Not a great solution, not least because of the reference voltage is one diode drop. The LM334-based design offers a 64mV drop across the current-sensing resistor, and the biased current mirror isn't much worse.
Regular diodes don't have a particularly well-specified forward voltage drop which means that R2 would have to be tweaked, and the transistor is being asked to dissipate several hundred milliwatt, so it will be running appreciably hotter than the diodes.
Exactly. 82R/680R should get you in the ball-park. You'd need to trim a 200R pot in series with 560R to get exactly 300mA through the LED at any specific battery voltage.
.pdf
Any circuit is going to stop working when the battery voltage gets too low; as the right-hand transistor goes into saturation not only will its Vbe rise but also its current gain will fall.
The purpose-built LED drivers that use a switched inductor to control the LED current should give a longer battery life than anything we can get out of a simple bipolar-based circuit.
I wouldn't like to drive even these LEDs from a voltage source - the resistance sn't all that high, and the forward voltage is temperature dependent and falls with increasing temperature, which - since the junctions run fairly hot - offers the chance of developing a nett negative resistance over periods longer than the thermal time-constant of the package.
snipped-for-privacy@ieee.org wrote: > On Jan 30, 3:25 am, Fred Bloggs wrote: >
As a first cut you can look at the datasheet for the cited Zetex job and locate VBE,ON for IC=300mA at VCE=2V, or active mode, and then go into VB,SAT at that VBE,ON level and read IC, it is roughly 100mA, an upper limit on how well you will do a Vbatt > The purpose-built LED drivers that use a switched inductor to control > the LED current should give a longer battery life than anything we > can get out of a simple bipolar-based circuit.
Zetex has a low voltage current mode boost controller described in DN61 IIRC which purposely tapers the current off at low batt, makes me wonder if that's not a good thing to extend the operating time at low batt versus a much more abrupt shutoff at continued high current.
The DS25 have a thermal of 15oC/W j-c, and easily heatsinked for a thermal of 30-60 oC/W j-a, so they will be a bit toasty at 1W dissipation. But if the D1 in the 431 circuit is snugged up to the LED on the heatsink, then Tcase will be the same for both components, making for a temperature difference of only 15oC at 1W, or -30mV for VLED. And since the sensitivity of Iled to both VF of the 1N914 and the VLED are the same and oppose at roughly 1.1, that makes for a 33mA net increase in Iled steady state, which you obviously adjust out with the pot R4.
I've bought a handful of LM334s and breadboarded a circuit. I'm using a short piece of nichrome as the reference resistor, and a ZTX1149A. The page linked to above suggests 390 Ohms and 0.1 uF to prevent oscillations, but I'm still seeing some oscillations with those components: about 200 mV at 50 kHz. This has reminded me that I don't know anything about filters! (Yes, I'm digital.) I could probably make it work by tweaking things - presumably reducing the resistor and/or increasing the capacitor - but I would prefer to actually know what I'm doing. Can anyone help?
Multicore arax (available through Farnell) can be used.
Since the flux is very aggressive, I wouldn't recommend it for PCB mounting. I prefer to use the arax wire to tin the ends, then clean it up, and use regular stuff for assembly.
Thanks. I picked up some "fluxite" zinc-chloride flux paste from the local hardware shop (quicker than ordering from Farnell) and it has done a great job.
Google suggested another alternative is to copper-plate the ends of the nichrome. I might have to try that idea too, just for entertainment...
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.