simple 150uA constant current supply

"John Fields"

Makes a big difference....

...... Phil

If it is a micro power op-amp, the bypass does little to help. The internal compensation workings of the op-amp make the PSRR to the minus rail so bad that the bypass on the 3.9K doesn't gain much.

How about this for a crappy circuit:

Dambit! my cursor just went away so I describe not ascii art the circuit Assume a micor power rail to rail op-amp.

If the minus rail of the op-amp goes to the load, there is a resistor in the plus rail and the inverting input and output are also hooked to the plus rail, you get a truly awful constant current source.

The opamp's output will pull down on the resistor sinking current to the load on the minus power.

at=20

here=20

=20

Nice!

--=20 KEith

--=20 Keith

You'd be best to simply ignore inflamatory traffic on the net. It seves no useful purpose.

What kind of LED are you working with and would an LM10BL with one sensing resistor work for you? The reference is 200mV and supply current ~300uA.

Others might have ideas for low current, low voltage references that might complement the tlv2760's 20uA supply current.

RL

I found an LT6650 that combines a 400 mV reference and an op-amp with a supply current of 5.6 uA typical, in a 5 lead SOT23 package. I made a SwitcherCad simulation, with one 2.7 kOhm resistor, that holds LED current to 148.524 uA (exactly), from 2.2 volts to 5 volts. Even better, it draws only 17.75 uA from the battery, so you get an efficiency of 770%. Yes, 39 uW in and 300 uW out. There must be an error in the model, but I think it will run on 170 uA. Of course the LT6650 costs $1.88, but *what a deal* for an "over unity" part!

This has been an interesting thread, whether or not it has helped the OP. The designs offered are useful in general, and I've learned a few things.

Paul

=====================================================================================

Version 4 SHEET 1 880 680 WIRE 112 144 -80 144 WIRE 112 208 112 144 WIRE -80 224 -80 144 WIRE 320 256 224 256 WIRE 384 320 384 256 WIRE 384 320 224 320 WIRE 384 336 384 320 WIRE -80 416 -80 304 WIRE 112 416 112 368 WIRE 112 416 -80 416 WIRE 384 416 112 416 WIRE -80 464 -80 416 FLAG -80 464 0 SYMBOL SpecialFunctions\\\\LT6650 112 288 R0 SYMATTR InstName U1 SYMBOL voltage -80 208 R0 WINDOW 123 0 0 Left 0 WINDOW 39 24 132 Left 0 SYMATTR InstName V1 SYMATTR Value 2.2 SYMATTR SpiceLine Rser=.1 SYMBOL res 368 320 R0 SYMATTR InstName R1 SYMATTR Value 2.7k SYMBOL LED 320 272 R270 WINDOW 0 72 32 VTop 0 WINDOW 3 0 32 VBottom 0 SYMATTR InstName D1 SYMATTR Value QTLP690C SYMATTR Description Diode SYMATTR Type diode TEXT -114 506 Left 0 !.tran .1 startup

I have taken measurements that are almost identical. The first LED video screen I worked on had a circuit that measured the LED forward voltage. The way it was configured pumped 1uA into the LEDs, and was (barely) visible. It turned out to be easier for me to measure sub-microamp "leakage" currents with a vacuum cleaner. We had bugger all test gear (almost everything I used I was renting to the company), and the "nook & cranny cleaning" pipe from the end of the vacuum cleaner was thick black plastic that tapered down to a narrow slot; black tape made it narrower still. By placing it over a led with my eye firmly over the other end, I could see well below 1uA.

the problem turned out to be easy to fix, and was inherent on the design of the voltage sensing circuit (not mine), which I duly modified. 3 years later I left, and the guy who designed the original circuit changed it back the next PCB revision (he never did like me). Funnily enough, they had a huge problem with contrast (black isnt very black when all the LEDs are glowing faintly), and had to do another PCB respin. He didnt use my exact fix, rather its dual, which ended up being more expensive. pants-wettingly funny. That firm made all their technical staff redundant this year (including him), and no longer build anything.

Cheers Terry

Top marks.

RL

Let me guess - they went onto fabless dsp-intensive AV controllers.... with a new crop of wonder boys, known to the new venture capitalists.

RL

Just be careful not to suck your eyeball out.

I've met a few really pig-headed engineers like that. They can't stand for anybody else to be right. He's probably selling insurance or something now.

I love that British term, "redundant."

John

even worse. they re-badge shit chinese product, and rent out advertising.

and (with one notable exception ;) pretty much the entire original design team is now designing a new one, for some other people. entirely new & original, but boy does experience help.

Cheers Terry

there

I just used the plastic nozzle. But I'll be careful :)

he had a serious NIH problem, and was permanently unhappy that he was never put in charge of R&D. The first time he mentioned wanting it,

he was remarkable in that he could be guaranteed to find the most complex, most sensitive, most expensive, least useful solution to any problem. Asking his opininon and then specifically not doing that is a sound design strategy :)

Cheers Terry

=====================================================================================

That's okay but you would want to buy yourself a bit more headroom with something like this: View in a fixed-width font such as Courier.

. . . . . . LTC6650 . . |\\ . -----|+\\ ~~ . | | >-+----+--|>|--. . 400mV .-|-/ | | | . | | |/ === [110k] | . | | 1u| | | . --- | | | | . '------|----+-[22k]-+ . | | . | [300] . | | . --- --- . . . . . . . .

=====================================================================================

Yes, that could be done if you really needed it to work below 2.2 volts supply. But comes at the expense of more parts, less efficiency, and not quite as good current regulation if the diode Vf varies due to temperature, production lot, aging, etc. The battery is probably just about shot at 2.2 volts, and a gradual dimming might be a "good thing".

I didn't notice the requirement for the 1 uF capacitor. Thanks for pointing that out. Now the simulation shows a little start-up oscillation, but the model is still flawed, since I am still drawing only 18 uA from the supply for 150 uA output. And it's not April 1 anymore.

Paul

================================================================================

Version 4 SHEET 1 880 680 WIRE 112 144 -80 144 WIRE 112 208 112 144 WIRE -80 224 -80 144 WIRE 256 256 224 256 WIRE 320 256 256 256 WIRE 384 320 384 256 WIRE 384 320 224 320 WIRE 384 336 384 320 WIRE 256 352 256 256 WIRE -80 416 -80 304 WIRE 112 416 112 368 WIRE 112 416 -80 416 WIRE 256 416 112 416 WIRE 384 416 256 416 WIRE -80 464 -80 416 FLAG -80 464 0 SYMBOL SpecialFunctions\\\\LT6650 112 288 R0 SYMATTR InstName U1 SYMBOL voltage -80 208 R0 WINDOW 123 0 0 Left 0 WINDOW 39 24 132 Left 0 SYMATTR SpiceLine Rser=.1 SYMATTR InstName V1 SYMATTR Value 3 SYMBOL res 368 320 R0 SYMATTR InstName R1 SYMATTR Value 2.7k SYMBOL LED 320 272 R270 WINDOW 0 72 32 VTop 0 WINDOW 3 0 32 VBottom 0 SYMATTR InstName D1 SYMATTR Value QTLP690C SYMATTR Description Diode SYMATTR Type diode SYMBOL cap 240 352 R0 SYMATTR InstName C1 SYMATTR Value 1µ SYMATTR SpiceLine V=16 Irms=1.816 Rser=0.031 MTBF=0 Lser=0 mfg="KEMET" pn="C0805C105K4RAC" type="X7R" ppPkg=1 TEXT -112 504 Left 0 !.tran .05 startup

I don't know that a simple LED indicator drive requires all that expense and precision. You're the one who went with a $2 versus a $0.45 part. I don't see that it's justified unless something else in the product requires a reference. I did jump the gun on the first circuit, absolutely terrible dI/dV, unusable actually. Taking the self-referential opamp circuit as a minimum acceptable performance, with dI/dV=-I/V, the reference circuit meets it with overheads, VOH, in excess of Vref/2. This gets you good-enough regulation down to 2V, and you have a reference you can tap into if necessary. View in a fixed-width font such as Courier.

. . . . . . . . |\\ V . .-----|+\\ + - Vref-VOH . | | >---+---|>|----. k= -------- . Vref .-|-/ | ~ | V . | | |/ [R1] || I . | | | |v . | | | | dI I Vref-VOH . | '--------+--[R2]----+ + -- = - - * -------- . | + - | dV V VOH . | kV [Rcs] VOH . | | - . '-----------------------' > Vref . => VOH = ---- . 2 . LT6650 . . |\\ V . .-----|+\\ + - . | | >---+---|>|----. . 400mV .-|-/ | ~ | . | | |/ [120k] || I k=0.1 . | | | |v . | | | | . | '--------+--[13k]---+ + . | + - | . | kV [1.2k] 200mV . | | - . '-----------------------' . .

I don't know that the 1u is necessary unless there is capacitive loading somewhere, the amp is unity gain stable, and they seem to go to great lengths to filter the power supply because the PSRR is *real* bad on that part...

================================================================================

Fred Bloggs a écrit :

=====================================================================================

I've not seen the whole thread, so...

You don't need the ref source and get Vbatt independence by using the LED as the ref voltage.

___ ___ .-|___|---+---|___|--. | | | === | |\\| | GND '-|-\\ | | >-----+ .-|+/ | | |/| .-. | === | | | GND | | | '-' (*) | | ___ '----------+----|___|--- +V | NC V ->

- | === GND

(*) using a GND supplied opamp with a bipolar output stage ensures startup, or add this resistor.

=====================================================================================

Right, we covered that earlier, and have been calling it the self-referential circuit. That is the one with dI/dV=-I/V , I and V are for the LED. The OA was a micropower type.

e:

=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D= =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D

The LT6003 seems to be happy with no bypass capacitor even though Linear doesn't seem to think so.

You can also flip the circuit around so that the LED's anode is at Vcc. This would let you use a "single supply" op-amp.

Or perhaps with higher battery voltage: (Modified version)

Many R-R input op-amps still work if the inverting input is pulled below ground while the non-inverting input is at ground. This circuit uses the op-amps idle current to ensure starting. The down side is that the op-amp needs to be a very low voltage one.

I made a self-referential circuit using a 2N3904 and 2N3906. It works in the SwitcherCad simulator down to 2.1 volts. It may not be very stable with temperature, but it's cheap and easy.

Paul

===========================================================================

Version 4 SHEET 1 1172 680 WIRE -16 0 -256 0 WIRE 80 0 -16 0 WIRE -16 32 -16 0 WIRE 160 80 80 80 WIRE 80 96 80 80 WIRE 160 112 160 80 WIRE -16 144 -16 112 WIRE 16 144 -16 144 WIRE -256 176 -256 0 WIRE -16 208 -16 144 WIRE 80 224 80 192 WIRE 160 224 160 192 WIRE 160 224 80 224 WIRE 80 256 80 224 WIRE 80 256 48 256 WIRE -16 320 -16 304 WIRE 80 320 80 256 WIRE -256 416 -256 256 WIRE -16 416 -16 400 WIRE -16 416 -256 416 WIRE 48 416 -16 416 WIRE 80 416 80 384 WIRE 80 416 48 416 WIRE 48 432 48 416 FLAG 48 432 0 SYMBOL pnp 16 192 M180 WINDOW 0 49 26 Left 0 WINDOW 3 38 52 Left 0 SYMATTR InstName Q2 SYMATTR Value 2N3906 SYMBOL res 0 416 R180 WINDOW 0 36 76 Left 0 WINDOW 3 36 40 Left 0 SYMATTR InstName R1 SYMATTR Value 50k SYMBOL res 64 -16 R0 SYMATTR InstName R5 SYMATTR Value 1.5k SYMBOL voltage -256 160 R0 WINDOW 123 0 0 Left 0 WINDOW 39 0 0 Left 0 SYMATTR InstName V2 SYMATTR Value 3 SYMBOL npn 48 208 M0 SYMATTR InstName Q3 SYMATTR Value 2N3904 SYMBOL LED 64 320 R0 WINDOW 3 23 71 Left 0 SYMATTR InstName D1 SYMATTR Value QTLP690C SYMATTR Description Diode SYMATTR Type diode SYMBOL res -32 16 R0 SYMATTR InstName R2 SYMATTR Value 40k SYMBOL res 144 96 R0 SYMATTR InstName R3 SYMATTR Value 4.7Meg TEXT -224 440 Left 0 !.tran 1m startup