SS or tube voltage amplifying stages in an amp with tube output?

[ilimzn]

Originally posted by ray_moth
IMHO, the main advantage of a tube amp over an SS amp is the ability of a tube output stage to clip relatively benignly, especially if care is taken in the design to ensure that this will be the case. Most SS amps clip suddenly and sound very harsh at overload.

Actually, I find this is largely an overgeneralization. With AC coupling and GNFB, it is more than possible to produce rather bad clipping effects in classic tube amp topologies, due to varius bias changes resulting from coupling caps charging via suddenly positively polarized grids. On the other end of the argument, it is not too difficult to devise topologies for SS amps that are very much resistant to clipping after-effects. These mostly stem from uncontrolled charging of various diffusion capacitances (and hence stored charge effects) under clipping. A diode here or there or limited currents to do the charging solve this problem. That being said, once an output transformer is in the picture for tube amps, it is practically unavoidable to have them clip differently, if benignely in both cases.

[b]
However, I don't think tubes really offer any advantage over SS in an amp's front end or preamp, assuming good design and use of appropriate devices.
So, my question is: does it really make any difference whether voltage amplifying stages are built using tubes or SS devices, in your experience?[b]

It should be noted that when you say 'voltage amplification' you are more often than not, talking about two different things when it's referring to SS vs tubes, and even then, solely looking at SS, it's different when referring to BJT and FET. It is very difficult to talk about this in general terms.
For instance, a triode can be phenomenally linear when CCS loaded, as a voltage in, voltage out amplifier, meaning, input impedance is high, BUT output should ideally have an infinite load impedance. With BJT devices, since their most linear characteristic is current amplification, the load impedance is often lowered as much as possible. Quite often it's done with FETs too. Hence, the way you build a hybrid amp topology (i.e. one with a tube input and SS output as in your question) will be (possibly very) different, if you want to exploit the best characteristics of your active devices. From that standpoint, we soon come down to an apple vs orange comparison.

Originally posted by SY
A tube input/driver stage will result in an amp with better linearity than the same amp with ss voltage amplification. OTOH, for follower service driving output stages, silicon rules.

Although i think I know what SY is pointing out, this is a rather generalized statement, begging for the definition of the term 'better linearity'. For one, comparing to costs of tube circuits, it is dead easy to add dosens of extra SS devices to improve linearity, and also, it is MUCH simpler to add oodles of NFB in SS circuits. Looking at the gross maths of it all, you can make a SS circuit more linear per buck spent, but things change if we remember that the ear ultimately is not linear in the classical mathematical sense. i remind you, all of this assumes that you gave the proper attention to the peculiarities of both active elements and theri interfacing - just implementing one topology with the other type of active component simply will not do. It has been my, though limited, experience, that tube inputs can yield comparativelt more linear results with less NFB (local, global, or if one looks at it from the standpoint of lower OLG inside the same NFB loop network) with a minimum of amplifier stages. That's even not considering the more 'esoteric' mechanisms of distortion as mentioned by Miles Prower below.

Originally posted by Miles Prower
[B]
The main advantage of hollow state over solid state is the greater linearity and the freedom from distortion mechanisms such as parametric frequency multiplication.
If you want to build a great amp, the place to start is with the open loop performance. If you do the open loop design right, then you don't need enormous amounts of gNFB to sweep your design flaws under the carpet -- and along with them, lots of the musical detail as well.

This is one of the things i think i have already referred to in the other thread, about SS front ends and tube outputs. Nonlinear reverse transfer capacitances are a big problem in SS devices and require topological steps to minimize (but note, not dispense with, some part of that non-linearity always remains). With tubes, the capacitances are as linear as they can be. Some SS devices also have the emntioned minority charge storage problem, tubes don't. A well chosen tube also has great linearity with no NFB applied, BUT in general this is payed for by lower gain. So, while some mechanisms of distortion are not present, it's not a free ride. once again, I cannot stress the need to consider all of this in designing a hybrid topology.

Originally posted by gingertube
I've been known for the annoying habit of going back to first principals when designing anything.
The triode tube is the most linear amplifying device ever invented. Need we say any more?
Cheers,
Ian

Actually we do - depends on the triode, and the implementation. Good triode, CCS loaded plate, veru high load impedance - and you get great linearity with no NFB for large voltage swings, but relatively limited gain (note that I am using 'gain' in the general sense, for instance, including gm).

[SY]

Although i think I know what SY is pointing out, this is a rather generalized statement, begging for the definition of the term 'better linearity'. For one, comparing to costs of tube circuits, it is dead easy to add dosens of extra SS devices to improve linearity, and also, it is MUCH simpler to add oodles of NFB in SS circuits.

To clarify, in a tube power amp, the limiting factor on feedback is the output transformer (15-20dB is typical, 25dB is a LOT). So that massive open-loop gain of a solid state input stage will mostly be useless. The feedback will have to be applied locally (i.e., not including the output stage) or in nested loops. Local application is nice, but doesn't take care of the biggest distortion contributor. Once could apply feedback around the input stage only, then put in an overall loop. But the overall loop will generally reduce the effectiveness of the local loop, so we go 'round and 'round again.

By comparison, a tube input stage will have more modest gain, so less opportunity for distortion reduction by feedback. BUT it can be made to be orders of magnitude more linear than the output stage. At the 70V p-p swing typical at output stage grids, a good tube will show better than -80dB third and vanishingly low higher order odd harmonics (the even order are largely canceled in the output stage). With 20dB of feedback applied, the well-designed tube driver ceases to be a noticeable contribution to the amp's distortion.

[jlsem]

Good triode, CCS loaded plate, veru high load impedance - and you get great linearity with no NFB for large voltage swings, but relatively limited gain (note that I am using 'gain' in the general sense, for instance, including gm).

Having said that, vacuum tubes have enough gain. Regarding the watts per dollar initial cost, if tubes and transformers were cheaper than transistors and capacitors, there would be no need for the latter in home audio.

John

[kstagger]

just a little historical note - ARC used all solid-state front ends in the late 80s with their line of 'Classic' amplifiers

schematic of the Classic 30 is here :
http://www.arcdb.ws/CL30/CL30.html

edit : I see they still use a 6CG7 to drive the 6550s

[refference]

[QUOTE]Originally posted by kstagger
[B]just a little historical note - ARC used all solid-state front ends in the late 80s with their line of 'Classic' amplifiers


Excellent example !!!!!!