[MrCurwen]

I am specifically not dealing with single triode tubes,
because for me the only gain would be more linearity(non issue becaue, as MrCurwen stated we have feedback)

I'm sure most have heard me quote the old "NFB is trading gain for linearity". There is another level to this; within this linearity that is acquired via NFB. It trades lower overall THD to higher proportion of higher order products.

So let me clarify.

If you have a circuit that produces 1% THD consisting of only 2H and 3H and you apply gNFB, you then get lower (let's say 0.1%) THD, but the spectrum of harmonics changes so that you now have very small amounts of 4H, 5H, 6H, and 7H.

What this means to you is up to your ears and brains. Just something to think about. (I'd like to reiterate that "the SS sound" in my opinion is not caused by SS, but by NFB.)

To bring it back to the quote above, it's not completely irrelevant how inherently linear your gain stages are, even if they are inside a loop. It is mostly irrelevant, but not completely, and for BEST RESULTS HIGH END AUDIO it definitely is not at all irrelevant.

My personal opinion is that is is always best to not create any distortion in the first place than to try to fix it after the fact.

The issue is that in this driver circuit it doesn't really apply because we are still within the same channel.
So the theoretical & possible gains of triode isolation would result in maybe better noise cancellation ,
wich is already non detectable..
And in this specific amp " balanced " design,
the driver tube actually NEEDS the two triode sections as similar as possible , so within the same tube glass,
is actually a better chance of uniformity.

For balanced SS circuits it is very important to have the devices heat coupled (same heat sink is best). For tubes, heat does change the specs a bit, but not as much as SS components. Having the tubes inside the same glass envelope is not important at all for the matching, especially since 99.99% of the heat inside a high rp low current tube is from the heater anyway, not the actual Ia or Pd.

I'd say single triodes are for best results in circuits where tube matching is critically important, because you cannot get double triodes with identical tube sections.

I don't think in practise this makes much difference in your circuit, but just a thought.

Better still would be a driver tube with common cathode.
I tried some but haven't found any noticeably good ones.

6SL7 has a shared cathode version (for radio duty) 6SC7.

http://www.mif.pg.gda.pl/homepages/frank/sheets/093/6/6SC7.pdf

http://www.mif.pg.gda.pl/homepages/frank/sheets/127/6/6SC7.pdf

It has essentially identical specs and identical linearity.

Very nice metal envelope, JAN, super high quality for military: http://www.ebay.com/itm/1-x-Ken-Rad...894410?hash=item5b349f88ca:g:2mIAAOSwMTZWR08f

These are a bit more expensive than I'd like to pay for a tube, since they are common in early 1950's Tweed style guitar amplifiers. I love tweed amps, I simply use russian 6SL7 for sensible price. Quality just as good.

Russians also make a copy of 6SC7, it's 6N10 but it's not available for sale. Probably only a few were ever made.

Why do you think having the cathodes connected together inside the tube envelope makes any difference compared to connected on the outside?

I am finding out a right combination is best...
Like brighter driver tube will liven a thicker powerube down the line...
So its a balance .

Yep. My own philosophy would be "don't mess with the frequency response in the first tube stage, so you don't have to try to fix it in the second tube stage".

Yes there is lower power output, although I am running them near max bias.

What's your Ia?

What I found was that although all these type tubes have the annotation of "P" for pulse,
Which is an advantage for increased power output..

Kind of. Pulse capability doesn't increase RMS power at all, but it can (in a properly designed circuit) increase transient response which brings about the sensation of power in the listener.

In a circuit that doesn't speficically take into account the pulse capability, they will probably sound worse. Let's go into that:

In my trials upon listening, the music dynamics & on transient attacks, it seemed exaggerated ..
I found that the music dynamics seemed artificially expanded in minute levels at times.
Of all the 6n6p or 6h30 or 6n30dr types I tried,
The bottom line was the russian 6n6p (oldest version without the gold grids)
Were the most natural sounding... In this amp.

Yes. I find this easy to believe.

Back some time ago I wrote about the grid parasitics, with specific mention to high pulse capability tubes (such as these russian tubes and also TV sweep tubes which I've been dealing with).

Basically the bigger the pulse capability of the tube, the bigger the electron cloud hovering around the cathode. Now there electrons hit the grid, and they must find a way to escape then. The circuit around the grid must accomodate this.

Worst case scenario, there is a large resistor from grid node to ground (or other reference that is applicable). This trapped current must then go thru that resistor to escape; current thru resistor causes voltage (Ohm's law). So that shifts the voltage at the grid.

What is the voltage at the grid? It's the audio signal.

This grid current phenomenom happens a little bit on zero signal situations, depending on bias point. In high mu high gm supertubes like 6E5P and other frame grid tubes it most definitely happens even at zero signal conditions. This is because the frame grid is positioned very very close to the cathode.

But of course with signal peaks you understand is where most of this grid current happens. In audio terms it affects transient response.

Now your circuit is not feeding the grid from a low impedance source (i.e. a source that is able to deal with the current easily) so it is no wonder that pulse tubes cause transients to sound 'wrong'. What you are probably hearing is extra distortion on transients, this makes your ear catch them more, not in a good way.

They had both the most natural, realistic and most holographic sound of the bunch... Even over the "supertube" version.... All hype.

100% circuit dependent.

I find that pulse types (such as TV sweep tubes) make THE BEST audio output tubes, provided they have adequate grid drive. Source followers in practise. If the transients are clean, they don't catch your ear in a bad way, instead they make the music more 'lively' and dynamic.

My current workhorse amp has 6E5P output tubes. Incredibly dynamic and also superbly clean. Have a look at the triode curves, sure is a sight for sore eyes aren't they.

I have the amp on high gain as well, with higher bias on both driver and power stages, and using the higher gain type driver tubes.. All to help with "gain".

I always tend to end up either with too little or way too much gain. Current amp has a voltage gain of 3500 (12AX7 -> 6E5P) which is simply stupid and an invitation for oscillation problems. I need my lab set up so I can build a new amp...

Also the increase MU of the 6SL7 type or 6C8G type , for me gives more life and detailing , wich would or could be lost in any non-optimal situations..
So I prefer the choice of a higher gain driver tube .

Yes all things being equal everybody prefers the louder one. That's why A/B tests need to be strictly volume matched.

 

[gug42]

Hello,

Sorry if i interrupt your discution .... hard for me to follow. I'm a future owner of a mk6+ and I was looking for tube.

I have ask the question on the mk6/mk8 thread, but someone told me to ask here for confirmation and explanation :
Can it be a probleme to not have closely matched 6SL7 driver tube ? Around 15-17 % of differences between the anode ?


Thank you by advance !

 

[MrCurwen]

TUBE SELECTION.. (! heh)

One of my most favourite subjects! Always on the lookout for new tube types.

6SL7 is a more linear tube, than a 12AX7 ( which has spikes and dips in it's EQ )...

Could be in your circuit, but in general; no, not at all. There is no inherent difference in linearity in the most commonly used op point area of these tubes. Possible differences in EQ and linearity come from poorly matching the resistor values to the tube rp, and from poor output impedance. 12AX7 does have a bit higher rp and so if it is used to drive a load (such as a capacitor with parasitics or a hungry tube grid) it will do worse. A decent buffer solves this.

Octals were more popular with hifi DIY'ers for RIAA phono preamps,
because the 12AX7 is a horrible match with RIAA correction(!)..

Not in DIY circles, not at all. Where are you getting this?

Smaller 9pin tubes may also be vunerable to problems such as cross talk between the two sections and the heater.

Possibly. Not in practise in my experience. At RF these things matters, not so much at audio frequencies.

Also the plates of the two sections are closer together because of the narrow confines of smaller tube...

This was used to keep the two sections more isolated,
which can also be especially useful for high frequency use or issues (oscilations)..

Yes these things do matter a bit, but the most significant difference with regards to oscillation is mu, which is distance of grid to cathode. Closer to cathode, higher mu, higher parasitics, higher coupling and chance of oscillation.

Still, the smaller size and higher gain factor and headroom of the 12AX7 type increased its popularity over the Octals, in guitar amps in the '60's.

No. Guitar amps used 12AX7 because it was plentiful and cheap. It had become plentiful and cheap because of NFB. Early 1950's saw the onslaught of gNFB audio circuits, which necessitated a lot of gain. So, 12AX7 became hugely popular. Also it was hugely popular because it was the tube of choise for jukebox RIAA correctors, and jukeboxes became very common.

But that's were advantages end and there more bad news ...
The square plates in tubes,
are found in the later (cheaper) constructed tube years, which
is why they usually don't command as high a price as the rectangular
or "round plates".

Ah, tube electrode geometry! One of my most favourite subjects!

Round plates are more desireable ...
Although the round plates don't have as much maximum power handling as the rectangular or square plates of same type,
the focus is sound quality...

It's not so simple as round = better and not round = not so good.

Some of the best tubes have all planar construction. Think of the directly heated classics; 45, 47, 300B, even the 2A3.

Or, the most linear tube; 4P1L: all planar.

Then again, some of the most linear tubes also have all round construction (think especially directly heated transmitter tubes and also these small signal tubes you talk about here).

The most important thing for linearity is that the electrodes are the same form. The reason for this becomes obvious once you visualize what is going on inside the tube.

The round plates have a more "even distribution" between the cathode, grid and plate,
which improves the amplification quality and details of the tube...

What does this mean, I didn't understand?

Placing of the grid between cathode and anode determines mu of the tube. The closer the grid is to the cathode, the higher the mu.

But, the closer you put the grid to the cathode, the more dense the grid wires must be. Each wire emits an electric field that is stronger near the wire and weakens when going outwards.

If you look at the electric field made by the grid assembly standing on the cathode; the further away the grid is, the more uniform the field is.

The closer to you the grid is, you begin to see weaker spots in the electric field: between the grid wires!

This causes the effect in the curves where when you go left, the curves are wide apart (each grid wire has a small negative charge, close to 0V, between the wires there are openings with almost no charge, many electrons rush thru them) and going right they bunch up (the openings close because of strong negative field from each wire).

Solution? Make the grid more dense. This causes massive grid current (electrons accidentally hit grid wires and get trapped in the grid node) and capacitive parasitics (capacitance is determined by the area of the grid relative to the area of the cathode; more wires means more area). These bring all kinds of nasty problems.

Real solution? Frame grids. Not made of wire, but one single piece, usually needs to be solid gold. Super thin with a mesh like texture, so maximum electric field coverage with maximum routes for electron stream to physically pass the grid without hitting it.

Frame grids allow the grid to be placed super close to the cathode, making some pretty extreme tubes such as 6E5P and D3a and the like. They still eat up A LOT of grid current though.

Don't tell me, you already knew, lol

Indirectly heated small signal tubes usually have a round cathode, so as to heat it most effectively. I.e. the heater is about as close to all points of the cathode. This means for uniform geometry the plate should also be round.

The grid is not round though. Some transmitter tubes have round frame grid grids.

This is probably why the ultimate linear tubes such as 4P1L and 47 have all planar construction; each element, grid included, can be planar. Uniform.


So I guess, yes, I already knew. =) I'm being silly but I love tube geometry.

 

[MrCurwen]

Sorry if i interrupt your discution .... hard for me to follow. I'm a future owner of a mk6+ and I was looking for tube.

Not at all, it is me who is interrupting so no need for you to apologize. Let me try to help you.

I have ask the question on the mk6/mk8 thread, but someone told me to ask here for confirmation and explanation :
Can it be a probleme to not have closely matched 6SL7 driver tube ? Around 15-17 % of differences between the anode ?

What do you mean a problem? What are your goals?

It will work just fine, just not with best sound quality.

 

[gug42]

Hello. Thank you to take time. Well i'd like to know the impact.
I was thinking mostly about channel imbalance between left and right. But some ine says to me that the op will auto adjust.... Strang thing for me but why not ...


What did you mean by sound quality ? Please i need a little bit more explanation

Thx

 

[baronbeehive]

I have tested as many driver tubes as I could before I blew out my transformer heater coil output winding to get that opinion.
So just recently I went online to see if any correlation to this,
And I found out that my findings actually did correlate to what was known already in general about some 9pins and Octal pin tubes.
Specifically the tubes used for these amps driver stage...

I am finding out a right combination is best...
Like brighter driver tube will liven a thicker powerube down the line...
So its a balance .

There is another way...
If you really want to see if it is your opamp,
Just swap that specific one which controls your defective stage,
To the other channel on the same stage (positibe or negative stage)
Then try and see if the meters are different .

...................

Very nice work, I really like this experimetal approach.

You are right about the driver/power tube matching. My friend has a mix of RCA and Mullard EL34 power tubes which are quite thick, and JJ driver tubes, 12AX7's,which are very lively sounding. The combo works really well. When I replaced the JJ's with my favourite TS5751's, which are a bit more tubey it didn't work as well as on my other amp which has Amperex EL84 power tubes which are very clear sounding and the TS driver tube, and the JJ's were too SS sounding on mine.

I can check out the opamp as you suggest however I would have thought it would affect both right hand circuits not just one. I have not done anything yet because I'm still too busy and the amp is having to take second place for the moment.
Edit: Oh wait I didn't realise there was one opamp on each tube, sorry.

But coin is ready and waiting as soon as I can get around to it I think!!

Edit: Oh, and turtles are really great, just waiting to book my holiday in the Galapagos!

 

[baronbeehive]

Hello. Thank you to take time. Well i'd like to know the impact.
I was thinking mostly about channel imbalance between left and right. But some ine says to me that the op will auto adjust.... Strang thing for me but why not ...


What did you mean by sound quality ? Please i need a little bit more explanation

Thx

You needn't worry as MrCurwen says it will work fine, and will autobias within limits. You are correct in that the only effect you might notice is channel imbalance which will affect the sound quality by making one channel louder than the other. More important is for the power tubes to be more closely matched.

 

[Maxx134]

One of my most favourite subjects! Always on the lookout for new tube types.



Could be in your circuit, but in general; no, not at all. There is no inherent difference in linearity in the most commonly used op point area of these tubes. Possible differences in EQ and linearity come from poorly matching the resistor values to the tube rp, and from poor output impedance. 12AX7 does have a bit higher rp and so if it is used to drive a load (such as a capacitor with parasitics or a hungry tube grid) it will do worse. A decent buffer solves this.



Not in DIY circles, not at all. Where are you getting this?



Possibly. Not in practise in my experience. At RF these things matters, not so much at audio frequencies.

Also the plates of the two sections are closer together because of the narrow confines of smaller tube...



Yes these things do matter a bit, but the most significant difference with regards to oscillation is mu, which is distance of grid to cathode. Closer to cathode, higher mu, higher parasitics, higher coupling and chance of oscillation.



No. Guitar amps used 12AX7 because it was plentiful and cheap. It had become plentiful and cheap because of NFB. Early 1950's saw the onslaught of gNFB audio circuits, which necessitated a lot of gain. So, 12AX7 became hugely popular. Also it was hugely popular because it was the tube of choise for jukebox RIAA correctors, and jukeboxes became very common.



Ah, tube electrode geometry! One of my most favourite subjects!



It's not so simple as round = better and not round = not so good.

Some of the best tubes have all planar construction. Think of the directly heated classics; 45, 47, 300B, even the 2A3.

Or, the most linear tube; 4P1L: all planar.

Then again, some of the most linear tubes also have all round construction (think especially directly heated transmitter tubes and also these small signal tubes you talk about here).

The most important thing for linearity is that the electrodes are the same form. The reason for this becomes obvious once you visualize what is going on inside the tube.




What does this mean, I didn't understand?

Placing of the grid between cathode and anode determines mu of the tube. The closer the grid is to the cathode, the higher the mu.

But, the closer you put the grid to the cathode, the more dense the grid wires must be. Each wire emits an electric field that is stronger near the wire and weakens when going outwards.

If you look at the electric field made by the grid assembly standing on the cathode; the further away the grid is, the more uniform the field is.

The closer to you the grid is, you begin to see weaker spots in the electric field: between the grid wires!

This causes the effect in the curves where when you go left, the curves are wide apart (each grid wire has a small negative charge, close to 0V, between the wires there are openings with almost no charge, many electrons rush thru them) and going right they bunch up (the openings close because of strong negative field from each wire).

Solution? Make the grid more dense. This causes massive grid current (electrons accidentally hit grid wires and get trapped in the grid node) and capacitive parasitics (capacitance is determined by the area of the grid relative to the area of the cathode; more wires means more area). These bring all kinds of nasty problems.

Real solution? Frame grids. Not made of wire, but one single piece, usually needs to be solid gold. Super thin with a mesh like texture, so maximum electric field coverage with maximum routes for electron stream to physically pass the grid without hitting it.

Frame grids allow the grid to be placed super close to the cathode, making some pretty extreme tubes such as 6E5P and D3a and the like. They still eat up A LOT of grid current though.



Indirectly heated small signal tubes usually have a round cathode, so as to heat it most effectively. I.e. the heater is about as close to all points of the cathode. This means for uniform geometry the plate should also be round.

The grid is not round though. Some transmitter tubes have round frame grid grids.

This is probably why the ultimate linear tubes such as 4P1L and 47 have all planar construction; each element, grid included, can be planar. Uniform.


So I guess, yes, I already knew. =) I'm being silly but I love tube geometry.

Good stuff to digest, thanks for info...
: popcorn:

What's your Ia?

Yes I wrote it in my notes I will find much later I have to go for now.

 

[baronbeehive]

........ incidently I think that bit about the archbishop on page 1 should have been left in........... how can we mod without all the facts?

 

[gug42]

You needn't worry as MrCurwen says it will work fine, and will autobias within limits. You are correct in that the only effect you might notice is channel imbalance which will affect the sound quality by making one channel louder than the other. More important is for the power tubes to be more closely matched.

Thank you. Good news. Well 15% could be in thoses "limits" or not ?

I suppose channel imbalance could be compensed by equalizer or by my unbalanced ear (true by the way ) !

 

[baronbeehive]

Thank you. Good news. Well 15% could be in thoses "limits" or not ?

I suppose channel imbalance could be compensed by equalizer or by my unbalanced ear (true by the way ) !

If I remember right you're out by 5%, so not a problem!

 

[gug42]

Yes, generally admited that matched pair is under 10% of differences.
Thx

 

[MrCurwen]

If I recall correctly it's one envelope (one double triode) per tube, so that each tube section inside the envelope handles one phase of that channel.

That being the case, having a 15% mismatch in transconductance (that is how tubes are matched) between two triodes that share an envelope will not be affecting channel balance, meaning relative volume between channels. Not at all.

It will affect linearity of that channel, meaning more gm mismatch -> more distortion. It's mostly 2H but as discussed above the NFB then spreads the harmonics around a bit.

I get that gug42 you won't probably be understanding a lot of this stuff so let me make it clear; having a normal 10 to 15 % mismatch between triodes inside a double triode is not going to be the sound quality bottleneck of your amplifier. It is not good for the sound quality in objective terms, but some people even prefer mismatch and the distortion it causes. It's often called "tube sound". Whether or not this sounds good or bad to you, only you can tell. So try it out.


As an aside, the percentages are in fact gm, not mu. Gm does by proxy affect gain a little bit (depending on circuit), but not that much. Tubes are usually not matched by mu.

Most human ears are 100% ok with 10% or even more volume mismatch, they simply adjust to it. 30% or more, ok you might have a problem then.

 

[MrCurwen]

If I recall correctly it's one envelope (one double triode) per tube, so that each tube section inside the envelope handles one phase of that channel.

What I meant to write was:

If I recall correctly it's one envelope (one double triode) per channel, so that each tube section inside the envelope handles one phase of that channel.

 

[gug42]

Hello,

@MrCuren : Thank you for thoses explanations.
=> Internal mismatch can be more anoying than tube Left/Right mismatch


Another point : Can you point me to a fan model that fit in ? Need a 80x80x20
Noctua doesn't enter in

Thx !