Circuit designed to draw specific amounts of current?

[cupoftea]

Hi,
We have to make a circuit that draws between 25.4uA and 27.7uA from a source which could be anywhere between 22V and 38V.
So we did the attached (LTspice and .PNG attached)
Do you agree that this circuit will always draw between the stated values?
If not then what might the limits of the current draw actually be?
Would any variation outside of 24.4uA and 27.7uA be more to do with hfe variations, or to do with vbe variations?

 

[FvM]

Why you need ~40 components for current source? If they are implementing additional circuit features, please specify.

 

[KlausST]

Hi,

yes, I agree with FvM.
***
Circuit:
Obviously there are three paths drawing current:
* R35
* R22
* emitter of Q12

If it was my task to keep total current consumtion constant I´d rather monitor R36 voltage drop.

Klaus

 

[cupoftea]

If they are implementing additional circuit features, please specify.

Thanks, yes, actually stating the real purpose of this circuit would do no harm to anyone whatsoever, but i cannot do so unfortunately.
(I have actually removed some capacitors from the circuit, none of the removed capacitors were series capacitors...all were from shown nets to ground or to another net. It was pointless of me to remove them, as it still would do no harm to anyone, but just to be a "good" boy so to speak i removed them.)
If you look at the 5v6 zener, and then look at some nets which are a "bit away" from that, you can likely see what the simple circuit is about.
I bet if i said what the circuit actually does anybody would think it was stark staring madness!

Not that it wouldnt be a simple job for anyone to buy the product and just reverse engineer it....let alone that, get in touch with a Chinese manufacturer and ask them to make there own version of the circuit for half the price. They wouldnt need to reverse engineer it....they'd just do it......same as anyone else would.
Non disclosure paranoia is running at high levels everywhere now.....i even got banned from fixing a problem on a circuit once....in case i "stole" the design.....but then they relented and let me work on it...it was a bog standard linear regulator...and yes, it was easy to fix the problem.

Also, May i ask, Do you think that the above shown circuit may show oscillation in some of the BJTs?

It would actually be OK if the shown cct drew less than the stated current range, but its of interest whether the actual drawn current would be capable of varying to outside of the stated current range? (24.4uA and 27.7uA)

If it was my task to keep total current consumtion constant I´d rather monitor R36 voltage drop.

Thanks but sorry i should have said we are not at liberty to add any more circuitry. The fact that you say this indicates that you may suspect that the current could indeed stray above the stated range? (certain batchs of BJTs, temperature variation, etc etc)

The Q12 and Q20 based current clamps will obviously ship more current with BJTs with increased Vbe.

Also the 10MEG resistor R23 conducts just 2uA of current at 22V input....i am of the impression that you can get 2uA of current in the solder resist around the resistors...which would dramatically adjust the "effective value" of this resistor. Would it be worth not adding solder resist, but then coating the whole circuit with a good quality conformal coat that would not conduct any leakage current around the resistors?

Noting that pages 10 to 11 of the following show that parasitic leakage currents in the PCB can easily lead to more than 1uA of leakage current in the solder resist..

https://www.power.com/sites/default/files/documents/tinyswitch-4_family_datasheet.pdf

..this can be prevented by good SMD practice and handling but you never know when the assembly house is going to have an off day and accidentally use no-clean flux etc. Not to mention contaminants getting onto the PCB at the customer premises. What if the PCB assemblers have high humidity when they are laying down the solder resist on the PCB....?.....the solder resist will have low impedance and those "10MEG" resistors wont be 10MEG resistors at all in real terms?

Surely any design with 10MEG or 4MEG7 resistors in it must be subjected to severe cross-questioning? Not least due to potential oscillations in the BJTs.