Off the same KA431 data sheet.
Look familiar

 

"Showing how critical it is to precisely control the LED intensity. "

that circuit shows that you don't need to precisely control the less intensity. It works this way: when the output voltage is too low, the tl431 outputs a logic high and the led is off.

when the output voltage is too high, the comparators output is low, turning on the led, which typically controls the duty cycle thus reducing the output.

digital mode totally works here.

 

Here's the feed back circuit with optocoupler.
View attachment 92380
Showing how critical it is to precisely control the LED intensity.

Does it? It controls the LED current, but that doesn't mean the LED intensity is tightly controlled. The intensity vs current curve is not flat. It varies with a number of parameters, including temperature and process variation. Plus, it does nothing about variation in the coupling gain between the LED and the phototransistor on the other side.

 

"Off the same KA431 data sheet."

the same analytical approach I outlined earlier would work for this circuit too.

 

Perhaps I'm not asking the right questions ?
Why do I need this circuit to provide constant current, when a BJT transistor is inherently already a constant current source.
The common-emitter circuit I showed earlier does this.
It also has a resister Re in the emitter that acts to cancel variations in Ic due to changes in beta due to temperature.
That is, without it, Ic = β.Ib. With it, Ic = (Vb - Vbe) / Re (Vb = base bias voltage).
Without confusing myself, or anybody else, how does adding the extra element improve this.
Obviously it acts to draw current away from the base as Ic increases, that is, negative feedback.
But Vbe is sitting right in its knee, or threshold region, and must change slightly.
This brings a large change in Ib. Have I just answered my own question ?

 

"The common-emitter circuit I showed earlier does this."

you asserted it to be a constant current source. You haven't proven it to be, yoy cannot prove it to be, because it isn't.

 

Anyone who sees the title of this thread and expects it to be useful with "understanding a constant current circuit" will be very disappointed .

 

dannyf
I had a look at your web page. I'm building a high power brushless motor controller to drive a car alternator to drive a buggy.
I noticed that it was important to use low ESR electrolytics to snub back EMF.
Anyway, I want to put my own web page up.
Did you write the HTML yourself, or use a web page builder.
And if so, which one. I'm trying a few out.

 

Anyone who sees the title of this thread and expects it to be useful with "understanding a constant current circuit" will be very disappointed .

Abandon all hope, Ye who enter.

 

The common-emitter circuit I showed earlier does this.

It doesn't, for obvious reasons: once the b-e junction starts to conduct, any increase in supply voltage would cause the base current, thus the collector current, to increase. The feedback introduced by R3 in your circuit helps but doesn't eliminate that problem.

So to make it a constant current source, you have to have a way to clamp the potential on the base. The npn does that, as does the TL431.

Another solution obviously is to replace R2 with a diode of sufficient voltage drop, like an LED.

 

Did you write the HTML yourself, or use a web page builder.

I basically just picked wordpress.com and never tried any other services. Simply typing it away. Very easy - if I can do it, anyone can do it,

 

That makes sense.
The second element keeps Vbe clamped.
I'm doing a full on analysis.
Be back in a couple of light years.

 

Perhaps I'm not asking the right questions ?
Why do I need this circuit to provide constant current, when a BJT transistor is inherently already a constant current source.
The common-emitter circuit I showed earlier does this.
It also has a resister Re in the emitter that acts to cancel variations in Ic due to changes in beta due to temperature.
That is, without it, Ic = β.Ib. With it, Ic = (Vb - Vbe) / Re (Vb = base bias voltage).
Without confusing myself, or anybody else, how does adding the extra element improve this.
Obviously it acts to draw current away from the base as Ic increases, that is, negative feedback.
But Vbe is sitting right in its knee, or threshold region, and must change slightly.
This brings a large change in Ib. Have I just answered my own question ?

It all comes down to how "constant" is constant, which is another way of asking when is "good enough", good enough?

The original circuit in this thread uses feedback to hold the voltage across R2 at just the voltage needed at the base-emitter junction of Q2 to get that much current out of Q1 and into R2. But the needed base-emitter voltage of Q2 is a function of how much collector current is flowing in Q2 -- as well as many other factors including the collector-emitter voltage across Q2, the temperature of Q2, and process variation from transistor to transistor. Those all result in an error between the desired R2 current and the actual R2 current. There are other effects as well, perhaps chief among them is the tolerance and stability of the resistance value of R2 itself. All of these have to be considered and placed into either the category of "it matters" or "it doesn't", depending on whether that factor, either alone or in conjunction with the others, has the potential to take the circuits performance from "good enough" to "not good enough".

The use of the KA431 or similar circuit let's us eliminate (or at least greatly reduce) many of these factors. It is nothing more than a circuit that presents whatever effective resistance between its two terminals is needed in order to maintain the reference input at a closely maintained 2.5 V. So that eliminates Vbe variations with temperature, current, Vce, beta or whatever else you want. You now only have (in that part of the circuit) the variations due to the finite gain of the amplifier within the KA431 and the error in the 2.5 V reference itself. At this point, you have almost certainly reduced the dominant error to the actual resistance of R2, which can first be mitigated by choosing a tight-tolerance resistor and, if that's not good enough, then controlling its temperature. At some point in there the variation in the beta of Q1 becomes an issue since what you really care about is not the emitter current going through R2, but rather the collector current going through all of the LEDs.

 

Please bear with me. I'm a total newb to electronics. I apologize if this question is stupid.

A stupid question is one not asked.

Where are you dsharp02, you started the thread.

 

An opto-coupler used in a critical feedback path of a switch mode power supply, for instance, would have to have its intensity precisely controlled.

As several others have already commented, the use in the posted references with a KA431 shunt regulator are not using the optoisolator in a linear mode.

 

,Realy.
But when it is switched on.
Does it have a constant current ?

 

Almost every real world off-line SMPS used optocoupler and TL431 in the feedback loop
https://www.fairchildsemi.com/application-notes/an/an-4137.pdf

This is (currently) throwing a 404 Error.

 

But when it is switched on.
Does it have a constant current ? [vulgar language snipped]

If you're responding to my post, the feedback loop only depends on the opto emitter being turned on or off; that's digital, not linear

In a linear application, the current in the emitter would be varied to affect the amount of light detected by the photo diode/transistor. The amount of light coupled between the emitter and detector is defined as CTR. CTR degrades over time, primarily due to emitter degradation, so designs using optocouplers in linear mode need to compensate for this or minimize the effects.

 

Sorry. It is linear. At least for the short time it transitions from off to on.
Sorry about the language.
Turrets. ?

 

This is (currently) throwing a 404 Error.

I downloaded the file before the link became broken...