Noise floor of Tubelab SE

[wrenchone]

If you're running out of poop even with a 100V tap, it sounds like there's still not quite enough voltage at the transformer outputs for the headroom you need. The 50 Hz line frequency adds insult to injury by increasing the output ripple for a given load/output filter capacitance. Using Schottky rectifiers and higher filter capacitance are both steps in the right direction to get a little more voltage to work with. Adding a little boost to the primary in the manner I described might also give you what you need. If it were me, I'd bitch at the guy who wound your custom XFMR.

[Nikon1975]

I think the problem of running out of headroom for the filament regulators is solved. As you said schottky and extra cap did the job. You can see the big difference between the noise I posted few days ago, and the one from today, showing the 1kHz tone.

The thing I did not decide yet, is if I may replace the 100kOhm potentiometer with a lower one. I remember George saying that 100kOhm was too much for the 5842.

Thanks,

Davide

[wrenchone]

If you ever run into a similar problem, it's another way of skinning the cat...

[Nikon1975]

Thanks,

It's an interesting suggestion, but I am not sure I got it right. Do you have a connection diagram ?

D.

[Nikon1975]

Sorry, I got it.

D.

[wrenchone]

For those that didn't get it, here's a diagram. The buck/boost transformer has its primary hooked across the line voltage. Its secondary is placed in series with the primary of the transformer you want to trim. Depending on the connection of the buck/boost transformer, you can either buck or boost the primary voltage of the transformer you're trying to trim. The secondary current rating of the buck boost transformer must be at least capable of handling the primary current of the transformer you're trying to trim, meaning that in most situations, a modest filament transformer will work just fine.

[Nikon1975]

Basically you are using it like autotransformer, so that the main transformer sees higher or lower voltage.

D.

[wrenchone]

Something like that - its a common technique.

[tubelab.com]

The thing I did not decide yet, is if I may replace the 100kOhm potentiometer with a lower one. I remember George saying that 100kOhm was too much for the 5842.

Sorry, I spent most of the last 3 days in a car driving 1200 miles, so I was away from this forum.

The resistance of the pot combined with the Miller capacitance of the 5842 can limit the high frequency response at mid volume levels. If you listen with the volume pot near minimum or near maximum with the typical low output impedance CD player, you will never see an issue.

Play some music with a lot of high frequency information and rotate the volume pot. See if you notice a loss of high frequencies near mid volume. If so reducing the value of the pot will fix the problem.

[Nikon1975]

No problem George, no need to apologize.

I did some homework to improve my understanding:

I estimated the Miller capacitance as Cm=Cgk+Cgp*(A+1)

I ended up with a value of 120 pF for the 5842.

I then simulate the input of the tube lab SE to see what you mean. These are the results.

This is in the worst condition. I see what you say. Maybe 50k is a better choice, without going to 10k, as I have one source with small (1-5uF) output caps.

I hope I will have again the 300B running in few days.

Thanks,

Davide