|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
79.222.135.42
In Reply to: RE: High transconductance tubes sound as transistors... posted by Aldovan on March 29, 2014 at 19:32:10
Hi!
It is all about the circuit and how you use the tubes. May be you do too much tube rolling rather than optimising a circuit for the specific tube? You can't compare all these tubes directly since they require very different circuits to work optimally
In my experience: If it doesn't sound good it is not the fault of the tube, but of the designer.
Best regards
Thomas
Follow Ups:
"If it doesn't sound good it is not the fault of the tube, but of the designer."In keeping with that;
The single ended stages in audio amplifiers are Class A out of necessity.
Part of the definition of Class A (not the short definition at the beginning of the chapter but the full descriptive definition found in the main text of the chapter) is that for low harmonic distortion, the tube is "only operated in the most linear part of the dynamic curve".
The operating condition (voltage, current, load and max. input signal) will not only determine if the tube is operating in the most linear part of the curve but also how linear and long that part of the curve is.
For low harmonic distortion it's up to the designer to make sure that the tube is being operated in the most linear way possible.
For me, this usually means relatively high idle current and active plate loads that keep the load line horizontal. This insures the tube operation stays away from the non-linear cutoff region.
A designer can stray from this and "color" the amp however he wants to for taste.
At that point the discussion regresses away from science and all bets are off.
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
Edits: 03/30/14
Unfortunately, tubes and circuits that use them are - at least a little - nonlinear EVERYWHERE. So for this to be meaningful, there must be some arbitrary boundary between "highly linear" and "not so linear".
One time-honored definition is the point where 5% THD occurs, into the specified resistance load. Here are some problems with this definition:
1) If this is useful for no-feedback SET, then it is probably not useful for push-pull, or pentodes or beam tetrodes or transistors or FETs or anything with feedback - they all have different harmonic distortion signatures.
2) This is specified for a resistive load, thus is less relevant for a loudspeaker load. How exactly do you specify a load region, or linearity within such a region?
3) Even for a single stage of no feedback SET, clipping or compression on one side (grid current) sounds different from the other side (plate curve compression).
Another time-honored tradition applies to high feedback amps, where you can call it linear as long as no clipping occurs. But you still have to define "clipping" pretty carefully, and you still have to address the load reactance questions.
I'm not happy with either of these, but I don't see any definitions that are better either.
"Unfortunately, tubes and circuits that use them are - at least a little - nonlinear EVERYWHERE. So for this to be meaningful, there must be some arbitrary boundary between "highly linear" and "not so linear"."
I guess I shouldn't be surprised when I make absolute statements and people take me seriously. I wasn't trying to be specific, I was really just speaking in general.
All I was really trying to say was there is no excuse for running triodes in very non-linear ways.
I was in a Chinese 300b amp the other day and the first stage was a 6sn7 running at about 3.4ma with a 62k plate resistor with 70 volts on the plate direct coupled to another 6sn7.
Or how about a 7b4 running less than .5ma with a 285k plate resistor at about 210 volts plate trying to drive an output tube with about 70pf of Miller capacitance.
Swinging the grid just 1.5 volts peak to peak causes a -55 and a +43 volt plate change.
Neither of those are what I would call highly linear operating points.
In fact they are both butt ugly.
With the 7b4 and a CCS plate load at 1.2ma and 200 volts plate, we can get -48 and +48 plate change from a 1 volt peak to peak grid swing.
I call that linear. (It still won't drive the 70pF though)
Oops, there I go again, making absolute statements that someone's going to parse and rip apart.
Oh well, I have broad shoulders and thick skin. I can take it.
Tre'
Have Fun and Enjoy the Music
"Still Working the Problem"
Using a Pimm-style mu-follower CCS could probably deal with the 70-peeks of grid capacitance.
cheers,
Douglas
Friend, I would not hurt thee for the world...but thou art standing where I am about to shoot.
Thanks, man. Yeah, I was expressing some frustration of my own - I apologize if the flak hit you accidentally. :^)
But I must say, you brought up a serious and important question, which it was my intention to further address - what (exactly, measurably) is a good definition of linear operation?
My own rule of thumb for power triodes - I can't call it a definition - is to design for a load line that gives twice the quiescent current, at zero grid-cathode volts. With a really linear tube (see below) that gives about 3% second harmonic at full output. Less linear tubes will have more distortion, I've calculated as much as 10% based on published curves, but those tubes will distort at practically any operating point! The reason I chose this is that even with huge negative feedback the tube will remain in Class A throughout a full sinewave cycle.
My model of tube linearity is based on a distribution of mu along the cathode. If the mu is everywhere the same, then cutoff (zero plate current) happens at the same grid voltage everywhere. I call such a triode "linear" because it follows the 3/2 power law exactly. That law is of course not a linear function, but you can't in theory do any better with a triode.
Of course, if the load impedance is much greater than that which produces the most power - which is the case for drivers if they have a high-impedance plate load such as a choke or current source - then it is easier to operate the tube in a region where the mu does not vary significantly, and you can this way approach truly linear operation.
This leads to the idea that there can be a region of nearly constant mu, within which tube operation is actually quite linear. But you still need a number - perhaps a 10% variation of mu is acceptable? Open for comments!
If you are comparing something like a 12AX7 loaded with a 100K resistor against something like a 6C45PI with an active load, each stage will sound very different.
If you like some extra distortion from your driver stage, then stick with the low transconductance tubes with less than ideal loads.
If you want a reasonably transparent driver stage for your DHT finals, these tubes are nice choices.
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: