Best low noise regulator?

[syn08]

This is how I'm wiring this regulator. If the regulator is in the same case with the phono stage you may delete 3 and the shielded cable T1, and simply star the output/sense and the grounds/sense at the low noise input stage terminals.

I guarantee that using this schema will improve the performance. Can't say about the sound, but if there's any impact it will definitely be larger than the effect of an overnight burn-in. Also, if the MOSFET you are using hass Ciss lower than the NTP in my schematic, you may want to proportionally increas the gate resistor, to keep the pole unchanged.

[Uchi Deshi]

Put resistor between inverting input and voltage ref source to isolate it from cap.

[AndrewT]

Syn,
why have you duplicated the route 3 and route 4 to the load sense point?

This is one of the very rare occasions where the designer has properly drawn the schematic to show how to implement connections/traces that ensure the design works as intended.
I do wish all designers would do this.

Uchi's extra resistor should be 125r to match impedances at the two opamp inputs.
But, this value only applies if the effective impedance of the LED ref is zero.
Any idea what the LED ref impedance is?

[syn08]

Originally posted by AndrewT
Syn,
why have you duplicated the route 3 and route 4 to the load sense point?

This is one of the very rare occasions where the designer has properly drawn the schematic to show how to implement connections/traces that ensure the design works as intended.
I do wish all designers would do this.

Uchi's extra resistor should be 125r to match impedances at the two opamp inputs.
But, this value only applies if the effective impedance of the LED ref is zero.
Any idea what the LED ref impedance is?

Because you don't want any DC trough the shielding. 4-4 should have a lower impedance, so current goes through there. Another idea is to leave the right 3 open, but I usually prefer to connect both ends.

If possible, I would avoid the 125 ohm resistor. It adds noise to the output.

You are correct, e.g. without proper wiring something like the PGP amp would never be possible. See the wiring schematic on my web site http://www.synaesthesia.ca/Wiring.html

Edit: As it is, the output voltage is about 30V and of course, the raw power supply has to provide at least about 35V. You may want to adjust the R3/R4 divider for other values.

[martin clark]

Any idea what the LED ref impedance is?

When I've used LEDs for voltage reference purposes (at 1-3mA) I've seen impedances c.35-50ohms. Pretty low actually, and significantly better than a zener at these currents

[analog_sa]

hi syn08

After examining your circuit again i have to admit i don't quite understand it and will appreciate some hints.

1. Why bypassing R5 important for the high frequency performance? Is R5 not there simply to supply current for the voltage source? And presumably can be replaced with a ccs for better performance?

2. Why is the connection from R5 taken to the sense point? I understand the connection to R4 but again, is R5 not simply supplying stable current to the led?

3. I would probably not ask 1 and 2 if i could get the circuit to work as shown, but i can't. As soon as there is a cap above 100n across R5 i get low-frequency oscillations seemingly determined by the LED impedance and the cap.


I don't know how important any of this is. Fed from a ccs instead of a resistor this shunt sounds quite spectacular.

[syn08]

Originally posted by analog_sa
hi syn08

After examining your circuit again i have to admit i don't quite understand it and will appreciate some hints.

1. Why bypassing R5 important for the high frequency performance? Is R5 not there simply to supply current for the voltage source? And presumably can be replaced with a ccs for better performance?

2. Why is the connection from R5 taken to the sense point? I understand the connection to R4 but again, is R5 not simply supplying stable current to the led?

3. I would probably not ask 1 and 2 if i could get the circuit to work as shown, but i can't. As soon as there is a cap above 100n across R5 i get low-frequency oscillations seemingly determined by the LED impedance and the cap.


I don't know how important any of this is. Fed from a ccs instead of a resistor this shunt sounds quite spectacular.

1. If you think in terms of sound, I have no explanation. Otherwise, the output impedance and the line regulation are both much, much better with R5 bypassed. Think of the caps as being included in the loop and the opamp acting as a capacitor multiplier. See the next two messages for results.

2. For the same reason as above. See also the Jung regulator articles, they use the same schema for connecting the reference current source.

3. I'm pretty sure you have grounding issues. From my experience, it is also unlikely you'll ever have good performance for this regulator built on a breadboard. Try point to point with short connections.

[syn08]

Line regulation, LED bypassed

[syn08]

Line regulation, R5 bypassed. Which one would you take?

Vertical scale is in dB.

[analog_sa]

Line regulation is something i largely don't care about. Plus i have preregulators. Do you have similar pictures for noise?