What are the pitfalls of the 1mA constant current source shown below?

Thanks.

 

1. R2 is way too large, as it will only allow about 12 μA through ZD1; while ZD1 will "kinda-sorta" function as a Zener diode at that current, it won't be very stable or predictable. I'd make R2 about 1.2 kΩ which will give you about a mA of bias current for ZD1.

2. Since Vbe of a transistor has a negative temperature coefficient of about -2 mV/°C, there will be a small positive temperature coefficient to the 1 mA output current (800 nA/°C, roughly).

Other than those two things, I can't see anything that stands out unless you're out to achieve very high precision.

 

It only gives a couple of volts of compliance (voltage swing) on the load.

 

What are the pitfalls of the 1mA constant current source shown below?

Thanks.
View attachment 97930

See attachment for an example of how much the voltage across a Zener changes with current through it. At that low a current you may only get about half the rated voltage. So about 0.5 mA out and only about 1 Volt (?).

 

You might consider a constant-current diode.

 

You might consider a constant-current diode.

Wow... I didn't even know those existed. Kind of expensive, though, at $2.28 a pop.

 

You might consider a constant-current diode.

Current regulator diodes are J-FET's with the gate tied to the source. Because of this the minimum voltage allowed across the "diode" is the gate-source threshold voltage of the FET. The 1ma diode in the link looks like it needs around 3 to 4 volts minimum.

 

Look at my Blog.

http://forum.allaboutcircuits.com/blog/low-temperature-drift-jfet-constant-current-circuit.579/

2N4416, page 4, second graph down on the right Id/Vds
That one runs a zero temp comp point at about 800 ua
It won't drift a lot if you fudge it to 1 ma or you might change your other circuit to run at 800 ua.

 

Wow... I didn't even know those existed. Kind of expensive, though, at $2.28 a pop.

And for that money you get a very inaccurate current. The 1ma diode in the link can vary between 0.88 and 1.32 ma!

 

What are the pitfalls of the 1mA constant current source shown below?

As previously noted, the zener current is too low for the zener to regulate voltage.

What are the requirements? Will supply voltage vary? Do you require temperature stability? Do you need exactly 1mA? How much headroom do you need? ...

 

Fair enough. Just thought I'd throw it out.

 

Fair enough. Just thought I'd throw it out.

Glad that you did. You just hit a sensitive spot with me.

I was frustrated when I wanted accuracy and compliance that current regulator diodes could not deliver. But that's how we learn.

 

But that's how we learn

Don't we all? ... and there are no shortcuts!

 

You can use a TL431 voltage reference to make a fairly low-cost and stable current sink as shown here (Figure 14).
Unfortunately there's no easy way to make it into a high-side current source to a grounded load.

 

See

 

Another approach you might consider is a simple current mirror, as in the attached. R1 sets the output current, while emitter ballast resistors R2 and R3 mitigate the effects of any Vbe mismatch between Q1 and Q2. Assuming the +4.5V supply doesn't vary, output current will be constant within a couple percent over a load voltage range of 0V up to within a couple of hundred millivolts of the supply voltage.

 

Lots of good suggestions. Thanks to all for the input.

 

Maybe not what you have in mind.

I played with several variations. This is one, bipolar, from the Walt Jung's book, with buffered Vsens.

Sure it must be simplified and could be to a high degree. The unconnected resistor is there to play with two ranges of input signal. Worked OK. Sensible to spikes in Vdd. Solved with good filtering.



/EDIT to add:

Load is R31 which for tests I made 4 Ohms.

EDIT/

 

If supply voltage will vary, you could combine the TL431 reference from @crutschow with the current mirror from @OBW0549; replacing the ballast resistors and 2 transistors with a dual transistor that would give better matching.