Geoff Thornton wrote:



> Another way if you dont want to use all the gates and transistors is

> to use a 555 timer astable running at about 9kHz to drive the diode

> pump (ie the two diodes and capacitors shown on Will's circuit.



  The NE555 is commonly touted as the answer to every problem.  It is

quite cheap.  It is however, rarely the optimum answer.  For *this*

particular problem:



 1} The task is to drive a 10k bias potentiometer.  This (1.3mA) is the

total load.  The source is specified as 13.6V, presumably automotive

(thus, needs to be protected by a 22 ohm resistor and 15V 5W Zener or

equivalent).



 2) An ICL7660/ Si7660 is *not* rated for this voltage.  You need the

7661.  It is the minimum component count (three capacitors; input,

output and shuttle) and most compact solution but it is not a regulator.
No other suggestion is either.



 3) If you have a MAX232 (which you often do!) fed by 5V, you *have* a

-10V rail, (don't get confused; it is *not* -12V or -15V, but no matter

for this application), moderately regulated (depending on the RS-232

line load).  A MAX231 has in effect the 7661 included, mirroring the

supply voltage.



 4) The transistors on Will's circuit are unnecessary in this

application; you will most likely obtain a greater voltage without them

(at 1.3mA that is!).  I would suggest instead, a 40106 with one gate as

oscillator and the other five paralleled as the current driver.  A B-

series gate may be preferable.  This is probably the *cheapest* option.



 5) An NE-555 is a poor performer in this circuit, it's pull to the

negative rail is about 0.4V, while it's pull to the positive rail is

short by some 1.5V or so.  A 7555 is much better, but has little more

drive (not sure on this) than the 40106, costs a lot more and has the

same component count.  To assert the 7555 circuit is simpler is facile;
it has a smaller footprint.  It is more accurate but in this circuit
this is again irrelevant.



  You figure!

--

  Cheers,

        Paul B.