Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Jim Williams]

Most of these older non-low dropout style regulators used a zener clamp diode with an opamp for the adjustable reference. That's where most of the internal hiss is generated.

The LT 1085/6 style low dropout regulators are much more precision and have lower output noise. I stil use a large el cap on their outputs to reduce that, 1000uf is a good starting point. Add a small film .01 uf to trap hf spikes and other noise.

Good heatsinking is equally important to reduce the noise. Cool semiconductors gererate less hiss than hot semiconductors.

Alternatively, one can add a transistor shunt after the regulator to further reduce ripple/noise. Feed the collector, run the base to ground with a cap and exit the emitter. Use a small 2k resistor from the collector to base. Observe polarity and don't let the smoke out.

[klaus]

Most of these older non-low dropout style regulators used a zener clamp diode with an opamp for the adjustable reference. That's where most of the internal hiss is generated.

So is this strictly the electron noise generated by the amount of (bad) components in a (bad) circuit implementation? Or is there switching noise involved?
I was always under the impression that the constant voltage adjustment to regulate the constantly changing incoming voltage generates the audible noise in these regulators. But I don't want to perpetuate that notion if it's wrong.

[Kai]

I was always under the impression that the constant voltage adjustment to regulate the constantly changing incoming voltage generates the audible noise in these regulators.

The key word is constantly. An analog, so called linear regulator does NOT wait for a change, then have a delay and then react.
Look at it as an analog amplifier (which in fact it is), and as every of such it regulates immediately, constantly, within its maximum speed range.
So there is no switching noise generated, because there is no switching or stepping going on. It greatly reduces the noise that is on the unregulated power supply side.
To get the best out of this a proper grounding scheme for the whole circuit is necessary. Regulation is in reference to its reference voltage input, this needs to be the center of the grounding starpoint.
A common mistake is to use a point at the unregulated side as grounding starpoint, where hum and noise currents are floating.
Further filtering (by caps) is best done close to the supplied circuit, to reduce supply impedance.

[Jim Williams]

In a good regulator design you will see around 1 mv of noise, that's a bit more than -60 db down, not bad. It's benign hiss, steadystate white noise. There are no switching artifacts as there is no switching. It shows more hiss than a quiet opamp does.

That hiss does not translate to the active element increasing it's noise as PSU filtering prevents that. With opamps you also get that PSRR ratio spec that nulls out that noise contribution another -80 db or more.

In as this is for a DC heater design, most of the audio concerns can be eliminated.

[Kai]

In a good regulator design you will see around 1 mv of noise, that's a bit more than -60 db down, not bad. ...
In as this is for a DC heater design, most of the audio concerns can be eliminated.

Some tube microphones use the DC heater for grid biasing too.
In this case additional filtering by a large cap is desirable.
Even if this is not the case, noise current introduced into the common ground connection might creep into the audio path.
A 4700uF / 16V cap doesn't cost a fortune.

[MikeCheck]

Update:
Built and installed Uwe's circuit. It's working great, and I was able to fit it in the original location of the Stabilyt cells.

[uwe ret]

Congratulations!
As I never considered this forum to be a DIY tutorial, I forgot to mention a recommendation to use a LM317-T in the TO220 package and mounting it to the metal chassis for proper heat dissipation.

As to the concerns about noise raised in previous posts to this thread:
a) the LM317 is an analog regulator with the reference far removed from the output path;
b) the filament is too sluggish to let its temperature follow mV noise variations in its supply;
c) to ameliorate any doubts, some data sheets suggest an additional capacitor of 1µF or larger
    placed between the 'Adjust' terminal and ground (I never found it to make any difference);
d) anyone undertaking any equipment modifications should be qualified enough to possess a
    basic understanding about proper circuit layout and grounding principles.

[MikeCheck]

Thanks Uwe for all your help and information!

I will change to the 220 package.

[uwe ret]

As the filament current for a single AC701(k) is only 100 mA you will probably be fine with your present arangement. If it is working satisfactorily without getting too hot to touch comfortably, there is no need to replace the regulator IC.

[Jim Williams]

Use an el cap to bypass the reference/adjust pin to ground on a LM317/337/LT1085, etc. It lowers the noise. I use 47 uf here, a Panasonic FR series tantalum replacement. Those have lower impedance specs than a tantalum cap.