Tube vs. SS CCS...

[nickds1]

Any opinions on this - other than aesthetics?

I like the all-tube thing, but will it gain me anything over a FET CCS?

Thanks

[SY]

Well, there's the esthetics...

In some uses with some CCS topologies, there's a theoretical reason why tubes could be superior (capacitance modulation). When actually trying it, I haven't found a measurable or audible advantage. This assumes that the solid state CCS is a good one and not a cheap'n'cheerful one transistor design.

[Miles Prower]

Originally posted by nickds1
Any opinions on this - other than aesthetics?

I like the all-tube thing, but will it gain me anything over a FET CCS?

Thanks

Not really. For CCS duty, solid state is better, and cascoded BJTs are better than FETs (especially those horrible constant current "diodes" -- actually a JFET biased internally to draw some specific current -- avoid like the plague). The reason for this is that the BJT is much higher gain device than any pentode, so it will present a higher impedance to the load. There is also the additional advantages of not needing another hole in the chassis, and the likely possibility that the SS CCS won't require an extra negative DC rail (depending on where you're using it) or that the negative rail can be easily derived from the heater xfmr since it won't be needing 100+ volts to operate.

[Brian Beck]

Like many interesting things in life, it depends! It depends on how well the CCS is implemented of course, no matter whether tube or SS.

But it also depends a lot on WHERE it's used, especially from a non-linear capacitance loading point of view (capacitance from CCS). More specially, it depends on the total resistance seen at the node to which it’s connected. Think of it in terms of an RC time “constant”. A CCS that might be OK in the low-impedance cathode of a tube such as a 6DJ8 when used as a CF might not work as well in the plate circuit of the same tube when used as a common-cathode gain stage. This is because the plate resistance is mu (actually mu+1) times higher than the cathode impedance and therefore the capacitive modulation of phase will occur at a mu-times lower frequency (mu times greater/worse effect).

Similarly, a FET-based CCS might be OK in the plate circuit of a low-rp tube like a 12B4 with rp = 1K, but a similar (although lower current) CCS might sound terrible in the plate circuit of a 12AX7 with its rp = 80K. Picofarad-for-picofarad, the phase modulation effect is 80 times worse in the 12AX7 circuit.

[EC8010]

Another point is that a CCS is effectively a high gain amplifier that amplifies a DC reference voltage and has your circuit as its load. When a valve CCS feeds the cathode of another valve, the voltage gain is pretty low, but it isn't zero, and one of the things that can be amplified is noise on the CCS valve's heater.

The best time to use a valve CCS is when you already need its output to be at +100V or so and the power to be dissipated is quite high. In the same situation, a transistor cascode has heatsink problems, either losing the heat, or stray shunt capacitance.