Constant Current Source Design

look for current Regulating diodes.

a couple of Fets and biasing resistors will do it which is really all the CRD are.

those there don't come in a .1ma size but will give you and idea of how stable they are.

if the range is not tight enough then i suggest using an OP-Amp as a comparator as the limiter circuit. basicly you couple the - and + inputs with a resistor which feeds also your output. when the inputs swing apart due to loading, the Op-Amp out put will bias a transistor that is supplying the source. something to look into.

100 V power supply seems impractical in my case. So instead of 100 uA, I could go for 10 uA or even 1 uA. This would reduce the voltage required to 10 V or 1 V.

only 1k, the >>transistors themselves will have to stand off that much voltage and still provide a >>reasonably close 100uA?

-- I couldnt figure out the circuit you are talking about !!! Could you please elaborate on this?

Certainly. Ohm's law tells you how much voltage is required to push a given current through a resistance.

The concept is to also have the current pass through a second resistor and a variable resistance device like a mosfet or junction transistor.

Some additional circuitry must subtract the voltage drop across the second resistor from a reference voltage and amplify that difference, using that result to control the variable resistance device so that it burns up all the extra voltage not needed to push the same current through the first resistor. If the amplifier has high gain, the current will vary very little as the load resistance changes. How fancy this circuit gets depends on how accurate and temperature stable the current must be and how fast the load resistance can change.

Some examples: See precision current source on page 7 of:

See bilateral current source on page 10 of: Note that the current is proportional to Vin (that you will have to provide).

You might also browse Google for [current source] or [constant current] or [current regulator]

--
John Popelish

Let's see....

At 100uA,

1k requires 1k*100u = 0.1V 1M requires 1M*100u = 100V

So, you require a voltage compliance of three decades and a voltage of 100V across your 1M load. And, when you don't want the 100V across the load, because the load is only 1k, the transistors themselves will have to stand off that much voltage and still provide a reasonably close 100uA?

Is that about it?

Jon

Here's a constant(ish) current source using a PNP transistor:

Current in load resistor RL = (Vcc * R1 / (R1+R2) - 0.7) / Re

"Confused Soul" schreef in bericht news: snipped-for-privacy@f14g2000cwb.googlegroups.com...

What do you really need? If you reduce to 0uA you need nothing at all.

petrus bitbyter

I guess there are two answers to your question. First, it's relatively easy to make a fairly stiff current source out of a voltage source with a series resistor. For the example you specified, though, you require a very high voltage (view in fixed font or M$ Notepad):

Stiff Current Source From Voltage Source ` + ___ ` o----|___|-------. ` 60 megohms | ` | ` o-----. ` | | ` R(L) .-. | ` 6000V 1K-1meg | |

"Confused Soul" skrev i melding news: snipped-for-privacy@c13g2000cwb.googlegroups.com...

Simple constant current device using tvo PNP transistors and two resistors:

___ .____/ | | \___| | .-. | | | | '-' o Constant current | | | | | | | |/ o-----------| | |>

| | | | \| | |-----------o