crake

modified 8 years ago

voltage resistant current sinks

211

02:31:25

test voltage: 10-70V
current through the led changes very little ~5mA or less over the test voltage range. Contrast that with the current sink on the far left.

published 8 years ago

zorgrian

8 years ago

Ok so which of these circuits apart from the left hand one is best? What are the relative merits of each topology?

hurz

8 years ago

To compare the performance I would put them on a XY plot like this http://everycircuit.com/circuit/5602875530280960

zorgrian

8 years ago

Yeah thats good, but what are the relative merits of each of these circuits? I just plotted the bias current in a similar way. However, it seems that these topologies have different characteristics, maybe each one is good for certain applications?

zorgrian

8 years ago

I am guilty of using cheap LED constant current chips and even those low cost modules on eBay. This is great if you just want to light up the garden. What if you have a whole panel of control LEDs to light that need to be switched on and off?

zorgrian

8 years ago

I also admit that in these circumstances, i just use rule of thumb resistance calculation. Sometimes, this results in variation in luminosity which is annoying

abobaker

8 years ago

Even more improved current source. Note that the improvement will depend on the output resistance of the JFET .... http://everycircuit.com/circuit/5263851749900288

crake

8 years ago

Thanks @Hurz, using the XY plotter is something i always seem to forget about. @Zorgian it was expected that these would perform pretty much the same, but the zener version was a surprise. @abobaker dude, this is great. Definitely going to remember this one. Thank you.

hurz

8 years ago

Zorg, read my description and @carke, I dont agree with the zener surprise. The zener is much higher in reference then the other examples. So lower the voltage and compare again the performance.

hurz

8 years ago

@abobaker, what you have there with nmos and voltage source is already quite exact a.model for a JFet! I dont know why users here always ask for JFets and that easy its already on.

crake

8 years ago

I see your point hurz

abobaker

8 years ago

@crake you are welcome ;) , also good point @hurz is making about lowering the voltage, both the feedback and the zener versions are pretty close although I think the zener version would be better against temperature variations

abobaker

8 years ago

@hurz I thought the same about the JFET :)

zorgrian

8 years ago

If i need something like this I would choose the circuit with 2 diodes. This is because i have a large box of signal diodes. Not because its the best circuit

abobaker

8 years ago

After some testing, in real world, the best circuit is the Feedback one with a JFET bias, the zener version is only better here because zener emission coefficient is asumed = 1 which is not the practical case .. So I was wrong about zener being better against temperature .. Also using two diodes will also give emission coefficient = 2 so still feedback is better

crake

8 years ago

@Zorgian Kinda the same situation here. This circuit comparison was constructed using circuit components I already have plenty of.

zorgrian

8 years ago

Yeah i do simulate stuff like this, but if I use something i know works i just go get components from the workshop

zorgrian

8 years ago

I.e. just draw it out and build it

zorgrian

8 years ago

I build using RF dead bug style but actually its very robust. I use punched circles of PCB material for lands... Good for up to 5 leads of through hole components. I use this for almost everything save microcontroller stuff

zorgrian

8 years ago

Look, the two diode thing is practical, low cost, not perfect but will work

hurz

8 years ago

Mostly I do a current sink with a LED as reference, this gives enough headroom for the R to tune Vled-Vbe=3V-0.7V=2.3V which is a good value to tune almost any current.

add comment in editor