Discrete JFET regulator problem

[lineup]

Hello.

I have made this 3 Transistor regulator.
One JFET ( BF245A ) controls voltage.

The problem is I do not know how to get it stay at 15.000 Volt.

I post some image to shows this dilemma.
This is the circuit I use.
One normal 3 pin reg.

[lineup]

Here is result when no load.
It is perfect 15.000 Volt !!
At 2 x 9 Volt input.
( It takes here 0.98 mA = Reg current consumption )

[lineup]

Here is my problem
I do not know howto fix this.

The voltage sinks, when I add a Load of 10 mA.
That 1k5 resistor.

Any ideas ???

Thanks, Lineup

[wakibaki]

I think the problem is that without the load the jfet current source sees 3300 ohms but with the load it sees 3300||(27296+1500).

This means that the reference point of the circuit moves.

I don't think there's an easy way to improve the circuit without using a reference such as a zener to compare with your output tap.

The jfet current source depends on a parameter which is not reliable i.e. the output voltage will be different with a different jfet.

w

[Christer]

There will be even more problems in practice, unless you hand-pick the JFET. Better use a BJT, where Vbe doesn't vary too much between individuals of the same type, and then add a zener, LED or similar on the "wire" between the base (which is now a gate) and ground. The voltage divider may have to be adjusted appropriately. This is still simple, but should give much more consistent behaviour. Load regulation might still be too bad, depending on your requirements. Even better would be to run the LED separately with a CCS instead of using the collector current of Q2.

(Hope I got it right, I should really have gone to bed by now).

[Nordic]

I think diffirent loads will always affect the rails to some extent as the voltage source is never ideal... so, bear in mind that fluctuating the load does the same to the source voltage... (to some extent... the base of U2 is connected directly to the DC supply without any regulation... so any changes in supply voltage affects output of U2 which in turn affects the fet...

I suppose it is a very simplistic way of seeing things, but lacking education, it is sometimes easy to see the trees from the woods...or the dirty river (R1 and R2) running through it...

[sawreyrw]

lineup,

All voltage regulators require some type of voltage or current reference device that does not change with input voltage or load current. The is none in your circuit.

Rick

[h_a]

Hi Lineup,

you definitely need a voltage reference. Currently you have an output impedance of about 1R

You can use a simple source follower with a vertical mosfet, this simple configuration has an output impedance of only 0.1R.

Have fun, Hannes

[lineup]

suppose so, Hannes
I am playing lot with my Simulator discrete Circuits. It is real fun & grosses spass.

In real life, I would use TL431, most probably. Using a small filter at the output of TL431 or TL1431 gives close to perfect VREF. And adjustable Voltage, too.

TS 431 ST Microelectronics is one variant.
It is limited to 1.24 - 6 Volt Reference and Precision is high!

TS431 Datasheet: Low voltage adjustable shunt reference
http://www.st.com/stonline/products/...re/ds/5558.pdf
TLV431 Datasheet:
http://focus.ti.com/lit/ds/symlink/tlv431b.pdf