1 mA Current Source

The line you quote from Page 2 is an output current error spec, not a maximum. It says that at 500 uA Iabc, you'll get between 350 and 650 uA of output current when the input stage is fully switched.

Not terribly encouraging accuracy, but there it is.

The output noise graph on P. 6 shows data for Iabc = 1 mA.

Cheers

Phil Hobbs

That's a big improvement over the datasheet circuit, but there are still a number tempco's, Vbe's, Iq's, and other things to worry about.

This is the way I'd usually do that:

V+ -+- | | +-------------. | | __|__/ .-. / /\ LMV431 | | R1 -- 1.24v | | 1.24k | '-' | | | .-----------+ | | . | | | |\ U1 | | '-|-\ |--' | | >- | | R.load | | R2 .--->

'-' | | | | | =3D=3D=3D =3D=3D=3D

U1 has to have an input range that includes its positive rail. A darlington PNP can be used in place of Q1, with some small loss of accuracy. Adjust R1 to trim. A dual op-amp, an extra FET, and another resistor gets you a dual source.

-- Cheers, James Arthur

Communication is surprisingly hard in person, and even more so in print.

I daresay few people opposed to cutting the federal budget understand that last month our federal outgoes were 3.01x our federal income.

At least in electronics we can measure stuff /then/ argue ;-)

-- Cheers, James Arthur

You need to examine ALL THE LINE of the '650 uA' stated limit; it isn't the output limit at all, it's the upper error bound for a PARTICULAR VALUE of program current; that isn't the value of program current you'd use for a 1 mA=20 source, so the relevance of that number is nil.

Basically, it's their commitment to accuracy of the=20 device functional equation; if it's as good as this range, they won't call the device faulty and won't give you a refund on your purchase price.

Thanks. That's the typical Vref & op amp current source, and clearly a better performer than the transistor addition to the original. The fun was in converting with a transistor, to keep it closer to the original sink circuit posted, and to see how it performs.

That's what got me wondering if there was a dual 1 mA source chip available. All they had to do was put the fet, Vref & r2 inside the chip. 8 pins, 2 sources, each programmable with an external R. Why they (apparently) haven't done that, I don't know. IMO the thing would sell, even if you needed just the single source.

Ed

Thanks. That clears it up for me.

I see that, thanks. It means they tested the thing at 1 mA and it actually operates there, so it is in fact a 1 mA dual source, just as whit3rd stated.

Ed

Thanks. You and Phil have cleared this up for me. Indeed, it will provide a dual 1 mA source, just as you stated.

Ed

Hard to do, making a precision dual source dissipating that much power inside.

If you used a 2.5v reference with the ASCII circuit above, you'd get the i/o voltage requirements within range of many ordinary op-amps. That's probably the most practical answer.

-- Cheers, James Arthur

Yes :) Use LM358 or LM324, fairly easy, and the 0.5% '431 part (TL431B, FAN431L) might save you adding calibration trimmers.

Grant.