Why are Constant Current Sources Awesome

[magnafuzz]

This might be a bit of a newbie question but oh well. Im trying to learn more about tube amps and amp design.

What is the benefit of using a CCS instead of the output resistors. I'm guessing the answer is... constant current... but i was wondering why the constant current vs resistor load makes such an appreciable difference as it seems that the CCS upgrades are well established as worthwhile and noticeable upgrades.

I read somewhere (I forget where) that the fluctuations from the output signal pulls the reference voltages up or down whereas with a constant current source they would remain constant. Is this correct/close/completely off? Is there more to it?

Also, are CCS's only effective or more noticeable on certain topologies?

[porcupunctis]

This will not be the most knowledgeable reply but might help explain things in a less detailed way.  Tube based guitar amps were built to be inexpensive and easily manufactured (for the most part.)  When you laid into the bass or a big power chord on these things it drew a lot of current and the voltage sagged dramatically.  Musicians make the most of what they got and turned this effect into a whole genre of bluesy-rock sound.  Great for blues guitarists but bad for high fidelity reproduction.

This is my simple understanding of what a constant current source will do for an amp.  It compensates for these changes in current draw due to changes in signal.  This keeps the output true to the input regardless off what the signal is doing.

You don't have to understand the physics to appreciate the difference.  You can hear it.

I would like to understand it all a little better myself and I'm hoping Paul or Doc will chime in with one of their classic explanations that make me want to do bunch of additional research.

[Beefy]

Great article here: http://www.ecpaudio.com/pdf/parafeed_basics.pdf

Refers largely to parafeed, but has a great section on the different types of tube loads.

[VoltSecond]

Here are some old notes on CCS ( Constant Current Sources )

http://www.siteswithstyle.com/VoltSecond/CCS_college/CCS_College.html

Where to use the CCS?

I don't see the need to CCS the grid of a tube nor do I recommend it.

On single ended designs, you CCS where the "ac" energy is coming out. On a cathode follower, you CCS the cathode. On a stage with plate output, you CCS the plate.

On push-pull designs, you can CCS the cathode of the "long tail pair" or where the cathodes of a stage are tied together.
http://www.siteswithstyle.com/VoltSecond/K-12M_AMP/K-12M_Push_Pull.html

On triodes, only CCS one pin.  The other pin normally will need to go to an "ac ground"  An "AC" ground is ground or a capacitor going to ground. It is called an "ac" ground because at "ac", it is attached to ground.

On a Pentode, you can also have a CCS fed shunt regulator on the screen.

On push-pull designs, you can CCS the cathode of the "long tail pair" or where the cathodes of a stage are tied together.

[Doc B.]

Great article here: http://www.ecpaudio.com/pdf/parafeed_basics.pdf

I would say that fellow has studied our designs quite a bit ;^)>

[magnafuzz]

Thanks for the info, interesting stuff. looks like i have some bedtime reading ahead of me

[magnafuzz]

So, lets see how my understanding is...

For a resistive load, an increase in plate voltage will decrease current through the tube, reducing the cathode voltage which causes the current to rise and the plate voltage to stabilize. in other words the resistive load has a feedback loop which stabilizes the plate voltage by changing current.

The CCS will hold the current and cathode voltage constant without the feedback loop.

I would expect that the feedback nature of the resistive load would cause peaks to be somewhat smoothed or dulled due to the feedback loop sort of interfering with the output, reacting, counteracting and stabilizing it. I would expect that the CCS would tighten up and improve the impact of these signal peaks. Plus the improved PSRR.

[Caucasian Blackplate]

For a resistive load, an increase in plate voltage will decrease current through the tube, reducing the cathode voltage which causes the current to rise and the plate voltage to stabilize. in other words the resistive load has a feedback loop which stabilizes the plate voltage by changing current.

This little nugget explains some other choices Bottlehead made.  To resist that change in cathode voltage, convention has suggested a large capacitor across the cathode resistor to stabilize its voltage, but this capacitor sees signal current, so instead carefully chosen LED's were used instead that maintain nearly a constant voltage over a wide range of current draw.

A CCS does also keep this voltage constant, and the PSRR can't be ignored.

The last piece of the puzzle has to do with load lines, I suspect that is buried in the links provided.

[Maxwell_E]

That really is a great article, Doc. I always wondered about where the C4S goes in the circuit. I have the original schematic for my S.E.X. amp, but when I installed the C4S I never saw an updated one. Whatever, I'm too busy listening to it to re-diagram it.

[magnafuzz]

For a resistive load, an increase in plate voltage will decrease current through the tube, reducing the cathode voltage which causes the current to rise and the plate voltage to stabilize. in other words the resistive load has a feedback loop which stabilizes the plate voltage by changing current.

This little nugget explains some other choices Bottlehead made.  To resist that change in cathode voltage, convention has suggested a large capacitor across the cathode resistor to stabilize its voltage, but this capacitor sees signal current, so instead carefully chosen LED's were used instead that maintain nearly a constant voltage over a wide range of current draw.

Yes! i was quite pleased with myself when i was looking over the Crack schematic and noticed how the LEDs would be beneficial.

The last piece of the puzzle has to do with load lines, I suspect that is buried in the links provided.

I haven't spent too much time with tube load lines but ill give it a shot. The CCS line should be flat, or close to it. Would the advantage here be that the resistive, negatively sloped, load line crosses over the less linear portions of the current vs voltage graph? i believe that relates to distortion.

[Caucasian Blackplate]

That is along the right lines.

With the flat-ish load linenyou will tend to see a very even spacing of the intersection of the load line and the grid voltage line (lower distortion).

This is an extreme oversimplification, PJ can undoubtedly fill in a lot of gaps here if necessary.