Best low noise regulator?

[jan.didden]

I would think that using the current source in the place of R6 would immediately give a huge gain in loop gain and a huge gain in performance. The 10 ohms would then need to be like 60 ohms, of course.

Syn08, what do you think?

Jan Didden

[Salas]

Gimme your best upgrading ideas and I will surely try them. But I forewarn you...My top criteria is always that elusive anti engineering sanity, non cross checked thing, ''sounds better or not''.

[syn08]

Originally posted by janneman
I would think that using the current source in the place of R6 would immediately give a huge gain in loop gain and a huge gain in performance. The 10 ohms would then need to be like 60 ohms, of course.

Syn08, what do you think?

Jan Didden

Of course, another JFET instead of R6 will likely work much better. It will be at Vds = Vgs(on) Q4 which should be fine (it's a few volts). A nice simulation exercise, unfortunately I'm pretty tied right now... Loop stability should also be checked, unity CL gain with high OL gain things can get tricky...

[Salas]

Will try first chance I remake one! Thanks! Engineers are cool!

[iko]

Until then, this is how it sims (jfet replaced R6).

[iko]

And while we're on the subject of simulation, if we're to believe any of the fiction our simulators tell us, who would like levels of line regulation like the attached?

[jwb]

Originally posted by Bonsai
Looking at your last post, if I'm reading it correcly I'd say you have problems. There should be no overshoot or ringing.

First, let me thank you for the stimulating conversation.

I think the entire power system needs to be addressed when looking at these regulators. For example, in the project I am working on today, there is a Jung-type regulator, but there is also an array of 8 39µF OS-CON caps, 8 1µF ceramic caps, and an 800pF capacitance due to the ground planes of the PCB. Output impedance of the whole system is < 100mΩ out to 80MHz and < 1Ω out to 400MHz, which I think is acceptable considering the passband only extends to 96kHz.

Regarding your comment, I agree the ringing was excessive in the plot I posted. It turns out that the AD825 has far better transient behavior than the AD820 (in simulation). I use the AD820 for low voltage outputs since it can swing rail-to-rail and run from a single 5V supply, and I use the AD825 for outputs ≥ 9V.

As you can see from this plot, the transient load response is rather good. For a (quite improbable) 25mA load step with a 10µs rise time, the output swings only about 10µV. That's -68dB @ 100kHz.

[jackinnj]

Originally posted by jwb


Regarding your comment, I agree the ringing was excessive in the plot I posted. It turns out that the AD825 has far better transient behavior than the AD820 (in simulation). I use the AD820 for low voltage outputs since it can swing rail-to-rail and run from a single 5V supply, and I use the AD825 for outputs ≥ 9V.

I think that ADI posited 2 SPICE models for the AD825 depending upon the rail voltages.

[jwb]

True, but the models are for ±5 and ±15. Because of the 825's limited output swing, it's impractical to use it below 9V output. The 820, on the other hand, will swing down to the negative rail, so you can use it for a 5V regulator.

[Salas]

Originally posted by ikoflexer
Until then, this is how it sims (jfet replaced R6).

Looks nice to me. How much better in dB it got versus the resistor? How many mA did you feed to the BC550C?