The 6c5 is a 6sj7 wired as a triode at the factory. That means it has the screen grid and shield of a pentode.
In the end we are still having to wrestle with what “sounds better” actually means. Earlier in the thread someone mentioned that they prefer the sound of the round plate Sylvania 6j5gt to others. Well, I prefer the Raytheon 6j5wgt with flat plates to the round plate Sylvanias. I also prefer the Sylvania T plate 6sn7w to the round plate Tungsol 6sn7. Is that a legitimate refutation of the whole premise? Until there is some sort of agreement on what “sounds better” means I think it does. There isn’t a single data sheet in existence that mentions sound quality. And none of the engineers involved in designing and manufacturing the tubes ever used it as a guide. How could they? “Sound quality” has everything to do with the circuit involved and the transducer(s) at the end of the chain. Would you really want the engineers in the 30s and 40s design in sound quality based on the speakers they had back then?
No, if there is an advantage to cylindrical plates there has to be some sort of mathematical, theoretical, or even manufacturing advantage that can be quantified and spelled out. Clearly it is not a necessary condition to get good sound. That alone should tell us that there isn’t any particular advantage to them.
No, if there is an advantage to cylindrical plates there has to be some sort of mathematical, theoretical, or even manufacturing advantage that can be quantified and spelled out. Clearly it is not a necessary condition to get good sound. That alone should tell us that there isn’t any particular advantage to them.
Data sheets don't give any description of sound as has been said, all you have is the curves and the data. But for the same data, users may well have preferences for the actual tubes they try out in their equipment. And it isn't just round anodes, it's mesh anodes, top caps and various other factors. Tube construction is fascinating, and there are successes like the 45 and the 27, 37, 56, 78 types, a lot of very average 9 pin tubes and some dogs. Not surprising considering the huge number of different tubes that were manufactured over 100 years or so.
All tubes are essentially parallel tubes. Many of you did not think of it that way.
There are many parallel electron paths that lead to a finite area plate (not a plate of zero area).
Many electrons flow according to the shortest path, but are also "steered" according to the angle of the largest field field strength that the electron "sees".
One question is . . . are all parallel paths the same length?
That depends on the geometry of the filament/cathode, grid, and the plate.
Grids are essentially uni-potential (same voltage all along the total grid structure).
Plates are essentially uni-potential (same voltage all along the total plate structure).
Indirectly heated cathodes are essentially uni-potential (same voltage all along the cathode structure).
Directly heated cathodes (Called Filaments) [no extra cathode element] are essentially Not uni-potential (whether DC or AC powered; AC changes the potential distribution for each Alternation [1/2 cycle] of the AC).
It is all just Physics.
Study fields, study electron paths, etc.
If you think the engineers of the 20s and 30s did not do that . . . well they are all dead, so you can not ask them.
But I am willing to go to Las Vegas and bet my fortune for those early vacuum tube engineers.
There are many parallel electron paths that lead to a finite area plate (not a plate of zero area).
Many electrons flow according to the shortest path, but are also "steered" according to the angle of the largest field field strength that the electron "sees".
One question is . . . are all parallel paths the same length?
That depends on the geometry of the filament/cathode, grid, and the plate.
Grids are essentially uni-potential (same voltage all along the total grid structure).
Plates are essentially uni-potential (same voltage all along the total plate structure).
Indirectly heated cathodes are essentially uni-potential (same voltage all along the cathode structure).
Directly heated cathodes (Called Filaments) [no extra cathode element] are essentially Not uni-potential (whether DC or AC powered; AC changes the potential distribution for each Alternation [1/2 cycle] of the AC).
It is all just Physics.
Study fields, study electron paths, etc.
If you think the engineers of the 20s and 30s did not do that . . . well they are all dead, so you can not ask them.
But I am willing to go to Las Vegas and bet my fortune for those early vacuum tube engineers.
Indeed, the 6F6S has round plate but has more gm in the same conditions ( like 3mA/V vs 2) and uprated specs: 410V plate voltage, 315V g2 voltage, 12W plate dissipation and 4W g2 dissipation. Can squeeze more power, especially in triode mode with higher plate voltage. Something like 2.5-3W for triode in class A1 and 5-6W for pentode.
I have just had a look and there are places where it's still relatively cheap at $13-15 each. But I have never heard of them before.
I have just had a look and there are places where it's still relatively cheap at $13-15 each. But I have never heard of them before.
Last edited:
happened lots of times, take the NEC 8045, to me it looked like a beam tetrode inside but factory wires as a triode....
Attachments
how to define? based on listening to a few bars of music? i had some audiophile friends come over for a tube rolling session, boy, when i replaced a tube and we played it, they were clapping after only a few bars!!! how quick to judge.....
Maybe the 6C5 is the 6J7 wired as a triode at the factory? Peter Lankshear's article attached for reference (and enjoyment).
That.
In any non round (actually non cylindrical) non concentric electrón paths from cathode to different points on plate won't be the same, flight time won't be the same.
Is it significant?
Given electron speeds and path length it will mean something at multi MHz frequencies.
Not the case for any Audio related frequency of course.
True for any Audio frequency.
Maybe significant at multi MHz frequencies where path differences become comparable to wavelengths involved, not within the Audio realm.
For filament heating frequencies? (Including DC)? : sure enough.
For Audio frequencies passing through those cathodes?
Think again.
AMEN BROTHER!..👍🏻
Same here +1000
Build a 2 sided cathode that has 2 square flat surfaces, with the filament between them. Connect the cathode sides together.
Build 2 frame grids, square with the same area as the cathode surfaces. Connect the two grids together.
Build 2 plates, square with the same area as the cathode surfaces, and therefore it is also the same area as the frame grid surface areas. Connect the plates together.
Now, build a "sandwich" parallel tube:
plate, grid, cathode, grid, plate.
Make the grids to plates spacing equal.
Make the cathodes to grids spacing equal.
Keep all planar surfaces parallel.
The electron path lengths from the cathodes will tend to be equal (shortest path is dominant).
This is a near perfect model of two parallel near perfect tubes.
Transit time of the electrons is not the issue, But cathode to grid spacing, and grid to plate spacing (equal potyiond of the path lengths are the issue:
Now, make the parallel tubes less perfect:
Space cathode 1 to grid 1 so it is 1.1 times the space from cathode 2 to grid 2.
Space on grid 1 to plate 1 so it is 0.9 times the space from grid 2 to plate 2.
Each half of the tube has equal Total electron path length (1.1 + 0.9), versus (1.0 + 1.0).
But the transconductance Gm, the mu (u), and the plate resistance rp, of the two halves of the tube are All different.
This is the im-perfect parallel tube model.
If you can follow the above, you had to work hard at it; but that means you now get my points in my earlier posts in this thread.
Thanks!
Build 2 frame grids, square with the same area as the cathode surfaces. Connect the two grids together.
Build 2 plates, square with the same area as the cathode surfaces, and therefore it is also the same area as the frame grid surface areas. Connect the plates together.
Now, build a "sandwich" parallel tube:
plate, grid, cathode, grid, plate.
Make the grids to plates spacing equal.
Make the cathodes to grids spacing equal.
Keep all planar surfaces parallel.
The electron path lengths from the cathodes will tend to be equal (shortest path is dominant).
This is a near perfect model of two parallel near perfect tubes.
Transit time of the electrons is not the issue, But cathode to grid spacing, and grid to plate spacing (equal potyiond of the path lengths are the issue:
Now, make the parallel tubes less perfect:
Space cathode 1 to grid 1 so it is 1.1 times the space from cathode 2 to grid 2.
Space on grid 1 to plate 1 so it is 0.9 times the space from grid 2 to plate 2.
Each half of the tube has equal Total electron path length (1.1 + 0.9), versus (1.0 + 1.0).
But the transconductance Gm, the mu (u), and the plate resistance rp, of the two halves of the tube are All different.
This is the im-perfect parallel tube model.
If you can follow the above, you had to work hard at it; but that means you now get my points in my earlier posts in this thread.
Thanks!
Heating has little to do with the potential, weather direct or indirect all that matters is the temperature of the filament/heater which is INDEPENDENT of AC or DC. The condition being that for long enough, typically some centimetres, the filament/heater temperature is the SAME along it. Uniform temperature gives rise to a uniform electron cloud. Then comes in the rest of the structure. Read the thread about AC/DC heating with scientific publication debunking myths and wrong thinking.
Last edited:
Equal path lengths will not have the same electron velocity, if the voltages are different.
Current is proportional to velocity.
The majority of electron path lengths from a 300B filament to the plate are equal.
But if one end of the filament is 0VDC, and the other end of the filament is +5VDC, the path from the 0VDC end of the filament will have higher velocity electrons, versus the electrons from the +5VDC end of the filament.
Higher velocity electrons means more plate current.
The 0VDC end of the filament has more plate current than the plate current from the +5VDC end of the filament.
Just imagine vertically sawing the 300B plate into two plates in the middle of the original plate structure.
The two Half Plates, would have unequal Gm and would have unequal plate resistance, rp.
The transconductance is greater at the 0VDC end of the filament, and transconductance is lower at the +5VDC end of the filament.
Likewise, the plate resistance is lower at the area where the electrons are coming from the 0VDC end of the filament, and the plate resistance is higher at the area where the electrons are coming from the +5VDC end of the filament.
With AC powered filaments, there are 3 extrema conditions:
1. One end is +2.5VAC (+3.5V peak), while the other end is -2.5VAC (-3.5V peak). Think of those two halves of the sawn up plate structure.
2. One end is -2.5VAC (-3.5V peak), while the other end is +2.5VAC (+3.5V peak). Think of those two halves of the sawn up plate structure.
3. The condition most people do not think about, is when from one end of the filament to the other end of the filament, it is All 0VAC and 0V peak (twice per cycle).
At the crest of the AC powered filaments, there is 7V peak from one end of the filament to the other.
Think what this might cause (At different filament to grid voltages, Gm and rp do vary, look at the tube data graphs).
Current is proportional to velocity.
The majority of electron path lengths from a 300B filament to the plate are equal.
But if one end of the filament is 0VDC, and the other end of the filament is +5VDC, the path from the 0VDC end of the filament will have higher velocity electrons, versus the electrons from the +5VDC end of the filament.
Higher velocity electrons means more plate current.
The 0VDC end of the filament has more plate current than the plate current from the +5VDC end of the filament.
Just imagine vertically sawing the 300B plate into two plates in the middle of the original plate structure.
The two Half Plates, would have unequal Gm and would have unequal plate resistance, rp.
The transconductance is greater at the 0VDC end of the filament, and transconductance is lower at the +5VDC end of the filament.
Likewise, the plate resistance is lower at the area where the electrons are coming from the 0VDC end of the filament, and the plate resistance is higher at the area where the electrons are coming from the +5VDC end of the filament.
With AC powered filaments, there are 3 extrema conditions:
1. One end is +2.5VAC (+3.5V peak), while the other end is -2.5VAC (-3.5V peak). Think of those two halves of the sawn up plate structure.
2. One end is -2.5VAC (-3.5V peak), while the other end is +2.5VAC (+3.5V peak). Think of those two halves of the sawn up plate structure.
3. The condition most people do not think about, is when from one end of the filament to the other end of the filament, it is All 0VAC and 0V peak (twice per cycle).
At the crest of the AC powered filaments, there is 7V peak from one end of the filament to the other.
Think what this might cause (At different filament to grid voltages, Gm and rp do vary, look at the tube data graphs).
Last edited:
Most electronic components such as tubes, transistors, many resistors and more are designed and manufactured without consideration of current (more complex than just closing or opening a circuit) and materials and resonances. This is audible.
Shape has influence.
By the way: AC or DC: the signal is audibly affected by the slightest voltage characteristics.
Shape has influence.
By the way: AC or DC: the signal is audibly affected by the slightest voltage characteristics.
@6A3sUMMER
If one end of the filament is 0V and the other is 5V has NOTHING to do with electron velocity because the electron cloud screens the filament/cathode from the grid and the plate.
As already said, AC or DC does not make any difference. The voltage gradient across the filament has nothing to do with emission which only depends on the temperature. Temperature is uninform if the filament/heather is long enough. Being slightly lower at both ends where normally the filament/heather is outside the structure. What matters most has been missing so far: the actual material and it's real quality/structure/uniform doping.
See my post #65 here:
https://www.diyaudio.com/community/...-to-heat-filaments-high-freq-ac.411444/page-4
Reference: https://www.sciencedirect.com/science/article/abs/pii/S0042207X9700170X
If one end of the filament is 0V and the other is 5V has NOTHING to do with electron velocity because the electron cloud screens the filament/cathode from the grid and the plate.
As already said, AC or DC does not make any difference. The voltage gradient across the filament has nothing to do with emission which only depends on the temperature. Temperature is uninform if the filament/heather is long enough. Being slightly lower at both ends where normally the filament/heather is outside the structure. What matters most has been missing so far: the actual material and it's real quality/structure/uniform doping.
See my post #65 here:
https://www.diyaudio.com/community/...-to-heat-filaments-high-freq-ac.411444/page-4
Reference: https://www.sciencedirect.com/science/article/abs/pii/S0042207X9700170X
Last edited:
AC powered DHT filaments:
Improperly adjusted hum pot causes hum.
50Hz power mains causes 50Hz hum when the pot is un-balanced.
60Hz power mains causes 60Hz hum when the pot is un-balanced.
Then the hum pot is properly adjusted and the 50Hz or 60Hz goes away.
But now, looking at the spectrum of a test tone, 1kHz for example:
There are residual lower and upper sidebands:
50Hz power mains: 100Hz sidebands; 900Hz and 1100Hz; spaced 100Hz from the 1kHz test tone.
60Hz power mains: 120Hz sidebands; 880Hz and 1120Hz; spaced 120Hz from the 1kHz test tone.
And, the sidebands are there, even if there is zero ripple on the B+ (keep reducing the B+ ripple, and the amplitude of the sidebands do not change)..
How to explain those 100Hz or 120Hz sidebands?
Perhaps the filaments are vibrating according to the alternations (alternations = 2 x power mains frequency).
Filaments that are magnetically attracted to the plate? (If the filament is closer to one side of the plate than to the other side of the plate)?
I am just asking for an explanation of those 2 x power mains frequency sidebands.
Definitely not the hum pots mis-adjustment.
DC powered filaments:
If the electron cloud keeps the grid and plate from "seeing" the filament;
And if there is no electron velocity difference as a result (higher velocity would cause higher current);
Then with fixed - bias grid voltage, why does the 300B plate current increase when the +V end of the filament is grounded;
Versus the lower current when the negative -V end of the filament is grounded?
Difference in velocity, difference in current.
No difference in velocity, no difference in current.
Improperly adjusted hum pot causes hum.
50Hz power mains causes 50Hz hum when the pot is un-balanced.
60Hz power mains causes 60Hz hum when the pot is un-balanced.
Then the hum pot is properly adjusted and the 50Hz or 60Hz goes away.
But now, looking at the spectrum of a test tone, 1kHz for example:
There are residual lower and upper sidebands:
50Hz power mains: 100Hz sidebands; 900Hz and 1100Hz; spaced 100Hz from the 1kHz test tone.
60Hz power mains: 120Hz sidebands; 880Hz and 1120Hz; spaced 120Hz from the 1kHz test tone.
And, the sidebands are there, even if there is zero ripple on the B+ (keep reducing the B+ ripple, and the amplitude of the sidebands do not change)..
How to explain those 100Hz or 120Hz sidebands?
Perhaps the filaments are vibrating according to the alternations (alternations = 2 x power mains frequency).
Filaments that are magnetically attracted to the plate? (If the filament is closer to one side of the plate than to the other side of the plate)?
I am just asking for an explanation of those 2 x power mains frequency sidebands.
Definitely not the hum pots mis-adjustment.
DC powered filaments:
If the electron cloud keeps the grid and plate from "seeing" the filament;
And if there is no electron velocity difference as a result (higher velocity would cause higher current);
Then with fixed - bias grid voltage, why does the 300B plate current increase when the +V end of the filament is grounded;
Versus the lower current when the negative -V end of the filament is grounded?
Difference in velocity, difference in current.
No difference in velocity, no difference in current.
Last edited:
While the current out of the terminals can be changed with the different filament connection, the bias voltage at every point along the filament is effectively the same. Otherwise you would get different anode curves with DC, center-tap and/or AC etc.... Instead they are all the same provided the reference for the bias is the correct one. If there is any imbalance of current on the plate is really neglectible.
The IMD has to do with iteraction of the signal with supply. If the filament sees an "infinite" impedance looking into the supply the side-bands disappear. Instead the AC supply (and non-current regualted DC supply) only reduce the hum but cannot get rid of sidebands because the anode curves are not symmetric at all whith large signal. @Rod Coleman can explain this better.
The IMD has to do with iteraction of the signal with supply. If the filament sees an "infinite" impedance looking into the supply the side-bands disappear. Instead the AC supply (and non-current regualted DC supply) only reduce the hum but cannot get rid of sidebands because the anode curves are not symmetric at all whith large signal. @Rod Coleman can explain this better.
Last edited: