LT108 Oscillation sounds like tube rush/white noise

[Hearinspace]

No reply expected or requested, I'm posting this because I couldn't find anything under the search terms I was trying when looking for info on a noise problem. Maybe this will be useful to somebody later.

A few days ago I wired up a VT-25A/10Y Preamp breadboard and though it sounded good it had some noise problems. After solving some buzz (grounding) and bias (mistake in circuit) problems, there was a much larger than usual tube rush /downpour on an otherwise quiet day sort of white noise. It was tolerable when there was music playing but too obvious between tracks.
I had wired the LT108 filament supply regulators the same as I had done on power amps I last built last year that also have the VT-25A's fil supply with a 10uF cap at the input of the LT108 and .47uF on the out. They are extremely quiet.

Not sure why the different result - maybe a weakness in the wiring layout on the new one tipped the scale . . . Just the same, today I put a 150uF cap across the output as per the LT data sheet (120uF) and now there is no white noise. Must have been hf oscillation.

I hope this is useful for somebody.
Thanks

[Arnulf]

I can't find any reference to LT108 on Linear's webpage so I'm assuming this is a low-drop voltage regulator (LDO) you're talking about (because people seem to be extremely fond of these here).

LDOs are far more prone to instability than the regular types so their datasheet requirements must be obeyed in full. If datasheet says 120 uF, then 120 uF it should have been all along. One can save a great deal of frustration by merely following manufacturer's suggestions, all neatly laid out in a simple schematic or two in the datasheet (ditto for capacitors close to non-LDO voltage regulator pins ...)

[kevinkr]

May also be a symptom of oscillation in one or more of the stages.. Grid stopper resistors right at the grid connection on the socket are advisable if not present..

[Hearinspace]

Originally Posted by Arnulf

I can't find any reference to LT108 on Linear's webpage so I'm assuming this is a low-drop voltage regulator (LDO) you're talking about (because people seem to be extremely fond of these here).

LT108 was a reference to the family of Linear regulators LT1083,1084,1085.

The value of 120uF was a typo - sorry. It was supposed to be 150uF as I actually put in. . . . . .

I was taught that functions of the pins don't really change whether the chip is being used as a voltage reg. or current reg. - either way it is always trying to regulate its own internal voltage.

However, using the capacitor combinations used on the sheet make it noisier as a current reg.in this circuit. In fact one of the quietest and best sounding circuits have a 10uF on the input and 0.47uF on the adjust with nothing on the output. (I am using the Linear designations for the pins here) My original reason for putting the cap on the adjust is that it is functionally the output in relation to the rest of the circuit and so I thought I'd give it a try. While it is quieter with 150uF in the same position a harshness starts to develop that is quickly tiring when listening to large orchestras and chorales.
I did try following the data sheet "in full" and it makes it worse.

I'd be happy to learn of better ways to use this chip. I spent a lot of time searching and though I found many references to it I didn't find any practical info on its implementation. If anybody can give me some useful guidelines for figuring how to set it up as a current reg. I'm all ears.



LDOs are far more prone to instability than the regular types so their datasheet requirements must be obeyed in full. If datasheet says 120 uF, then 120 uF it should have been all along. One can save a great deal of frustration by merely following manufacturer's suggestions, all neatly laid out in a simple schematic or two in the datasheet (ditto for capacitors close to non-LDO voltage regulator pins ...)

Originally Posted by kevinkr

May also be a symptom of oscillation in one or more of the stages.. Grid stopper resistors right at the grid connection on the socket are advisable if not present..


Thanks Kevin. I do have <100R stoppers on the grids of the 10Ys I'm trying as single stage line amps now. I''ll try some higher values to see if it nets anything.

[Arnulf]

It is my experience that normal regulators (of the non-LDO kind) are way behaved way better. If you have enough headroom (you'll need 1.25V + 2.5V or thereabouts) do try the venerable LM317.

[Hearinspace]

Arnulf thanks. Trying out various things here, not necessarily happy with the results yet but I'm really impressed at the influence the filament supply has on sound. I'll get a couple of LM317s next week.