If bandwidth was not an object of interest I would (and have) used an LM317 or LM317L as a good high impedance and stable current reference.

An LM334 also looks like a good, stable choice



It is hard to match new (or 25 year old) integrated circuits against discreet designs, though I have to agree that discreet is much more fun and interesting.

 

Again, I'm looking for the µa region, Not sure where I'm going wrong, I can't seem to get the LM4041 to work as expected, even with no load. <sigh> The LM334 it is then. I use the LM317 as a current regulator all the time. I'll be testing the LM1117 next (on a separate project).

 

OK, the result from the chip were so wonky I wondered if I had the right parts. So I broke out the USB microscope and took pictures. Do these look like they could be LM4041?

 

Until I get more LM4041 ICs I guess I'm done with this thread.

 

So, just for the heck of it I retried this schematic from post 15. It didn't register a current(.1µa resolution) until it saw 8.4V. Oh well , it will be the LM134 until I can get more LM4041s.

​

 

OK, next experiment:

​

I think our more experienced folk can see where I'm going with this

 

Results:

Minor modifications to the layout shown. After thinking about it the minimum voltage for a NE555 power is 4.5V this means the lower voltage range is 1.5V (1.5V minimum for 12V power it is 3V ( 3V minimum). This circuit did not meet the 1.5 Spec, but if I really cared I'd use Schottky Diodes for the bridge.

​

Measured Data

Input Voltage	Icc in µa	..........................................................................................................................................................................................
1.5	49.9
2.0	452.3
2.5	453.9
3.0	454.3
4.0	455

 

@DickCappels pointed out the LM134/LM234/LM334 3-Terminal Adjustable Current Sources that looks very promising.


I will order several of them ASAP. Clicking on the link will automatically download the .pdf datasheet. I can't figure out how to fix the blank screen the comes from clicking it but it works.

 

Doh! The brain damage really, really screwed with my memory.

 

So, just for the heck of it I retried this schematic from post 15. It didn't register a current(.1µa resolution) until it saw 8.4V.

Don't understand that.
My LTspice simulation below shows it entering the (sort of) constant-current region at about 2.5V.

Double check your circuit.

 

Then there is the LT3042. https://www.analog.com/media/en/technical-documentation/data-sheets/3042fb.pdf

 

i basically dropped this thread because i was discouraged. Now that I have finished the siren circuit I'm coming back to it, given my short attention span it may take a while to get back into the groove.

 

So I tried this experiment:

​

I decided to use schottky diodes because of their low drop. I got the following results:

Input Voltage	Current
2.5V	5.8µa	.............................................................................................................
2.0V	5.5µa
2.5V	5.8µa
3.0V	6.0µa
3.5V	6.7µa
4.0V	6.6µa
4.5V	6.7µa
5.0 V	7.0µa
5.5V	7.3µa
6.0V	7.5µa
6.5V	7.8µa
7.0 V	8.0µa

I am less than impressed by these results, anyone have any other ideas? Steam maximum and minimum voltages on the final product should be around 3 volts and six volts respectively.

 

Most of the graphs on the datasheet seem to stop at 10μA, suggesting to me that they really don't want you to know how well it works <10μA. Does it work any better at higher currents?

 

Something else that might be going on here:
This is the reverse leakage current for a 1N5818 schottky diode. Expect at least a couple of microamps of voltage-dependent leakage current even at low voltage.
I don't know what type of Schottky you used, but a BAT42 would get the leakage down by an order of magnitude, and a good old 1N4148 would be <10nA, but voltage drop might become a problem.

 

See

 

Something else that might be going on here:
This is the reverse leakage current for a 1N5818 schottky diode. Expect at least a couple of microamps of voltage-dependent leakage current even at low voltage.
I don't know what type of Schottky you used, but a BAT42 would get the leakage down by an order of magnitude, and a good old 1N4148 would be <10nA, but voltage drop might become a problem.

Truth to tell, he had not even occurred to me that the diodes might be leaky. I'll run a test on the LM334 without the diodes, see if it makes a difference. Thank you.

 

It hadn't previously crossed my mind that it might become a problem, mainly because, in the dim and distant past when this thread started, the premise was

So now I need to create a CC regulator that can go down to 100 µa

but now we're thinking about getting down to 5μA

 

That was just arbitrary number of course one of the points of this experiment was to see where the limits were.

 

@Ian0
you called it. I tried this layout:

​

You called it . It was a rock steady 4.6µA down to 0.88V. So I guess 1N4454's are in my future. Since the final design will be 3 volts to 6 volts( 12V PS on a 555). It simply never occurred to me that back biased diodes might leak.