[sdiy] Op Amp LED driver / Current Source problem/solution

[Tom Wiltshire]

Love it. A really nice example of collaborative design!

On 27 Nov 2014, at 12:24, Justin Owen <juzowen at gmail.com> wrote:

> Thanks to everyone that chipped in on this - much appreciated.
> 
> Here's what I'm going with for now...
> 
> http://www.sdiy.org/juz/unipolar_2.png
> 
> Roman's point about whether the DAC was actually going to *zero* was a good one. I'm running off an analog output at the moment to simulate the DAC but no, I can't guarantee that it will until I've plugged it in. 
> 
> I think part of this was also my perception of how dim the LED should be at lower voltages but as Neil pointed out - a decent LED will still be glowing down to the microamps range. 
> 
> So, the above version 'looks' like it's performing the way I would expect it to as an end user.
> 
> Thanks all,
> 
> - Justin
> 
> 
> 
> 
> -----Original Message-----
> From: Neil Johnson [neil.johnson71 at gmail.com]
> Received: 27.11.2014 08:51:20
> To: Justin Owen
> Cc: SDIY List
> Subject: Re: [sdiy] Op Amp LED driver / Current Source problem/solution
> 
> Hi Justin,
> 
>> That version doesn't turn the LED fully off when the input signal gets to 0V...
>> 
>> ...but adding a big old 1M R to the supply rail fixes that:
>> 
>> http://www.sdiy.org/juz/unipolar_LED.png
>> 
>> Is that a valid fix or is this potentially going to cause problems elsewhere in the curcuit/make bad things happen?
> 
> You've got a couple of problems with the circuit as presented.  You've
> specified a TL072 so at this stage we can discount input bias
> currents.
> 
> With just one diode in the feedback loop you are going to get awful
> open-loop behaviour under half of the input range (when the diode is
> reverse-biased).  The impedance of the diode becomes very large when
> reverse-biased, so the op-amp gain will be very large too.  The end
> result is that as the control input approaches zero it is highly
> likely the op-amp output will peg to the negative rail.  As an aside,
> this is probably not a good thing for the LED to experience - over
> time it will fail if it sees too much reverse voltage across it (they
> are typically specified for 5V or so).
> 
> Why does the resistor "fix" the problem?  You need to consider the
> op-amp input offset voltage (Vio) - think of it as a little voltage
> source in series with the Vin+ pin.  Now, depending on the sign of the
> offset voltage you wil either find that even with the control input at
> 0V you still have some total control voltage due to Vio.
> 
> If you have a very efficient LED then even a few microamps could be
> enough to produce a very small glow (I have some blue LEDs that need
> microamps to glow nicely).
> 
> Your extra resistor introduces an external bias current that pulls the
> Vin- pin up a little bit, so that when your control input is at 0V the
> op-amp is trying to pull the Vin- pin down, but it can't due to the
> reverse biased LED, which gets turned off very hard.  So no glow, but
> not good for the LED.
> 
> The simpliest solution is a 1N4148 anti-parallel to the LED to protect
> it, and keep the 1M resistor to bias up the Vin- pin.
> 
> Phil's suggestion of a capacitor would help stabilise the op-amp
> during the transition between diode conduction (the point where both
> diodes are not forward biased so the op-amp is operating practically
> open-loop).  And check if your "0V" really is 0V as Roman suggested -
> any output offset from the DAC just adds to the problem.
> 
> Hope that helps!
> 
> Neil
> --
> http://www.njohnson.co.uk
> 
> 
> _______________________________________________
> Synth-diy mailing list
> Synth-diy at dropmix.xs4all.nl
> http://dropmix.xs4all.nl/mailman/listinfo/synth-diy