LED brightness in circuits with multiple supported supply voltages

[aion]

I have an effect that I'd like to be able to run from 9v all the way to 24v. The circuit itself should be fine up until the opamp's max voltage which is north of 40V. My main concern is the LED. Is there a reliable way to ensure that the LED stays at roughly the same brightness regardless of whether it's 9, 12, 18 or 24 volts? Would a BJFE-style zener work well for this, or would it have trouble dropping the higher voltages? (This is in a 1590A enclosure, so I don't have space to add a voltage regulator.)

[Gurner]

How about something like a 7V regulator...

http://www.rct.ru/pdf/transistor/78l07.pdf

It'll always give 7V output & is rated up to 35V (though as you crank up the voltage the regulator might get a bit warm!).....then just use a series limiting resistor to drop the remaining voltage required for your led

Eg blue led approx 3.4V ...therefore you need to lose 3.6V across a series resistor .....work on say 10mA for the LED = 360R resistor.

You could use a lower value regulator like a 5V but then you're just placing more heat stress on the regulator....better to share the heat disappation workload between the regulator & the series current limiting resistor.

[GGBB]

I think I once read something about placing a reversed Zener diode in parallel with the LED to pass any voltage over the Zener rating directly to ground.  By reversed I mean the Zener's anode is at the LED's cathode.

[PRR]

If you have a lot of JFETs.....



Select one to give a nice current. Probably less than 20% change from 5V to 30V.

First time you run 24V, put your finger on the FET, see if it gets hot.

You can also run the LM317L (TO-92-size) with a resistor to make a constant current limiter.
http://www.ti.com/lit/gpn/lm317l  Figure 4

[R.G.]

(This is in a 1590A enclosure, so I don't have space to add a voltage regulator.)

Use a TO-92 package 100ma regulator, like the LM317L, about which see below.

How about something like a 7V regulator...
It'll always give 7V output & is rated up to 35V (though as you crank up the voltage the regulator might get a bit warm!).....then just use a series limiting resistor to drop the remaining voltage required for your led Eg blue led approx 3.4V ...therefore you need to lose 3.6V across a series resistor .....work on say 10mA for the LED = 360R resistor. You could use a lower value regulator like a 5V but then you're just placing more heat stress on the regulator....better to share the heat disappation workload between the regulator & the series current limiting resistor.

Good idea, good thinking. SOMETHING has to be regulated, either voltage or current, to get the LED to the constant-current that it needs for constant brightness.

I think I once read something about placing a reversed Zener diode in parallel with the LED to pass any voltage over the Zener rating directly to ground.  By reversed I mean the Zener's anode is at the LED's cathode.

Bad idea. What the zener does is steal current when the LED is forced so high in current that its forward voltage matches the zener. Very tough to do right. One other way is with two resistors in series to the LED, the zener from the junction of the two to ground. The LED conducts normally till the junction of the resistors and zener reach the zener voltage, then the zener and the top resistor eat the extra power. Still touchy to do well.

If you have a lot of JFETs.....
Select one to give a nice current. Probably less than 20% change from 5V to 30V.
First time you run 24V, put your finger on the FET, see if it gets hot.
You can also run the LM317L (TO-92-size) with a resistor to make a constant current limiter.

Good ideas. The JFET works if you select the JFET for the right current - which, if you think about it, sums up almost every JFET circuit. Pre-sorted JFETs used to be sold as "constant current diodes", but not any more that I'm aware of.

The LM317L is my choice. Takes one resistor if the biggest input voltage and LED current don't over heat it. Takes a resistor where the JFET doesn't, but then you can simply calculate the resistor and don't have to sort JFETs.  The JFET and LM317L will have to eat almost exactly the same power dissipation.

[PRR]

> Pre-sorted JFETs used to be sold as "constant current diodes", but not any more that I'm aware of.

They may be around. However DigiKey turned up only one small-quantity in-stock item, and it is 20mA and a whole buck:

http://www.digikey.com/product-detail/en/CLD20B-TP/CLD20B-TPMSCT-ND/2639874

http://www.mccsemi.com/up_pdf/CLD20B(DO-214AA).pdf - Page 2 shows curve-trace. 2V knee and 53V blow-up sure smells JFET-like. Page 3 shows LED limiting (plus AC feed). 17ma-23mA slop suggests decent process control (or ~~20% yield?).

Ah, DigiKey just isn't finding parts.

OnSemi has a range 10mA to 350mA. This is actually a "self biased transistor" (patent-pending), whutever that is. The 70% knee is down at 1V, which may/may-not be JFET inside. But what do we care? It is clearly aimed at LED limiting. And only half-buck.

http://www.digikey.com/catalog/en/partgroup/current-regulator/27882

10mA part: http://www.onsemi.com/pub_link/Collateral/NSI50010Y-D.PDF

LM317L seems better generally, since it also a half-buck, but is also useful in voltage-fixed work, so you may as well take the 10-pack price break, $0.38/each, leaves 12 cents for the resistor.

[R.G.]

I find I do tend to underestimate the prices of things, as I tend to buy quite a few at one time.

@OP:
If one has the space for two transistors and a couple of resistors, the standard old current clamp works.

One transistor (let's say it's an NPN, although PNP works if you reverse everything)  is set in series with the + power line, collector to +. A resistor from collector to base turns it on and lets current through. A second resistor in series with the emitter monitors the current and connects to the load, the LED anode in this case.

The trick is the second NPN, connected with its base to the first NPN's emitter, its emitter on the low end of the emitter resistor at the LED, and its collector connected to the base of the first NPN.

When the current in the emitter resistor causes a voltage drop of over about 0.5V, the second NPN turns on and steals the base current from the first NPN, limiting the current. You calculate the emitter resistor as R = 0.5V/(the current you want to limit at).

This set up is quite flexible. It's quite an old mainstay of the ham radio community; I think I first saw it in an ARRL handbook decades ago. If you use a MOSFET for the main current pass device, you can limit amperes and hundreds of volts this way. I designed a current clamp for tube amps which limits the current through the main rectifiers to prevent burning out tube rectifiers on switching out of standby, and to protect the power transformer in case of a shorted power tube or inductor.

[Perrow]

Source: http://www.girr.org/girr/tips/tips7/white_led_tips.html

Adjust the resistor to your liking.

From Tayda it's $0.13 (LM317L + resistor), led excluded.