[G Georgopoulos]

Hi

I have been trying to make tube amps more linear,using negative feedback,there is no way i can think of to achieve that,no way,period,even tube amps that use nfb are misleading,face it,tubes are very nonlinear devices,the magnitute of high voltages and resistances make linearisation impossible,couple that with low transconductance,render nfb useless,so tubes are just that nonlinear,good or bad,that's what i have found out,since you can't do anything about it,one should use them as they are,using simple biasing etc,the reason i'm posting this is i want to know if others know something i don't,so please share here any technique that makes tube more linear or just comment on what i just posted..

cheers

[FullRangeMan]

GM70 tube is famous to be a very linear tube.
Some users report adjusting bias once a year.
As you know linearity varies between types of
valves, as you did not say what is the tube it
passed the idea that all the tubes are unlinear.

If the power tube is an model know to be linear
the prob is in the driver circuit or in the OPT project.

[Steve]

Hi

I have been trying to make tube amps more linear,using negative feedback,there is no way i can think of to achieve that,no way,period,even tube amps that use nfb are misleading,face it,tubes are very nonlinear devices,the magnitute of high voltages and resistances make linearisation impossible,couple that with low transconductance,render nfb useless,so tubes are just that nonlinear,good or bad,that's what i have found out,since you can't do anything about it,one should use them as they are,using simple biasing etc,the reason i'm posting this is i want to know if others know something i don't,so please share here any technique that makes tube more linear or just comment on what i just posted..

cheers

I think this might help G. According to the RCA Radiotron Designers Handbook, page 611, Shorter's research concluded:

System C Just perceptible 2.6% RMS sum of Harmonics, Weighted Distortion 2.8%, no kink in linearity curves.
System E Just perceptible 0,62% RMS Sum of Harmonics, Weighted Distortion 1.3%, slight kink in linearity curves.
System A BAD 2.3% RMS Sum of Harmonics, Weighted Distortion 5.1%, Pronounced kink in linearity curves.

Weighted distortion, some engineers use a weighted distortion factor in which the harmonics are weighted in proportion to their harmonic relationship.
Some say as low as 0,7% with bandwidth of 15khz.

One may also lower the input stage/driver stage to as low of distortion as possible. This will keep the higher orders of distortion at a lower figure.

As one can see, there are several factors to consider when dealing with perception of distortion. This is not exhaustive, got to keep some secrets. 

Cheers
Steve

[Freo-1]

Tubes are more linear than sand devices.  This is well known and understood:

http://www.theaudioarchive.com/TAA_Resources_Tubes_versus_Solid_State.htm

[Mags]

Tubes can be extraordinarily linear, especially DHTs. Triodes have their own internal feedback mechanism to make them inherently linear, far more linear than typical transistors which is why transistor amps typically use gobs of global negative feedback, while a wonderful-sounding triode amp may be made with zero global negative feedback. The additional elements inside tetrodes and pentodes mitigate those internal feedback mechanisms and make those tubes more like transistors, so they typically need some feedback added to sound their best.

[Mags]

Take a look at these curves, for a triode-wired 6E5P tetrode. Evenly spaced curves indicate linearity. Spectacularly linear.

[G Georgopoulos]

this is for my friend Steve with all the secrets
the tubes use a resistor load amplifying voltage
because of the low transconductance of the tube
this generates distortion
no matter what the input z is at the other end
with ss we use bootstrap or current source for amplifying voltage
and of course the higher transconductance of ss
this lessens distortion + huge nfb even more

i have spent a few years with tubes
steve have you done ss at all

cheers...

[FullRangeMan]

George do are follow the schematic?

[Steve]

this is for my friend Steve with all the secrets
the tubes use a resistor load amplifying voltage
because of the low transconductance of the tube
this generates distortion
no matter what the input z is at the other end
with ss we use bootstrap or current source for amplifying voltage
and of course the higher transconductance of ss
this lessens distortion + huge nfb even more

i have spent a few years with tubes
steve have you done ss at all

cheers...

The reason I mention perceptibility in my last post was that super low distortion is not necessary. Secondly, what harmonics are present is also important as each harmonic is weighted; that is higher orders are more easily perceptible than lower orders for the same %, etc. We are also dealing with large signal tubes producing the most distortion.

I could design a tube amp with easily less than 0,1% distortion, .01% if you wanted such. Just design it with huge gain, and huge negative feedback. But why when one does not need it and the increase in parts count degrades the music? By the way, there is a wide range of global negative feedback where higher orders harmonics are produced, which is the last thing we want. The global negative feedback is also delayed through the amp by phase shift and also transit time through the amplifying devices, thus delayed back to the input. As one can see, there is a little more to contend with than simply HD and IMD. By the way, why does one need bootstrapping and current devices if transistors are such a low distortion devices?

Now take small signal input/driver tubes, such as a JJ E88cc tube. The JJ brand is about -78db (approx. .012%) in harmonic distortion with virtually zero feedback at 2 vrms, not boot strapped, no current sourced, or any other complicated configuration. (That is some 1/8th the HD of a 6SN7 in common cathode configuration.) This means less circuitry/parts and signal degradation. I have also designed circuitry to produce 200 v p-p with 0,05% harmonic distortion. I probably could have lowered the distortion even more, but why since I only need 50 v p-p or so?

(Notice how poor most recorded music is to real music and check the complexity of the recording equipment, even with its "low harmonic distortion" specs.)

If one checks the specs, the DHTs produce the same % harmonic distortion, if not more than the IDHTs. Besides that, one must consider the distortion produced from the input/driver tube(s) necessary to drive the ridiculously high signal drive requirements of most DHTs. The distortion from the input/driver combine with the output tube and output transformers distortion to add higher harmonics.

Let's take a simple two stage amplifier. The input/driver produces only 2nd and 3rd harmonics. The output tube stage also produces only 2nd and 3rd. The two stages will combine to produce 4th, 6th, and 9th orders of harmonics. If the input/driver produces virtually no harmonics, then we only have 2nd and 3rd from the amplifier (plus the OPT distortion %).

Yes, I believe there is a type of transistor that produces virtually no HD or IMD. Been a long time since I delved into SS. But does it really matter when one can design an amp with little global negative feedback, is not complicated with parts to degrade the signal, and what HD and IMD are virtually not perceptible.

By the way, most amps, in a typical relaxing setting are only supplying a watt or two. We are talking less than .1% for sure in most cases. At higher levels, the ear is not as perceptible to distortions, and the speaker also introduces its own distortions.

Cheers and all the best G.
Steve

[nickd]

I have been trying to make tube amps more linear

George,
I would not claim to be an expert. However as I understand it, The weakness in most tube designs (if you could call it that) is the standard of using coupling caps and output transformers. I suppose that's where some of the "beauty" of tube sound comes from. It is a bit of a black art to make either part affect sound quality for the better as neither part is totally "transparent". They always affect the sound. While the Tube may be Linear in its function, the total circuit is a bit more complex.

Having owned a Ayon and NAT tube gear in the past, I would say "Linear" circuit's using tubes to amplify voltage are not only possible, they already enjoyed by everyone who can live with the cost, heat and space requirements tube designs require.