AoE3 query re current sources

It probably got omitted because it's horrible. Essentially, your circuit se ts up a Vbe across the 110R resistor,and Vbe is ill-defined, and decreases by about 2mV per degree as the junction temperature rises.

It does have the advantage of using only 350uA to set up the 6mA current su nk.

Current mirrors can do quite a bit better. Australia can't get AoE3 yet, bu t AoE2 does include the Wilson current mirror, which can do much better.

If Q2 and Q3 were a matched pair of transistors, you could have done rather better by putting a 220R resistor between the emitter of Q3 and the -12V r ail, taking the emitter of Q4 directly to the -12V rail ,and tying both the bases and the collector of Q3 together.

This all rather depends on Q3 and Q4 having a current gain of at least seve ral hundred from 350uA to 6mA, but such transistors aren't hard to find as packaged duals

Arrays like the 3046, 3096, 3126 tend to be less good on current gain.

There are parts that are better, but they cost a bit

4-300/dp/1354175

--
Bill Sloman, Sydney

Thanks Bill, yes I did say it was quick and dirty but I think it rivals AoE3 fig 3.26A (that is two forward biased diodes setting base vref) which has the same tempco as my horror but a "floppier" current regulation.

piglet

Ring-of-two, that's a cute name, descriptive.

No, I think it's a fine circuit for some jobs, and I use it all the time. Somehow it got overlooked. For example it's an ideal current pulldown for an emitter follower. In this I disagree with Bill, "horrible" or "just fine," is a matter of perspectives and definitions. While one in general doesn't know the exact value of Vbe, and hence the exact current in the circuit, often it doesn't matter. Its tempco isn't all that bad, the current-determining BJT runs at a low power dissipation.

One nice feature, as you said, due to its internal feedback, it has a high output impedance; its current doesn't change much with output voltage. That's unlike ordinary mirrors and more like a Wilson mirror, which costs another BJT.

Hmm, maybe I can slip it into the Fig in a later printing.

--
 Thanks, 
    - Win

Thanks Win, I use it a lot too and was getting worried that somehow experts reject it for some reason beyond my knowledge.

piglet

This is fun:

TC may well be dominated, or at least confused, by the emitter resistor.

And it glows in the dark!

--

John Larkin         Highland Technology, Inc 
picosecond timing   laser drivers and controllers 

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

I use those a fair amount--they work great if you pick the right LED, and you get a free pilot light.

A LED driving an emitter follower makes a nice quiet 1V reference, too. (Not world-class for stability, but a good 20 dB quieter than your average bandgap.)

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 

hobbs at electrooptical dot net 
http://electrooptical.net

c_bb.JPG

c_data.JPG

80-90 ppm/C with a red or orange LED reference isn't bad, the ring-of-two BJT reference gives 2mV/650mV = 3300 ppm/C.

--
 Thanks, 
    - Win

The local feedback is a huge win compared to an ordinary current mirror, though. The output impedance isn't limited by the Early effect of the pass transistor, and it holds up well with frequency. For stuff like biasing emitter followers, it's pretty good.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 

hobbs at electrooptical dot net 
http://electrooptical.net

A patient person (certainly not me!) could use better resistors and then optimize the LED current, and maybe get better TC.

A resistor from emitter to ground gets interesting, too. One could probably null out the power supply sensitivity.

Somebody should make a decent resistor-programmed IC current source/limiter.

--

John Larkin         Highland Technology, Inc 
picosecond timing   laser drivers and controllers 

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

ED_Isr

ED_Isr

LM134, LT3092 ?

-Lasse

[snip]

TI used to have a line of two-terminal current sources, but they seem to have disappeared.

I seem to recall that National had a programmable one, but it was PTAT :-(

It would seem that a restructured TL431 architecture could do the resistor programmed source.

Is there a market? ...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    | 
| San Tan Valley, AZ 85142     Skype: skypeanalog  |             | 
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  | 
| E-mail Icon at http://www.analog-innovations.com |    1962     | 
              
I love to cook with wine.     Sometimes I even put it in the food.

The 3092 is fair, but it is fabulously expensive. And it's pretty slow.

LM134 is PTAT, so not very accurate.

--

John Larkin         Highland Technology, Inc 
picosecond timing   laser drivers and controllers 

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

I used something very similar, though with a FET for the pass transistor, to drive green LEDs from a LiIon battery, a few years back.

That's extra fun in a high-gain photodiode amplifier.

--

John Larkin         Highland Technology, Inc 
picosecond timing   laser drivers and controllers 

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

It's a feature--if you haven't got enough photons of your own, the TIA gives you extra. ;)

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 

hobbs at electrooptical dot net 
http://electrooptical.net

The pilot light aspect is nice when (NPN version of) the CCS is used at the emitters of a differential pair: the light goes off when the diff traansistor bases don't receive their bias. We see at a glance if the galvanic connection to the cryostat is OK.

Regards, Mikko

Well, any jFET with a source resistor and grounded gate qualifies. Also, the "source resistor = 0" variant of that circuit was a simple package change, and there used to be current limit diodes based on that.

One would think ALD could program a MOSFET to be a good current diode:

An LM317 with load resistor makes a regulated current (between input and common terminals, 'output' pin only connected to the resistor), if you can handle 4V of headroom and a 4 mA minimum current. This circuit is a good way to check out a three-terminal regulator on a curve tracer, by the way, if you have a PolyPaks sack to sort through.

Jfet current sources might vary 2:1 or maybe 10:1.

People still sell current limiter diodes, which are actually binned jfets. They cost, like, $4.50 each.

Seems like they could, although tempco might be poor.

A depletion mosfet and a source resistor is a mediocre current source.

--

John Larkin         Highland Technology, Inc 
picosecond timing   laser drivers and controllers 

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

Depends on your criteria. See our Figures 9.36 and 9.41. An LND150 is good to about 0.1% to 50V, and 0.5% to 150V at 100uA. And we have more convincing evidence in the x-Chapters, nicely using them in series to 2.5kV. Very useful in HV amplifier designs.

But oh dear, as it happens, I'm beginning to replace my decades-long fav LND150 with Infineon's sot-23 BSS126.

--
 Thanks, 
    - Win