12b4 bass rolloff problem

[6L6]

Thank you for the schematics!

Originally Posted by artosalo

Using two FET or transistor CCS with 12B4 is waisting of resources.

Do you mean that a 2-fet is more complexity than the 12B4 needs, and a simple CCS is sufficient?

[artosalo]

Yes.
The anode impedance of 12B4 is so low that the dynamic impedance of a simple one-fet CCS is more than sufficient. Typically 500k to 2Meg.

[6L6]

Artosalo -

A few questions, in the CCS schematic that you show, does the potentiometer (variable resistor) need to handle any current? I.E., can it be a small trimmer pot?

And does this pot control the current, or is it a function of the pot in concert with the zener string?

Is there any reason why the zener can't be replaced with more LED's to make a 6.6v string?

Thank you,

Jim

[zigzagflux]

Originally Posted by artosalo

The anode impedance of 12B4 is so low that the dynamic impedance of a simple one-fet CCS is more than sufficient. Typically 500k to 2Meg.

Depends. I thought the same thing until measurement confirmed improvement in both power supply ripple reduction, as well as IMD. Before and after attached; pictures of after condition in two images.

[artosalo]

The anode current of the tube passes thru the potentiometer, but since the voltage drop is only some three volts the power dissipation is small and the pot can be small too.

IRF9630 is a low cost enhancement type PFET and it has to be set to "open" stage with the "bias" voltage generated with the zener + diode string. The anode current of the tube forms a control voltage drop over the potentiometer which keeps the current constant. The diode in series (1N4148) with the zener is for temperature compensation.

I do not know how well the LED string can stabilize the voltage over wide temperature range, but this zener + diode works well.
Anyhow zeners need much higer current to operate.

[artosalo]

Originally Posted by zigzagflux

Depends. I thought the same thing until measurement confirmed improvement in both power supply ripple reduction, as well as IMD. Before and after attached; pictures of after condition in two images.

I can not analyze your pictures. What is what ?

What are the circuits you compared ? And the results ?

In one of the plots the distortion component is some -110 dB below fundamental. What signal source you used ?

[zigzagflux]

There is only one picture showing power supply ripple and IMD. That is the before. The other two are after the cascode was installed. You can see both THD as well as power supply ripple is changed when going from single FET to cascode. It does make a difference, even with a 12B4.
Source was just a transformer coupled DCX2496 with digital input, then feeding a P-P 12B4A preamp WOT.

[artosalo]

Sorry. I can not analyze your pictures. I can see in the first picture -110 dB hum component and some distortion components, but not such picture that makes the comparison possible.

-110 dB hum level is few tens of dB's below hearing level, so there is nothing to improve.

[zigzagflux]

So you don't see the two 120Hz sidebands around the fundamental in this picture? Those are getting through the single FET CCS from the power supply. It is most simple to see they are gone in the other images, no heavy analysis required. No need to know every last detail about the circuit; one has a cascode, the other does not. Yes, these are way down near the noise floor of the soundcard, but they are real. I don't trust absolute distortion measurements down this low, so don't choke on the harmonics too much; that's not what is being compared.

Whether things are audible or not will be a neverending debate. It is one thing to claim audible differences that cannot be measured. These can be measured, and for that reason alone I strive to improve upon it. Never made claims about audibility; but I do make the claim there is measureable improvement with a cascode CCS.

[6L6]

Originally Posted by artosalo

I do not know how well the LED string can stabilize the voltage over wide temperature range, but this zener + diode works well.
Anyhow zeners need much higer current to operate.

Thanks! I will try it as drawn first.

Zigzagflux -- Do you have a schematic of the CCS that you are showing these measurements? I would like to try that as well.