1 mA Current Source

I'm using LM334's for the .001 amp current sources I need, and they work fine with the tempco circuit on the datasheet. What I'd like even better is a _dual_ 1 mA source in a dip. A huge bonus would be tempco built into the chip. Anything out there, with or without built in tempco? My rummaging around in Digikey hasn't found one. I think there's a nice dual .5 mA from TI, but I can't change the 1 mA spec.

Ed

Reply to
ehsjr
Loading thread data ...

When I need something that light duty, I would simply use a dual comparator or Op-amp to construct a dual current source. You just need the Sense R on the low side of the load which also connects to the (-) in of the op-amp. the sense r then connects to the common.. (+) input is normally held at some reference voltage..

Those come in a variety of 8 PDIP.

Jamie

Reply to
Jamie

You haven't given any specs, but this might do for ordinary requirements:

Vcc --- | | R1 | | | | Q1 | |Q2 |Q3 | |/ |/ |/ +---------|-----|--//---| | |>. |>. |>. | U1 | | | .---' LMV431 | | | ^ | | | / \

Reply to
dagmargoodboat

2 in parallel?
Reply to
David Eather

LM13700 is a dual amplifier with current-source outputs; you can strap its inputs to saturation and get your two current sources that way. It needs a regulated (+) supply and program resistor to set the current.

Reply to
whit3rd

Can't the TI dual be used at other currents, especially in the mirror configuration (if i remember correctly)?

Reply to
Robert Baer

Configurable for several different currents as you indicated, but not above 400 uA. You made me look at the REF200 datasheet. :-)

So far, it looks like there's not a dual 1 mA dual source chip. The 334 works well - I was just hoping for something a bit "nicer": one dip to replace 2 334's & their resistors and diodes for the tempco.

Ed

Reply to
ehsjr

Well, the best i can think of is to use the (damn! obsoleted with no replacement) LM194/294/394 super beta matched pair in a DIY mirror configuration; the currents would track very well with no need for temp comp. To be dirtier, the LM3046 may be usable especially if trimmed emitter resistors were used...

Reply to
Robert Baer

Analog devices make a few types of matched pairs - they would do the job as well. The LM394 is still available from some retailer - futerlec is quite cheap.

Reply to
David Eather

With a dual transistor my suggestion is stable, decently accurate, and simpler than what he's using now. DMMT3904W would do it. 10 cents if you buy a reel at Digikey.

-- Cheers, James Arthur

Reply to
dagmargoodboat

I must be missing something. I'm looking for a dual _source_, in a single dip, not a sink with 3 active SMT components. A real bonus would be a chip with built in tempco.

That said, I'd like to know the numbers you have in mind where you said "stable" and "decently accurate". The 334 datasheet circuit has a tempco spec of -77 nA/degC, and .02%/V current regulation, again per the datasheet.

Maybe the 431 & transistor, configured as a source, would do better?

Thanks, Ed

Reply to
ehsjr

if

People usually say "current source" even for sinks. The LM334 can do either, so I wasn't sure.

As to the "3 active components," an LMV431+dual transistor is the same number of parts as the LM339+diode you're using now so that didn't seem like a problem.

If you need 1% or looser the resistors are the limiting factor. Standard grades of the LMV431 are +/-1.5%, whereas your LM334 specs

+/-6%.

Stable depends on the grade of LMV431, emitter resistors, and delta(Vbe) of the transistor pair chosen. The feedback gobbles part of the change in base current over compliance voltage, so output impedance >1meg should be easy.

So, it's all good except I've drawn sinks and you need sources. Offhand I don't see an easy fix that's as accurate.

Ooops.

-- Cheers, James Arthur

Reply to
dagmargoodboat

Those are duals, but no matching spec worth spit.

Reply to
Robert Baer

if

Define "spit."

formatting link
DMMT3904W -- MATCHED NPN SMALL SIGNAL SURFACE MOUNT TRANSISTOR * 2% Matched Tolerance, hFE, VCE(SAT), VBE(SAT) * Vbe matched to 1mV.

-- Cheers, James Arthur

Reply to
dagmargoodboat

I believe in this context, it's closely akin to "not worth a continental."

Hope This Helps! Rich

Reply to
Rich Grise

If the V_CE's are close, that can work OK, but if not, you can be off pretty badly unless you have a big emitter resistor. The thermal resistance is ballpark 300-500 K/W, so a milliwatt of differential dissipation gets you a ~3% error.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal
 Click to see the full signature
Reply to
Phil Hobbs

That's initial accuracy pre trimming, and it's +/- 3 percent for the

334, not +/- 6. When you trim it to the current you want, it regulates current to .02%, per the datasheet. In circuit, I measured a variation of up to 7 uA, which is .07%, way higher than the .02%, but within my error budget. The variation I measured is in circuit with noise, load variation, power supply variation, temperature etc. - not at the test bench.

On the 431, again the spec is initial accuracy. It should do a lot better than +/- 1.5% when trimmed, but I haven't verified that at 1 mA in source configuiration, and I don't see a current regulation spec on the 431 datasheet like the one ( .02%) that appears on the 334 datasheet. Sigh. I'll need to build a test jig to eliminate as many variables as I can and compare a source configured 431 with 3904 vs a

334 with diode to satisfy my curiosity.

I _still_ would like a dual source chip!

Thanks, Ed

Reply to
ehsjr

Those are separate dice in the same package. I use them in one place but they're not all that great. I'd ditch them if I had that board to layout again.

Reply to
krw

s if

Sorry, you didn't mention that you were trimming it, or needed any particular high order of accuracy. I think of the LM334 as "loose"; non-critical. The datasheet (I'm looking at it) says "Set current error: 6%." I infer they mean +/- 6%, not 0 current to .02%, per the datasheet. =A0In circuit, I measured a variation

Indeed. It could easily be done with a reference, a dual op amp, and a few parts, but that's more parts.

-- Cheers, James Arthur

Reply to
dagmargoodboat

d

ts if

It's not quite that bad here--there's about 600mV (give-or-take) across each sense resistor.

As drawn, and operating in a total vacuum as to the actual conditions and requirements, if we assume 1mA and |Vce1 - Vce2| =3D 10v, that's

10mW difference x 500K/W x 2.6mV/K =3D 13mV d(Vbe). 13mV / (600mV across the sense resistor) ~=3D2%.

That could be reduced greatly by using a 2.5v version of the reference, e.g. TL431.

Yep.

But this wasn't meant to be super-duper, just a pedestrian stab at an unspecified reqm't, simple & cheap, like the LM334. Super-duper's different.

-- Cheers, James Arthur

Reply to
dagmargoodboat

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.