Constant current circuit

[FT Dude]

Dear readers,

This is my first post on the forum, so let me introduce myself.
I am electronics hobbyist and student in mechanical engineering. I do have some experience with making circuits and writing basic programs for PIC’s. I am a longtime viewer of Dave’s vlogs, but not this forum.

I have tried Google and the search function of this forum to get an answer for my questions.

I want to make a constant current source or sink for 20mA LEDs. 24 white and 6 Red LED’s.
The circuit is used outside where the temperature fluctuates between -10 and 35 degrees Celsius.
The circuit is battery powered so low power consumption is a must. It’s either 5V or the 2s LiPo voltage. (RC car)
The total efficiency of the solution must be greater than using a current limiting resistor.
No thermal runaway.
All components are SMD if possible.
Low weight and size solution.

I have different solutions in mind: led driver IC (but which one???) or current limiting transistor circuits parallel. And I have different variations for the LED’s connection in mind at the moment (6 in series and 5 parallel etc).

On the internet and in some documents I came across some circuits using transistors, that I have put in Multisim. See attachment

I have the following questions about the circuits itself and about the values needed to calculate the appropriate resistor seizes.

1)   What are the ad and disadvantages of A vs BCD circuit?
2)   Why would somebody choose circuit B or C, if a zenerdiode (D) is more voltage stable when junction temperature changes? (Forward Voltage vs. Junction Temperature graph).
3)   What voltage values do I need to use in the ohm’s law formula to have the transistors operate in linear mode? That Vbe(on) right? not Vbe (sat)
4)   Do Q1 and Q2 need to be the same type? otherwise I could use a lower hfe transistor for Q2 and a darlington for Q1.
5)   Most Datasheets of zenerdiodes state a test current Iz of 5mA. How low can I go with the current (efficiency) if Iz minimum is not given?
6)   For normal diodes the Forward Voltage vs. Junction Temperature graph shows (large) swings in voltages when temperature changes. Is the lowest line my minimal If for that diode?
7)   What LED driver IC is suitable for my application?

White LED: OSRAM LCW JNSH.PC 3,05V 20mA
Red LED: LR T67F-U1AA-1-1 2,05V 20mA
Q1: BC847C
R1: 100ohm trimmer, so I can adjust the LED current down to almost 5mA tested in my Multisim circuit)

Kind regards,

FT Dude

[saike]

I have used a lot of these constant current diodes. Not sure how they perform at different temperatures but it may be worth reading a few data sheets for them
http://uk.farnell.com/w/c/led-lighting-components/led-driver-ics/led-drivers?no-of-pins=2pins

[drussell]

I have used a lot of these constant current diodes.

Wow, I knew things like that existed but I didn't realize there was such a range of that kind of device now!  Duly noted, I think those could come in handy in a bunch of different situations.

In this case, something like a NSI45020A would probably do the trick, though in this instance it seems that he probably wants to run from a low voltage, so will need to boost voltage anyway, might as well use some sort of combination booster/regulator setup.

[FT Dude]

evb149
With your advice I am currently looking at some drivers.
I will make transistor circuit A (see my attachment first post) and a switching CC circuit to measure the difference in efficiency.

saike
Those are handy components. I discovered that those are or either CCD (Constant Current Diode) or a CCR linear (Constant Current Regulator). CCD is internally built with a JFET and a resistor and the CCR is built with a self biased transistor. So both are inefficient.

I will keep you updated when I am further with the project.

[macboy]

For any linear LED driver, you have losses equal to a simple resistor. The current through the LEDs is equal to the current coming out of the battery.  The benefit to a constant current source, whether built from discretes as in your first post, or in the form of a single device, is the relatively constant current as voltage changes. The discrete versions in your first post are in fact very slightly less efficient than a single resistor due to the small current through R2, which is not going through the LEDs. All other current goes through the LEDs.

For lower loss, you need some type of switching DC-DC converter, where the output (LED) current current can be higher than the input (battery) current.

[FT Dude]

Thank you for the help guys.

I have chosen the following driver: TPS61169DCKR  http://nl.farnell.com/texas-instruments/tps61169dckr/led-driver-boost-1-2mhz-sc-70/dp/2542731 (I can't get the hyperlink to work).
I will use 2 of them with each 1 string of 12 white LED's in series.
When my Inventor design is in a later stadium I will know what my dimension specifications are for the PCB.

Can somebody still tell me the answers for the following questions:
1)   What are the ad and disadvantages of A vs BCD circuit? (transistor vs zener vs diodes)
2)   Why would somebody choose circuit B or C, if a zenerdiode (D) is more voltage stable when junction temperature changes? (Forward Voltage vs. Junction Temperature graph).
3)   What voltage values do I need to use in the ohm’s law formula to have the transistors operate in linear mode? That Vbe(on) right? not Vbe (sat)
4)   Do Q1 and Q2 need to be the same type? otherwise I could use a lower hfe transistor for Q2 and a darlington for Q1.
5)   Most Datasheets of zenerdiodes state a test current Iz of 5mA. How low can I go with the current (efficiency) if Iz minimum is not given?
6)   For normal diodes the Forward Voltage vs. Junction Temperature graph shows (large) swings in voltages when temperature changes. Is the lowest line my minimal If for that diode?

[Audioguru]

You say your white LEDs are 3.05V. No they aren't, they have a range of voltages and you did not indicate which range you have. It might be 2.8V or 3.4V or anything in between.
The LEDs are tiny, can you solder to them?

[FT Dude]

Yes I can solder those LED's. I have done PCB rework for a company in the past. The lack of special tools at home is the limiting factor. This will be my first time with my own tools.

You say your white LEDs are 3.05V. No they aren't, they have a range of voltages and you did not indicate which range you have. It might be 2.8V or 3.4V or anything in between.

Do you mean do group number in the datasheet? Or bin code and sub bin code for the following LED: CLM3C-WKW-CWBYA453 http://www.farnell.com/datasheets/2047854.pdf
If so, can you help me to figure that out. Because a part of the model number is referable in the datasheet, but the exact group or bin is not.

My chosen OSRAM LED: LCW JNSH.PC-BUCQ-5H7I-1 http://www.produktinfo.conrad.com/datenblaetter/500000-524999/521368-da-01-ml-LED_LCWJNSH_PC_BUCQ_5H7I_1_de_en.pdf

[hamster_nz]

evb149
With your advice I am currently looking at some drivers.
I will make transistor circuit A (see my attachment first post) and a switching CC circuit to measure the difference in efficiency.

saike
Those are handy components. I discovered that those are or either CCD (Constant Current Diode) or a CCR linear (Constant Current Regulator). CCD is internally built with a JFET and a resistor and the CCR is built with a self biased transistor. So both are inefficient.

I will keep you updated when I am further with the project.

All these will perform pretty much the same in terms of efficiency. Maybe A having a slight edge for temperature stability, esp if the transistors are closely matched.

A while ago I made a board of a 8-bit shift register, and sixteen discrete 20mA drivers (two per bit) using something like design A. It worked really well.

If you are powering from 5V, you won't have enough voltage to drive two 3V LEDs in series, and you will be wasting 40% of your power in the current limiting resistor and transistor.

If you are driving from 12V, with three 3V diodes in series then you will only be 'wasting' at most 3V in the current limiting resistor and transistor, so will already will have around 75% efficiency, so any extra expense in building drivers will only extend run-time by 33% with a perfect switching CC driver.

Would it not be simpler to reduce the LED current to match battery life, or get a slightly larger battery than add the complexity of using a switching LED driver?