current mirror

This one is cute, too:

Version 4 SHEET 1 880 680 WIRE -128 -144 -320 -144 WIRE 80 -144 -128 -144 WIRE 80 -112 80 -144 WIRE -128 -80 -128 -144 WIRE 80 -16 80 -32 WIRE 80 -16 0 -16 WIRE -320 16 -320 -144 WIRE 80 16 80 -16 WIRE -128 32 -128 0 WIRE 48 32 -128 32 WIRE 208 48 112 48 WIRE 0 64 0 -16 WIRE 48 64 0 64 WIRE 80 112 80 80 WIRE -320 144 -320 96 WIRE -128 144 -128 32 WIRE 208 144 208 48 WIRE -128 272 -128 224 WIRE 208 272 208 224 FLAG 80 112 0 FLAG -128 272 0 FLAG 208 272 0 FLAG -320 144 0 SYMBOL res 64 -128 R0 WINDOW 0 54 42 Left 2 WINDOW 3 62 73 Left 2 SYMATTR InstName R1 SYMATTR Value 1 SYMBOL res -144 -96 R0 WINDOW 0 -61 22 Left 2 WINDOW 3 -60 56 Left 2 SYMATTR InstName R2 SYMATTR Value 1K SYMBOL Opamps\\UniversalOpamp2 80 48 R0 WINDOW 0 27 -39 Left 2 SYMATTR InstName U1 SYMBOL current -128 144 R0 WINDOW 0 -104 48 Left 2 WINDOW 3 -338 99 Left 2 WINDOW 123 0 0 Left 2 WINDOW 39 0 0 Left 2 SYMATTR InstName I_IN SYMATTR Value PULSE(1u 10u 10u 1u 1u 1m) SYMBOL res 192 128 R0 WINDOW 0 67 45 Left 2 WINDOW 3 86 79 Left 2 SYMATTR InstName R_LOAD SYMATTR Value 200 SYMBOL voltage -320 0 R0 WINDOW 0 -91 35 Left 2 WINDOW 3 -91 71 Left 2 SYMATTR InstName V1 SYMATTR Value 15 TEXT 296 -32 Left 2 !.tran 5m TEXT 216 -120 Left 2 ;1000:1 CURRENT MIRROR TEXT 264 -80 Left 2 ;JL Dec 28, 2015

You're right, it looks obvious but no, I've never seen it before either.

Useful.

Cheers, James Arthur

How about this one? It's not a mirror exactly (polarity is reversed), more like a copier/booster. It's also obvious when you look at it right.

Version 4 SHEET 1 880 680 WIRE 80 -48 -128 -48 WIRE -128 16 -128 -48 WIRE 80 16 80 -48 WIRE 48 32 -32 32 WIRE 192 48 112 48 WIRE 336 48 192 48 WIRE 48 64 16 64 WIRE 80 112 80 80 WIRE -128 144 -128 96 WIRE 336 144 336 48 WIRE -32 160 -32 32 WIRE 192 160 192 48 WIRE 192 160 -32 160 WIRE 16 272 16 64 WIRE 144 272 16 272 WIRE 336 272 336 224 WIRE 336 272 224 272 WIRE 16 352 16 272 WIRE 336 368 336 272 WIRE 16 496 16 432 WIRE 336 496 336 448 FLAG 80 112 0 FLAG 16 496 0 FLAG 336 496 0 FLAG -128 144 0 SYMBOL res 320 128 R0 WINDOW 0 56 43 Left 2 WINDOW 3 63 78 Left 2 SYMATTR InstName R1 SYMATTR Value 1 SYMBOL res 240 256 R90 WINDOW 0 74 57 VBottom 2 WINDOW 3 81 57 VTop 2 SYMATTR InstName R2 SYMATTR Value 1K SYMBOL Opamps\\UniversalOpamp2 80 48 R0 WINDOW 0 27 -39 Left 2 SYMATTR InstName U1 SYMBOL current 16 432 R180 WINDOW 0 58 24 Left 2 WINDOW 3 28 -23 Left 2 WINDOW 123 0 0 Left 2 WINDOW 39 0 0 Left 2 SYMATTR InstName I_IN SYMATTR Value PULSE(1u 10u 10u 1u 1u 1m) SYMBOL res 320 352 R0 WINDOW 0 49 42 Left 2 WINDOW 3 79 79 Left 2 SYMATTR InstName R_LOAD SYMATTR Value 10 SYMBOL voltage -128 0 R0 WINDOW 0 -91 35 Left 2 WINDOW 3 -91 71 Left 2 SYMATTR InstName V1 SYMATTR Value 15 TEXT -304 344 Left 2 !.tran 5m TEXT -400 232 Left 2 ;1000:1 CURRENT COPIER TEXT -344 280 Left 2 ;JL Dec 28, 2015

...

This one looks interesting - I wonder if the first one would have start-up problems. It would require the compliance of the input current generator to extend all the way to slightly below zero.

kevin

Worse, I need my output/load voltage to go below zero, so the input sink has to be based on some negative voltage.

I probably can't use it (the Dec 27 one) just now, but I thought it was cute and wondered if anyone has seen it before.

Put e.g. 1mH in series with R_Load and 100pF in parallel, and you'll know why :-)

-marcel

I have seen it before... I did that in the late '60's, in essence it's half a Howland "pump", see...

(Patent) 3,644,835 Phase Detector and Digital Phase-Locked Loop

on the home page of my website.

Larkin needs to research more carefully before he claims creativity

...Jim Thompson

Lots of current sources will oscillate into a goofy load. It's not very different from a Howland, dynamically.

Actually, these sources fell out of my current project, which needs a super-wideband cc source. The answer is the obvious, a fast PNP inside an opamp feedback loop, with a string of various inductors in the collector. Wideband, high-current, load-tolerant current sources are tricky.

These are nice,

lots of current and a fraction of the price of conical inductors. Same idea, a ferrite/epoxy filled coil.

No problem, just use a current mirror to the negative supply. ;)

Cheers

Phil Hobbs

Voltage sources are *so* much easier than current sources. To make them wideband, just keep adding caps to ground.

Potential for CM problems with the second one. The first one is just an unbuffered differential voltage to current converter where you develop an input voltage with your signal current.

Cool! Definitely does not depend on transistor matching or temperature flim-flam.

I liked the previous one so well, that i had to stand on my head to like this one as well.. ..and now i like it better. They are both rather nice; cannot say which is my CURRENT fave..

Except for ESR, and ESL, and resonances between poorly chosen capacitors.

Just as easy to make better current sources -- just add inductors, minding the conjugate things, EPR and EPC, and resonances between poorly chosen inductors. (Remember that Picotronics bias tee: they used carefully chosen inductors, plus some C and R+C helpers.)

It happens to be a little harder to get very high impedances (10k+?) out to very high frequencies, than to get very low impedances (1mohm?) out to... well, modest frequencies (~1MHz). Well, it's not harder at all, it's just a matter of demand: low impedances for modest frequency power supplies are most common, while high impedances for special purposes (and therefore serious demands on impedance and frequency) crop up infrequently but are that much more of a challenge.

Tim

(snip)

How do you know? Please clarify.