re: LED current regulation

I just wopndered if the feedback given by the 4R7's is a waste of time?

Click to expand...

You should calculate instead of just wondering.

In the shown dimensioning, you get about 0.8 V voltage drop at the resistors, this seems to be sufficient to get e. g. < 10 % current matching with usual parameter variations. Cause you won't get guaranteed transistor parameters from a manufacturer, you may assume 20 % to have a safety margin. Thus the LED should have about 45 mA minimum current capacity.

Re: LED current regulation

eem2am,
This circuit regulates the total current thru the 4 parallel paths to 160 mA. This is OK if the LEDs are matched. Otherwise, if one path has one or more LEDs with a lower voltage at 40 mA, this path will have more current and the other paths will have less. To avoid problems due to LED mismatch, you need 4 independent current regulators.
Regards,
Kral

LED current regulation

Even if LEDs are made by the same manufacturer they are all a little different.
Some are made in a rush on Friday. They are poor quality.
Others are made on a Monday, and are not good.
Ones made on a Wednesday are the best but they are quickly bought by flashlight manufacturers.
The manufacturer will not tell you when they were made but have BINS with sorted forward voltage, brightness and colour.

Re: LED current regulation

I think you may as well just use 1 current limiting resistor in each string, it would probably give equal performance as your current setup.

The best way is what Kral suggested regulate each string. Either get a cheap quad opamp or some dual transistors for mirrors. TI also makes some inexpensive SMPS boost LED drivers,so you could just use 1 string.

But a simple current limiting resistor in each string might be good enough.

Re: LED current regulation

I like do emphasize my opinion, that it's basically a meaningful design. I wonder, why many contributors apparently don't understand the obvious stabilizing effect of the transistor current sources, that provide a LED current independent of individual LED voltage drop. Of course a quad OP can give better current matching, but I don't think, that it's required by the application.

Re: LED current regulation

If somebody goes through the time and expense for a constant current source I assume equal current through each string is important. If this is the case then at this point why not just do it properly he already has 4 BJT’s and 5 resistors.

I’m not disagreeing it doesn’t have advantages over a single current limiting resistor per string but I don’t think its that significant to warrant all the parts. My point is if your design can tolerate those tolerances then it could probably tolerate a simple current limiting resistor. It would also be more reliable.

LED current regulation

Well its good to see such debate.

And indeed, may i tell that this very type of circuit gets used in a LED product which sells the world over.....its mighty cheaper than most other ways of doing it.

I am trying to calculate the differences expected in current , bit i can't, as it depends on the individual transistor parameters which i do not know.

in fact, i missed off the diodes which limit current if a LED fails open...........

LED current regulation

hi

In the real circuit, the opamp (+) voltage is variable so that the LEDs can be dimmed.

I cannot use any more opamps since the offset needs seeing to and this requires resistor changing at final assembly and test time, so many op amps would not be practical.

By the way, say the LED strings contained different amount of LEDs -would the circuit still keep all LED currents similar? (within 10% of each other)?

Re: LED current regulation

eem2am,
I don't see why offset voltage would be a problem. You have a 1.6V reference. The offset voltage of any decent op-amp would be very small compared to 1.6V.
.
As to using different numbers of LEDs in each string; this won't work for the same reason that mismatched LED voltages would cause unequal currents in each parallel path.
Regards,
Kral

Re: LED current regulation

As to using different numbers of LEDs in each string; this won't work for the same reason that mismatched LED voltages would cause unequal currents in each parallel path.

Click to expand...

This has been said:

I wonder, why many contributors apparently don't understand the obvious stabilizing effect of the transistor current sources, that provide a LED current independent of individual LED voltage drop.

Click to expand...

As a numerical example. A BJT has e.g. an early voltage (slope of Ic=f(Uce)) of 100 V. This means a 5V LED voltage difference causes only 5% current variation, as long as saturated operation is maintained.

LED current regulation

A transistor displaying the Early effect has plenty of collector to emitter voltage and is far from saturation.

The circuit in this tread is an odd current regulator. Usually an opamp driving a transistor makes an excellent current regulator.

LED current regulation

Audioguru...i think FvM meant "avoided" not "maintained"?...since the NPN's should avoid saturation.

kral:- I appreciate what you ay.......though in the actual circuit the 1.6V is actually a variable voltage...........and there are three of the above subcircuits on each board (R,G&B).

.....now when the variable voltage is adjusted for minimally lit LEDs as our light_quality people say........the small offset voltage does then actually make a difference and some LED banks are lit and others are not....and we want them all to be lit at this particular low opamp input voltage....so we have to address the offset of each opamp........so we only use one opamp per LED bank.....because otherwise its a lot of resistors to solder in/out at final test time.

Re: LED current regulation

i think FvM meant "avoided" not "maintained

Click to expand...

Yes, thank you. I was think of unsaturated, active-forward operation but writing saturated.

The circuit in this tread is an odd current regulator.

Click to expand...

It has limited capabilities, but it isn't odd. BJT current sources are used in many electronic application fields, where there properties are sufficient, e.g. audio amp bias. In the LED application, the brightness differences due to individual LED parameter variation can be expected a multiple of the current variation effect, even with select exemplars, so it's pretty O.K., I think.

Regarding the above discussion of OP offsets, please consider a realistic current variation factor, e. g. 10:1 and a realistic offset, e.g. +/- 5mV. You'll see, that there isn't a real issue. But I think, the more simple circuit, as suggested, is just O.K.

LED current regulation

hello,

yes indeed it is a magic circuit........pse see this simulation where it magically keeps the three currents the same, in spite of different numbers of LEDs...



...waveforms for above schem


GREEN IS CURRENT IN 3 LED STRING
BLUE IS CURRENT IN 2 LED STRING
RED IS CURRENT IN 1 LED STRING

-------------------------------------------------
Now take away the 4R7 resistors and see how much worse it is........



...and the (much worse) waveforms).......


GREEN IS CURRENT IN 3 LED STRING
PURPLE IS CURRENT IN 2 LED STRING
RED IS CURRENT IN 1 LED STRING

---------------------------------------------------
Now see the currents with just a 160R resistor in each string..........



...Waveforms for above....


GREEN IS CURRENT IN 3 LED STRING
BLUE IS CURRENT IN 2 LED STRING
RED IS CURRENT IN 1 LED STRING

--------------------------------------------
-Please notice that current mismatch is WORSE when in the circuit with no 4R7 resistors than it is with NO TRANSISTORS at all ! (wow!)

--------------------------------------------------
--------------------------------------------------

And please notice in the circuit containing the 4R7 feedback resistors (first circuit above) that i used dufferent transistors and resistors to simulate tolerances variations.

----------------------------------------------------------
Also, as i mentioned, this circuit is really for use as a dimmer with a variable opamp reference voltage.......there are three banks of LEDs in the above configuration (R,G,B).......and if the offset of the opamp is not sorted, then at the most dimmest setting....some banks are ON and some are not.....and customers dont like this...so we have to address the opamp offset.............-And this in turn is the reason why we cannot use a separate opamp for each transistor....because it would mean doing too much offset adjustment at final test time, which takes too long.
---------------------------------------------------

anyway, i wondered if any reader has spotted a hole in the above top circuit?

LED current regulation

The three transistors are completely different. The tiny little surface-mount BCW60 is smoking at a dissipation well above its max allowed dissipation.

Try the simulation with the same transistors and without the 4.7 ohm resistors.

The tiny little surface-mount LEDs are also smoking and are operating at currents well above their absolute max allowed current (30mA continuous).

Reduce the current so that the parts are cooler.

LED current regulation

Regarding non-degenerated current sources (2nd circuit in your test) you should consider, that the different transistor types possibly use the same default model. To estimate worst case behaviour, you have to use the specified parameter variation from a datasheet (if specified at all). Also temperature differences occur, resulting from different Uce. Thus I wouldn't base a design on Vbe matching of individual transistors. But with degeneration resistors, it's O.K. for a lot of applications.