Use od LED for CCS and cascode stages

[Mambo]

From Your knowledge/experience are there any cons in using LEDs for current sources and cascode stages ?

Thank You

 

[Zldz]

Rod Elliott uses a green led in his famous PROJECT 3A.
http://sound.westhost.com/project3a.htm

Ciao
Lorenzo

 

[peranders]

From Your knowledge/experience are there any cons in using LEDs for current sources and cascode stages ?

No, not really but it looks cool. The disadvantage is the "unprecision". The voltage is very dependent of brand. It's better to use a normal silicone PN-junction or even better a 431.

But it's not bad when the LED is there and tweaked in, but it hopeless if you want repeatability and precision.

 

[mikek]

Re: Re: Use od LED for CCS and cascode stages

From Your knowledge/experience are there any cons in using LEDs for current sources and cascode stages ?

Thank You

No, not really but it looks cool. The disadvantage is the "unprecision". The voltage is very dependent of brand. It's better to use a normal silicone PN-junction or even better a 431.

But it's not bad when the LED is there and tweaked in, but it hopeless if you want repeatability and precision.

Per is right with regard to variable tolerances from one LED to another.....However one significant advantage is that LED's provide very effective temp. compensation for your current source....compared to ordinary diode-biasing.

 

[peranders]

The temperature compensation isn't so important in most cases as long as the current source has sufficient feedback and the environment is "home". IF it's important it's better to use a monolithic pair instead, cheap but hard to solder.

BC847BS at the picture
--- can't reach my homepage right now. But I have a picture at the bottom of the page
http://home.swipnet.se/~w-50719/hifi/layout_qsxm2.html
http://home.swipnet.se/~w-50719/hifi/qsxm2_layout.html
I'm not sure of the file name.

 

[peranders]

The links weren't right any of them.

An externally hosted image should be here but it was not working when we last tested it.

 

[bocka]

No, not really but it looks cool. The disadvantage is the "unprecision". The voltage is very dependent of brand. It's better to use a normal silicone PN-junction or even better a 431.

But it's not bad when the LED is there and tweaked in, but it hopeless if you want repeatability and precision.

No, I highly disagree with you. If you're not using bulk LEDs but only from one manufacturer the tolerances are very small, about +/-1% I measured in a lot of 200. The TL431 is not better.

A LED provides very low noise, too. I can't see any reason why using a TL431 instead of a LED. Use a brand like a TLLR3360.

 

[peranders]

I can, the dynamic resistance is under 1 ohms in a 431 and the temperature stability is better and the initial tolerance better, but it isn't the same type of component.

 

[bocka]

Without coupling the ref pin to emitter of a (NPN) the temperature stability is worse. Simply couling the TL431 with the ref to a PNP is not possible. Noise is higher. Initial tolerance is not better (refering to the datasheet). 1% of a LED is the same as the resistor used to set the current of the CCS. So I can't really see any reson why using the shunt reg, although it'll make the job.

I'd strongly recommend the LED.

 

[peranders]

I'm talking more from the perspective of series production. You can't really on purpose design in a part which is so special. Parts with no second source is a nightmare.

 

[bocka]

I don't think it's a problem with volume production, if you're only using LEDs from one manufacturer. You can change them. I've made some designs with LEDs as voltage reference, tolerances of a bjt are higher. I'm astonished how small the tolerances of a LED are. Maybe because they are selected to several intensity groups...

Parts with no second source is a nightmare.

Yes, it is. Today many comlex parts like communication controller, FPGA, memory, ADC etc. have no second or third source. Give me fits more and more

 

[mikek]

No, I highly disagree with you. If you're not using bulk LEDs but only from one manufacturer the tolerances are very small, about +/-1% I measured in a lot of 200. The TL431 is not better.

A LED provides very low noise, too. I can't see any reason why using a TL431 instead of a LED. Use a brand like a TLLR3360.

I agree completely...thanks for your input bocka..

 

[AndyM]

Not to forget that LEDs have a steeper I-V-characteristic than low voltage zeners or a double normal diode.

 

[L Daniel Rosa]

I like the complementary BJT CCS because it allows for a two wire inline regulator. I know full well that it requires tweaking, but the simplicity is it's prime appeal. There was some discussion recently at http://headwize2.powerpill.org/ubb/showpage5.php?fnum=3&tid=4179&fdays=20 and there are some illustrations as well.

The '431 is an excellent component but does have some liabilities in this application. The current to the '431 would have to be regulated as well and is prone to oscillation under some conditions.

 

[keyne]

that is a neat circuit!
thanks!
unfortunately a bit limited in max current, because the LEDs each take about half of it.
my simulator shows impedance of about 3 meg-ohm, very nice for this simple ckt. I will keep it in mind.