Cons / Pros

[Borat]

first let me say what i do NOT want this thread to be. i don't want it be an argument about which amps are better - tube or solid state. i think we have all witnessed enough such arguments and at least at the moment i would rather not have another one.

so lets avoid drawing any conclusions, but instead present only arguments.

i want to see a partial but somewhat comprehensive list of cons/pros of each technology.

for example arguments like " this type of device has more gain " or " this type of device has more bandwidth " or " this type of device allows for designs with simpler signal path " or " this type of device is more linear " etc.

in other words lets have some objective statements about each technology without sliding into comments like " this type of amplifier has no soul " or " people who like this type of amplifier like to eat babies "

also for both technologies let the context be that of a Class A amplifier.

[aardvarkash10]

You make a number of not entirely sustainable assumptions. In no particular order:

1 - partial can be comprehensive
2 - technology begets outcome
3 - music and the pleasure of listening to it can be derived from and defined by the technology that reproduces it ie the pleasure (an emotional response) is reliant on the technology and hence emotional outcome is the sole result of technology.

The discussion is pointless except for the purposes of a decent fist-fight between the extremeists for any technology. It makes as much sense as arguing whether the woodwind or the brass sections more truly represent the pure production of human air expelation derived music.

Its a straw man.

If I want pure *** kicking horsepower derived kilowatt requiring dance and rave music, I'm going for SS, if only as a matter of practicality.

If I want to sit in the dark, with just a glow in the corner and simple clear defined passages playing, I'll go for glass.

And for building, nothing beats the pleasure of surviving high voltage. Except when I'm teaching when an LM1785 can light up a 15 year old students face (and brain) in under a day.

It ain't no competition. Why make it one?

[kavermei]

High flammability...

[Miles Prower]

Originally Posted by Borat

also for both technologies let the context be that of a Class A amplifier.

This, right here, is unrealistic. Most SS amps don't run Class A (neither do most hollow state amps). Which, of course, brings up one advantage that hollow state has over solid state. VTs like to run hot. Getter material works better hot, and there is nothing that heat can really harm in a VT. Even if you run Class AB, you can get away with a higher static plate current to get the operation more towards Class A. Transistors get all hot and bothered when you warm 'em up. Too hot, and the dopants start to migrate. Therefore, Class A hollow state doesn't require the huge heatsinks transistors require. Class AB, especially running deep into Class AB, lets you get away with much smaller heatsinks.

VTs are also more linear, and better suited to analog applications. The less distortion you have open loop, the easier it is to get good performance.

VT clip behaviour on overdrive is a good deal gentler, producing much less of the nasty high order harmonics. With VT amps, you simply don't hear the occasional clip that would be highly audible with a transistor amp. Especially since these need much larger margins of feedback which makes clip behaviour even worse.

Being low gain devices, it makes a big difference what type of VT you choose, not just the circuit topology. The right type becomes even more critical if you intend to run open loop. With transistors, you can just about wire in whatever type you want, so long as it can process the proposed frequencies, or doesn't require some special feature (such as a very low noise figure). The high gain, even in a low gain application, makes performance nearly independent of device characteristics. Done many a project with "transistors anonymous" picked up at Rat Shack.

Being a low voltage, high current device, transistors are also LoZ devices. This makes it much easier to get wideband performance. If you need a high cutoff frequency to minimize phase errors, it's easier to get it with transistors.

Doesn't mean that you can't get good performance from solid state. I've done solid state designs that turned out very well indeed. Almost, but not quite, as good as hollow state designs. I intend to do more in the future, particularly doing MOSFET amps right. (Especially with these new SiC JFETs -- I sense a hybrid Circlotron design coming up if I can get ahold of some of these.) Most of these sound "off" because they're not using the devices correctly. BJT style OTL topologies should stick with BJTs. Weren't designed for MOSFETs, so it's no big surprise most MOSFET amps don't sound right.

Lastly, I'm not so adverse to using solid state where it does the job best. You can't beat cascoded BJTs for active loads, and nothing can slap around Class A*2 grids like a source follower. Much better current sourcing with Zo's the likes of which they couldn't dream back in "the day". Both go to reducing distortion greatly in this type of operation.

[Gordy]

Naughty Mr. Borat, this thread is heading in almost the same direction as the one you just posted in Solid State ( 'who makes amps that theoretically should sound best ?' ).

So let me quote from yourself:

Originally Posted by Borat

I'll tell you guys what MY philosophy is when it comes to amplifiers.

I believe the best sounding amp will be a very complex design with a very simple signal path.

For example starting with an average amplifier you can first give it separate power supplies for left and right channels. then you can perhaps split the power supply for the gain and the output stages. then you can make power supplies regulated. all of this would of course amount to a very large increase in complexity.

at the same time however you could be taking steps to SIMPLIFY the actual signal path. reduce the number of gain stages, eliminate any unnecessary parts such as inverter at the input ( use differential input directly ) etc.

switching to class A operation would also simplify the life of the signal so to speak while at the same time it may require increasing the complexity of the amp elsewhere. for example to avoid ripple / hum you may need to use a more complex power supply design and to avoid overheating you may need to use some sort of microprocessor control of bias etc.

so basically i believe the path to perfection is at the same time through greater use of technology in the *supporting* parts of the amplifier and through greater simplicity in the actual signal path.

I think I understand your positive sentiment in starting these threads, and I agree with most of what you say as quoted here. However, despite my previous post in your other thread, you still blindly miss the fundamentals. Remember that the amp operates in a system, not in isolation. Also, the final arbiter of performance is human intellect, not a calibrated test-set. I recommend that you consider investigating these first before deciding upon amplifier design...



PS solid state designers really do eat babies

[Borat]

Originally Posted by Miles Prower

VTs are also more linear, and better suited to analog applications. The less distortion you have open loop, the easier it is to get good performance.

is this statement universally true or topology dependent ?

Originally Posted by Miles Prower

Lastly, I'm not so adverse to using solid state where it does the job best. You can't beat cascoded BJTs for active loads, and nothing can slap around Class A*2 grids like a source follower. Much better current sourcing with Zo's the likes of which they couldn't dream back in "the day". Both go to reducing distortion greatly in this type of operation.

are you saying run VT gain with SS output stage ?

what are your thoughts on mixing VT and SS in the same signal path ?

[Borat]

Originally Posted by Gordy

However, despite my previous post in your other thread, you still blindly miss the fundamentals. Remember that the amp operates in a system, not in isolation. Also, the final arbiter of performance is human intellect, not a calibrated test-set. I recommend that you consider investigating these first before deciding upon amplifier design...

thats great but we can't transform our emotion into a circuit. if you think you can - good for you.

the rest of us lowly humans are going to have to be content with using things like logic and physics to design.

notice i am not asking how judge an amplifier but how to design one ...

[Gordy]

Originally Posted by Borat

thats great but we can't transform our emotion into a circuit. if you think you can - good for you.

the rest of us lowly humans are going to have to be content with using things like logic and physics to design.

Not only did you completely and utterly miss my point, but you also managed to be sarcastic in your reply.

Oh well, good luck with your quest.

[Bandersnatch]

hey-Hey!!!,
I don't buy the 'tubes have more pleasant overload' for a minute as the reason I like tube amps. I have speakers that do very nicely at a Watt, and the amps driving them are capable of a clean 50 Watts. Tube amps dound bad when clipping, and if you make them clip they sound bad.

I'll see about measuring them again when there is a set of measurements that are able to deliver predictions about what amps sound good and which ones don't. If it were just the current set of stuff, the SE amps would come in dead last, and in listening tests they usually don't. I don't want to live with them, but that's my preference...
cheers,
Douglas

[Miles Prower]

Originally Posted by Borat

is this statement universally true or topology dependent ?

Of course it's not universally true. It's true where low to medium gain triodes are concerned. However, there are other characteristics that are desirable for different applications. These would include a variable-u factor that makes for wide range gain adjustments by grid bias. Highly desirable for AGC of IF gain stages in superhet xcvrs. Plays hell with audio linearity (good for making FX boxes, however). Pentodes and high-u triodes tend to make more of the higher order harmonics, and you have to take more care to see that this isn't a problem in your designs.

As for linearing up transistors, you can use cascodes and/or local NFB. It takes even more effort, but every little bit of open loop performance increase does help.

Of course, a good device in a bad design will perform badly, so it's dependent on both the active device and the circuit design.

what are your thoughts on mixing VT and SS in the same signal path ?

Here's a case in point: Wolverine Main Schemo

The A Number One problem with the 845 is its notorious tendency to pull grid current even before Vgk actually goes positive. Originally, they figured t'hellwiddit, and specced a grid bias of -155Vdc with a signal input of 300Vp-p, keeping Vgk negative at all times. By using a MOSFET grid driver, this is no longer a problem. You can get a couple of extra watts from it by increasing the input swing to 310Vp-p, and you get a better behaved transient response as well since there's no lack of grid current capability. The MOSFET driver also has a much lower Zo than you'd get with a cathode follower.

Using an active plate load for the second gain stage not only helps improve linearity, it also allows you to develop the huge input swing the 845 requires, due to its very low u-factor, without having to resort to enormous DC supply voltages and high watt resistors, or by putting a coupling transformer in there.

"are you saying run VT gain with SS output stage ?"

As for running hybrids, I already said that.