Hi,
I would like to use this current source (schematic stolen from Wiki) for stabilizing current of IrLED in my remote:
The problem is maximum voltage over Load is Vs - VR2 (voltage over R2). I got "great idea" to put R2 between the voltage stabilizing LED and base of the transistor. In basic theory current through the Load will be fixed multiple of base current. Problem is exact ratio depends on so many variables.
I know it changes from one transistor to another - it is no problem, I can measure the one transistor I will use and use resistor to get the result.
It also changes with temperature - I will use it indoors and power dissipation should be small (~20 mA for a few seconds with average duty much less than 10%).
But there are definitely many other problems I don't know about but that may affect current through the load. Do you think it is possible for this to work with around +/- 20% error from original values? What I am afraid of at most is aging of the transistor - I may try everything on breadboard, it may seem optimistic but what about 2 years later?
I want to power it with coin battery. The battery supplies 3V when it is fresh (and probably close to 3V for most of its life) slowly decreasing to 2V. If I use current limiting resistor for 20 mA@2V it will be around (2-1.2)V/0.02A = 40 Ohm. With 40 Ohm current limiting resistor and 3V of fresh battery I will get (3-1.2)/40 Ohm = 45 mA current. I think it is not so important difference but I want (red) LED to indicate the remote is transmitting, current limiting resistor for it, transistor to drive the LED and resistor to its base to limit current anyway. If I use this circuit I don't need current limiting resistor saving one AND possibly get better results. I planned to use exactly this circuit (but R1 connected to Arduino pin instead of Vs) but now I "discovered" this upgrade helping to overcome problem when voltage is low: If LED is red with 1.8V forward drop then voltage over R2 should be about 1.2V which is too much with 1.2 forward drop of Ir LED and only 2V supply voltage.
If you want it's current to be accurate down to a 2v supply a Wilson current mirror would do it.
But by the time a 3v coin cell has dropped to 2v it's internal impedance will be high - it's nearly flat and won't deliver much current . For pulses a large reservoir capacitor would help.
Smajdalf:
The problem is maximum voltage over Load is Vs - VR2 (voltage over R2). I got "great idea" to put R2 between the voltage stabilizing LED and base of the transistor. In basic theory current through the Load will be fixed multiple of base current. Problem is exact ratio depends on so many variables.
Doesn't work, without R2 the base voltage is a diode drop above the emitter so the LED is completelt off.
What you can do is just have a high value resistor from base to supply to set the base current at
a roughly fixed amount. IE remove R2 and the LED, make R1 much higher in value. R1 will have
to be 'select-on-test' (ie each transistor will need R1's value determined by measurement).
But its still not really addressing the issue of accuracy as the current gain of a transistor varies dramatically
as the emitter-collector voltage drops and the device approachs saturation.
Sure it will work. Maybe I expressed it wrong. I want control base current instead of emitter current (one end of R2 between R1 amd LED, the other to base). The question is - how silly is this idea?
It will probably give less accuracy than just having a series current-limiting resistor on your load...
You could add the LED back in to provide a stabilized voltage from which to feed the base with
the high value resistor.
That's like moving R2 to between the junction of R1/LED and the base.
Perhaps that what you originally meant, come to think of it - but you didn't draw it!
MarkT:
Perhaps that what you originally meant, come to think of it - but you didn't draw it!
I have problems explain myself in this topic. The original picture is copied from Wiki without any change. Here is "better" schematic:
Currently I put a diode in place of "?" leaving R2 where it is drawn. It reduces voltage dropped over R2 so circuit works down to 2V but R2 also needs to be smaller - I think smaller R2 and another "constant voltage" diode drop introduces additional inaccuracy. If it could be solved by moving R2 from its original position in place of "?" (and make it much larger) it would be nice. But I am afraid beta of a transistor is not stable enough to allow this - so I asked for advice.
allanhurst:
A typical IR led drops about 1.6v.
MY IR LED has forward drop 1.35 typical and 1.6V maximum with pulsed 100mA (Datasheet says). With lower currents it is closer to 1.2V (DMM says). But it is not so important, it only limits maximum voltage R2 may drop in original configuration.
JohnRob:
To do what you want, I would look at a TL421.
Google didn't come with anything useful - can you post a link please?
CrossRoads:
How long do you expect it to run? Coin cells don't have a lot of capacity.
For years - estimated average current draw will be in order of microamps. Just sleeping when no button is pressed. It is why I am concerned about long term stability.
Original diagram is ok, except for the LED.
A LED drops too much for this low voltage application.
Replace the LED with two 1N4148 diodes in series.
Then you only have a ~.065volt drop across R2.
Calculate R2 accordingly.
Calculate R1 for 1/10 of the IR LED current, assuming the two diodes in series drop 2*0.65volt.
Leo..
I am not sure any of this is required.
Remotes normally use short pulses of high current to achieve thr required range.
Just use a suitably sized reservoir capacitor.
