Ultra low noise JFET voltage regulator

[Kochkurov Maxim]

This regulator provides noise lower than 0.5 uV (0.1-10Hz) and may uses to supply clock xtal oscillator, reclocking triggers or other digital devices. Some words about 2P903 (KP903 industrial version). It's russian military hi-frequency power JFET, which has very low noise(0.5nV/Hz^1/2). You may use any power JFET (max current >0.5A), but noise will be higher than in 2P903 case. Unfortunately, I don't know its equivalent, which has so low noise, maybe anyone knows?

[ukram]

http://focus.ti.com/docs/prod/folder.../reg103-a.html

How about the Burr Brown / TI REG103. It has 33 uV noise for 100 kHz. So it is about the same? (sqrt 100k/10 = 100)

[Fred Dieckmann]

The output fet is inside the feed back loop
of the OP-27. The input stage noise of the op amp will control the regulator output noise voltage in cases where the voltage reference is adequately filtered. Since the ADR40 voltage reference has a noise of 1.75 uV in the 0.1 to 10 Hz region and the RC values of the filter shown cause it not to filter noise until well above the audio band, I think your noise specs for the regulator are in question and are lower than the noise the regulator actually produces. With the filtering shown the voltage reference is the leading contributer to the noise in this circuit. The reference noise is also multiplied by the gain of the op amp.

In terms of transconductance and power dissapation A power mosfet would be a better choice than a jfet.

[x-pro]

Noise in this circuit will be defined by the reference DA3 and the op-amp DA4 . The JFET VT1 is inside the feedback loop and it's noise is not critical - the circuit most likely will work happily and produce about the same noise even with quite noisy MOSFET in place of VT1, i.e. IRF510, 520 or equivalent. The only nesessary change would be to replace VD1 with a 10V zener in an opposite direction (anode to the ground)

Cheers

x-pro

P.S. Fred did reply quicker than me . However the figure in the original post probably RMS, and in the spec of the reference is peak-to-peak, so no real error here.

[Kochkurov Maxim]

Yes, this is RMS. I measured noise of this regulator by selective millivoltmeter, which measure only RMS. The RMS value of ADR420 is 0.3-0.5 mv. All apologies, I had made error when I drew this schematics. R4 must be 1k, and C17 must be bypassed by 0.1mk X7R and OS-CON 10mkX10V.

In accordance with theory, feedback fully reduses noise of VT1, but it's not so good in real life. When I change 2P903 to IRF540, noise increases from 0.5 to 2uv.

[Elso Kwak]

Hi,
Unfortunately the picture did not show up anymore. But that can be repaired.
A 7815 has typical some 40µV/VO noise RMS. So that is 600µV for 15 V out.
Now I have no experience with shunt regulators but if that will be effective enough to get the low noise as claimed??.
Also I am not sure about the interaction of the 7815 and the opamp regulator. With a series regulator I got oscillations in some cases, depending on output caps size. My experience is that is is very difficult to get rid of the noise of the 7815, once it is in your system.

[Ryssen]

Unfortunately the picture did not show up anymore. But that can be repaired

Anyone have a Schema for this one?