Andyjevans said in Post # 10:
"dull in the extreme treble, noticeably on drumkit with snare rimshots and cymbals"
That got me thinking . . .
I often have noticed that cymbals do not sound correct in some of my systems, and others systems.
They sound slightly harsh, not like I remember they sound like when I am at a concert or a jazz group.
Is it the recording?
Is it the amplifier?
Is it the loudspeaker?
In my more recent sets of amplifiers and loudspeakers, I generally do not hear that harshness anymore.
Is it my hearing rolling off those higher frequencies?
Or, is it possible that my amplifiers are not producing lots of harmonic distortion and intermodulation distortion upper frequency products?
Would that lack of extra distortion make the sound as if the high frequencies are rolled off?
Is our perception just concluding that the higher frequency distortion products which are missing . . .
so we interpret the sound effect as if it is caused by high frequency roll off?
"dull in the extreme treble, noticeably on drumkit with snare rimshots and cymbals"
That got me thinking . . .
I often have noticed that cymbals do not sound correct in some of my systems, and others systems.
They sound slightly harsh, not like I remember they sound like when I am at a concert or a jazz group.
Is it the recording?
Is it the amplifier?
Is it the loudspeaker?
In my more recent sets of amplifiers and loudspeakers, I generally do not hear that harshness anymore.
Is it my hearing rolling off those higher frequencies?
Or, is it possible that my amplifiers are not producing lots of harmonic distortion and intermodulation distortion upper frequency products?
Would that lack of extra distortion make the sound as if the high frequencies are rolled off?
Is our perception just concluding that the higher frequency distortion products which are missing . . .
so we interpret the sound effect as if it is caused by high frequency roll off?
Last edited:
My hearing is a bit rolled off due to age but I can still tell differences in high frequency reproduction. As a pro bass player I spent many years right next to a drumkit so it's a very familiar sound, and not easy to reproduce. I'm a bit of a tube completist and have tried out innumerable tube and OPT comparisons. The 2a3 and 300b types are familiar and good, but only a gain of 4 so I've been using high gain output tubes so I can use DHT drivers like 10Y. The best, like EL12 in triode, don't sound harsh and have good treble so it can be done. I fully agree with higher primary impedance. Multiply by x4 at least. So a tube like EL12 in triode at 1.1K wants to see 5K or more. As usual, SE no global feedback.
andyjevans,
Push pull, balanced, and single ended:
As I said in a thread earlier this week, one of my design criteria is to pick larger primary impedances for my triode wired beam power tubes.
Most of my recent designs are push pull or balanced. But my later single ended designs used higher primary impedances too.
Push pull and balanced:
Sometimes I use good push pull transformers that have lots of inductance, and lots of laminations.
But I only drive them at much lower power than they are capable of.
The cheating I do is that I connect the loudspeaker to the next lower secondary tap. I put my 8 and 6 Ohm loudspeakers on the 4 Ohm tap.
The largeproary inductance, and the generous amount of laminations helps me to be able to do this without a large penalty.
Less distortion, and higher damping factor result. The tradeoff is lower power out, which I usually do not need.
Single ended:
Single ended output transformers might not have as good of an improvement when you put the loudspeaker on a lower output tap impedance.
That is because the primary inductance of air gapped output transformers is more likely to get in the way of low frequency response.
When you tap down, the secondary reflects a higher impedance to the primary, so it does not swamp out the primary inductance quite as well as connecting the loudspeaker to the correct output tap.
Just my $0.03
Push pull, balanced, and single ended:
As I said in a thread earlier this week, one of my design criteria is to pick larger primary impedances for my triode wired beam power tubes.
Most of my recent designs are push pull or balanced. But my later single ended designs used higher primary impedances too.
Push pull and balanced:
Sometimes I use good push pull transformers that have lots of inductance, and lots of laminations.
But I only drive them at much lower power than they are capable of.
The cheating I do is that I connect the loudspeaker to the next lower secondary tap. I put my 8 and 6 Ohm loudspeakers on the 4 Ohm tap.
The largeproary inductance, and the generous amount of laminations helps me to be able to do this without a large penalty.
Less distortion, and higher damping factor result. The tradeoff is lower power out, which I usually do not need.
Single ended:
Single ended output transformers might not have as good of an improvement when you put the loudspeaker on a lower output tap impedance.
That is because the primary inductance of air gapped output transformers is more likely to get in the way of low frequency response.
When you tap down, the secondary reflects a higher impedance to the primary, so it does not swamp out the primary inductance quite as well as connecting the loudspeaker to the correct output tap.
Just my $0.03
Last edited:
If the driver's internal resistance is well enough small, the load resistance can't determine basically the low frequency response extension.
Hi Andy,
My best sounding indirectly heated penthode is (are) PL519/ 509 using g2 as anode (anode not connected) auto-bias, Rk consisting of 390 ohm non decoupled in series with 680 ohm decoupled (220 mu + 0,1 mu) and Vb = 300v (Rg1 470k). It has a mug2g1= 3.5 so any driver used for a 300B or 2A3 can be used.
Some 20 yrs ago I wanted to simulate a mesh anode and thought may be a g2 could be used for that. This was succesful right away. I posted my schematic on this forum I believe 5 yrs ago. Jonathan Weiss of Oswalds Mill Audio has made a commercial version of this (a friend of mine had given him the schematic) and now it can be found through google under the heading Parallax Amplifier. I'm sure you won't regret trying this.
My best sounding indirectly heated penthode is (are) PL519/ 509 using g2 as anode (anode not connected) auto-bias, Rk consisting of 390 ohm non decoupled in series with 680 ohm decoupled (220 mu + 0,1 mu) and Vb = 300v (Rg1 470k). It has a mug2g1= 3.5 so any driver used for a 300B or 2A3 can be used.
Some 20 yrs ago I wanted to simulate a mesh anode and thought may be a g2 could be used for that. This was succesful right away. I posted my schematic on this forum I believe 5 yrs ago. Jonathan Weiss of Oswalds Mill Audio has made a commercial version of this (a friend of mine had given him the schematic) and now it can be found through google under the heading Parallax Amplifier. I'm sure you won't regret trying this.
I've been researching some posts on KT88 and 6550 in triode and the results were pretty negative. Examples....
6l6GC is generally low in 4th order distortion. Higher in the 2nd order. EL34 seems lower in 2nd and 3rd but higher in 4th. The KT88 in triode just seems generally higher in all orders.
I've tried KT88-98 and 6550C in triode with results too poor be explained by manufacturing variations - it just doesn't work, and never has!
I am not sure what the problem with triode connected 6550/KT88 comes from, but I just found them very "uninvolving" and slow sounding. Maybe the greatly increased miller capacitance was just too much for the wimpy driver stages I used at the time, both SE and PP.
6l6GC is generally low in 4th order distortion. Higher in the 2nd order. EL34 seems lower in 2nd and 3rd but higher in 4th. The KT88 in triode just seems generally higher in all orders.
I've tried KT88-98 and 6550C in triode with results too poor be explained by manufacturing variations - it just doesn't work, and never has!
I am not sure what the problem with triode connected 6550/KT88 comes from, but I just found them very "uninvolving" and slow sounding. Maybe the greatly increased miller capacitance was just too much for the wimpy driver stages I used at the time, both SE and PP.
This sounds plausible to me, wimpy driver stages is a classic way to ruin the sound from triodes.
A well documented phenomenon and I learned it myself the hard way some twenty years ago when I took the leap from TV tubes and junk transformers into my first "real" SET project, a 6B4G SET with Hammond 1627 OPTs. The first versions used 6SL7 and triode wired EF86s as input tubes, resulting in a quite dull sounding amp until I tried E280F and other high gm pentodes.
Nowadays I have developed a nasty habit of adding direct-coupled, high gm cathode followers between the gain stages and the output tubes.
Perhaps not the most elegant solution but it takes care of any problems related to input capacitance and even grid current.
Currently trying out a pair of Sovtek 6L6WXT. Not the 6P3S you liked - it has a staggered black base not a straight one. This sounds like the best yet. It's a lighter sound than the 6P3S-E but with good piano and tone, and doesn't seem to have any rolled off highs. Promising. All the 6L6 types I tried had better tone than the EL34 in triode. It looks just like the 6L6GB. My pair are older than the current ones.
Last edited:
FWIW I have found there to be such wide differences in sound quality between different tubes of the same tube type I’m not sure I could make a generalization about a certain tube type’s sound. I have been using a variety of 807 tubes in my amplifier in both triode and ultralinear. There’s no way to confuse a Cossor vt60a with a Sylvania 5933 despite supposedly being the same tube type. That also applies to the Russian made “Feranti”, Fivre, and French Mazda 807 as well. Let’s pile on supposed 807 analogs like the RCA 6bg6 and STC 5b/254m sounding different as well.
So yes, I have heard differences between running Sylvania 7027a (in a 6bg6ga bottle), Chinese el34, 6ar6, and Sylvania 6v6 as well but at this point I’m of the opinion that different tubes sound different because they are different tubes. I mean they are literally different. No two tube manufacturers made tubes the same way, even for the same tube type. And while the curves of the same tube type are supposedly close-ish it has never been clear to me how the curves look translates to how the tubes sound. Or if the curves can capture “the sound” at all. I try not to engage in audio woo too much but I wonder how two tubes made with different materials and different shapes could ever result in the same sound even if they look similar on a scope.
So yes, I have heard differences between running Sylvania 7027a (in a 6bg6ga bottle), Chinese el34, 6ar6, and Sylvania 6v6 as well but at this point I’m of the opinion that different tubes sound different because they are different tubes. I mean they are literally different. No two tube manufacturers made tubes the same way, even for the same tube type. And while the curves of the same tube type are supposedly close-ish it has never been clear to me how the curves look translates to how the tubes sound. Or if the curves can capture “the sound” at all. I try not to engage in audio woo too much but I wonder how two tubes made with different materials and different shapes could ever result in the same sound even if they look similar on a scope.
Indeed - tube inner structure is not surprisingly the major factor in sound differences, rather than the logo on the bottle. With 6SN7s I prefer all the angled plate versions whatever the name on the front is. This runs counter to tube folklore which differentiates between RCA, Sylvania etc.
So on to my last tube test - the Ruby labelled 6L6GCMSTR. This was similar in general sound to the 6L6WXT+ Sovtek, but a bit inferior across the board. Less clarity, slightly more wooden piano, slightly more shut in and a little coarser.
So out of my EL34 types and 6L6 types the best one was the 6L6WXT+ Sovtek. My pair are several years old, but still the current model available from several sources since they're used in stage amps. Price from Hot Rox is £35 each including tax in the UK. They are described there as ....
"Modelled after the vintage RCA 6L6GC "black plate," the Sovtek 6L6WXT+ features larger plate dimensions and improved grid structure for increased power handling capabilities. The 6L6WXT+ also features mica spacers with metal springs to eliminate tube rattle and microphonics. The Sovtek 6L6WXT+ yields a 20% higher output than the Sovtek 5881WXT and offers superior tone and overall performance to any 6L6 or KT66."
So higher power as well. Win win result.
So on to my last tube test - the Ruby labelled 6L6GCMSTR. This was similar in general sound to the 6L6WXT+ Sovtek, but a bit inferior across the board. Less clarity, slightly more wooden piano, slightly more shut in and a little coarser.
So out of my EL34 types and 6L6 types the best one was the 6L6WXT+ Sovtek. My pair are several years old, but still the current model available from several sources since they're used in stage amps. Price from Hot Rox is £35 each including tax in the UK. They are described there as ....
"Modelled after the vintage RCA 6L6GC "black plate," the Sovtek 6L6WXT+ features larger plate dimensions and improved grid structure for increased power handling capabilities. The 6L6WXT+ also features mica spacers with metal springs to eliminate tube rattle and microphonics. The Sovtek 6L6WXT+ yields a 20% higher output than the Sovtek 5881WXT and offers superior tone and overall performance to any 6L6 or KT66."
So higher power as well. Win win result.
"The Sovtek 6L6WXT+ yields a 20% higher output than the Sovtek 5881WXT"
Be very careful with these guys: I've got a bunch, and they belong to a very different species.
A normal 6L6 at Va = Vg2 = 250V requires -14V at the grid to yield Ia = 72mA and Gm = 6mA/V
The 6L6WXT+ needs -16V to yield Ia = 82mA and Gm = 7.8mA/V! (average of 16 pcs.)
You might get away with self-bias, but expect surprises with fixed bias!
IOW, re-biasing is mandatory.
Be very careful with these guys: I've got a bunch, and they belong to a very different species.
A normal 6L6 at Va = Vg2 = 250V requires -14V at the grid to yield Ia = 72mA and Gm = 6mA/V
The 6L6WXT+ needs -16V to yield Ia = 82mA and Gm = 7.8mA/V! (average of 16 pcs.)
You might get away with self-bias, but expect surprises with fixed bias!
IOW, re-biasing is mandatory.
Last edited:
Funny how different people are, I find the angled plates clearly inferior to various T plates and ladder plates in 6sn7 tubes.
I do think that as harsh as I was on curve gazing they are still important to figure out the general operating points of the tube. Over on head-Fi you’ll see people waxing poetically over how amazing certain high power tubes like 6550, kt88, or el34 sound despite running them at 250v and using milliwatt output power. If you don’t need the power and don’t need to run at 400+ volts there are a huge number of “good sounding” tubes that are relatively inexpensive. Metal 6l6 and 6v6 come to mind as well as the 6aq5. I tend to run my 807 push pull pretty conservatively in triode, 320v 40ma. Sounds great and is sufficient power for what I need. Haven’t felt compelled to pursue the more expensive and higher power tubes like the kt88 or even el34.
I do think that as harsh as I was on curve gazing they are still important to figure out the general operating points of the tube. Over on head-Fi you’ll see people waxing poetically over how amazing certain high power tubes like 6550, kt88, or el34 sound despite running them at 250v and using milliwatt output power. If you don’t need the power and don’t need to run at 400+ volts there are a huge number of “good sounding” tubes that are relatively inexpensive. Metal 6l6 and 6v6 come to mind as well as the 6aq5. I tend to run my 807 push pull pretty conservatively in triode, 320v 40ma. Sounds great and is sufficient power for what I need. Haven’t felt compelled to pursue the more expensive and higher power tubes like the kt88 or even el34.
Of course, it is elegant! Another huge advantage is good protection against thermal runaway, which caused by grid emission.
neodymium,
1. You said:
"If the driver's internal resistance is well enough small, the load resistance can't determine basically the low frequency response extension."
I Agree, but only as far as the generalization is true.
Quality output transformers work better for this generalization, Yes!
Poor quality output transformers suffer, and disprove the generalization.
Example of some poor quality output transformers:
Drive the output transformer primary with a solid state amplifier, output impedance < 1 Ohm. So far, so good, Right?
But, the low quality output transformer I purposely selected, has High DCR, and Low primary Inductance.
The high DCR is in Series with the low primary inductance.
No matter how low the solid state amplifier impedance is, the dominant low frequency pole is according to the inductive reactance, Xl, and DCR.
It is even worse for output tubes (not very many have < 1 Ohm plate impedances).
There is one saving factor, even with poor quality output transformers.
Consider most loudspeakers in the 10's of Hertz, below the impedance peak(s), both for ported and closed box.
Their impedance below those peaks, the woofer impedance is the same as its DCR. An "8 Ohm" rated speaker has a DCR of between 4 to 6 Ohms.
The good thing is that low impedance (the same as DCR), reflects back to the primary at all extremely low frequencies, but not at DC, and therefore the speaker DCR does reflect to the primary, and swamps out the primary inductance.
Read those loudspeaker reviews, and look at the impedance versus frequency graph (below the lowest woofer impedance peak, according to the cabinet).
Then get your DMM out, and measure the DCR of your speakers.
Know Thy Speaker.
2. Thanks for reminding us about thermal runaway!
Grid emission not only can cause thermal runaway; but even before that . . . it causes the driver to have to drive a lower impedance grid.
Could that cause driver distortion?
Depends on the lower grid impedance grid, and the particular driver that it is connected to.
Tube amplifiers require us to pay attention to details.
Have Fun!
Regardless of the output transformers.
1. You said:
"If the driver's internal resistance is well enough small, the load resistance can't determine basically the low frequency response extension."
I Agree, but only as far as the generalization is true.
Quality output transformers work better for this generalization, Yes!
Poor quality output transformers suffer, and disprove the generalization.
Example of some poor quality output transformers:
Drive the output transformer primary with a solid state amplifier, output impedance < 1 Ohm. So far, so good, Right?
But, the low quality output transformer I purposely selected, has High DCR, and Low primary Inductance.
The high DCR is in Series with the low primary inductance.
No matter how low the solid state amplifier impedance is, the dominant low frequency pole is according to the inductive reactance, Xl, and DCR.
It is even worse for output tubes (not very many have < 1 Ohm plate impedances).
There is one saving factor, even with poor quality output transformers.
Consider most loudspeakers in the 10's of Hertz, below the impedance peak(s), both for ported and closed box.
Their impedance below those peaks, the woofer impedance is the same as its DCR. An "8 Ohm" rated speaker has a DCR of between 4 to 6 Ohms.
The good thing is that low impedance (the same as DCR), reflects back to the primary at all extremely low frequencies, but not at DC, and therefore the speaker DCR does reflect to the primary, and swamps out the primary inductance.
Read those loudspeaker reviews, and look at the impedance versus frequency graph (below the lowest woofer impedance peak, according to the cabinet).
Then get your DMM out, and measure the DCR of your speakers.
Know Thy Speaker.
2. Thanks for reminding us about thermal runaway!
Grid emission not only can cause thermal runaway; but even before that . . . it causes the driver to have to drive a lower impedance grid.
Could that cause driver distortion?
Depends on the lower grid impedance grid, and the particular driver that it is connected to.
Tube amplifiers require us to pay attention to details.
Have Fun!
Regardless of the output transformers.
Last edited:
Somewhere between 10 and 15 years ago Antique Electronics Supply was selling these with the "bait" that "they couldn't blow them up no matter how hard they tried," so I took the bait. These poor guys have "seen the light" lots of times, and after lots of abuse, they are STILL ALIVE! Seen here at about 40 watts of dissipation. The coke bottle shaped tubes were replaced with some straight sided tubes that WERE far too easy to blow up. I bought a box full of "NOS" 6BQ6GA's from AES for under $1 each. A few were not NOS, so I took the crustiest oldies and stuffed them in an SSE and decided to see how much power I could extract from them. Let's just say that you can get 20 WPC from a 6BQ6GA, but you pay for it in lifetime. These lasted about 20 minutes before the THD started climbing, and a few more minutes before my bench power supply started complaining and smelling funny, so I killed the experiment. The tubes didn't seem to work very well the next day. I wonder why?
As for the sound of the different tubes mentioned, all of them can be dropped into a Tubelab SSE with no changes, but the big boys (KT88 and 6550) will benefit from some more current which was afforded by a resistor on alligator clips connected across the existing cathode resistor. For chick singer and a guitar type of "simple" music the EL34 in triode was the clear winner. The big boys in UL had the edge when trying to rattle the neighbor's windows with DSOTM cranked to the edge of clipping into some 15 inch Hawthorne Silver Iris OB speakers. Cathode feedback was used without any other feedback in both cases. The coke bottle 6L6GC's were OK sounding in an SSE, but screamed pretty good in a cranked guitar amp, so that's where most of them went.
For daily listening I tended to swap tubes whenever I got bored with whatever was in the amp. I think that a pair of old Electro Harmonix 6550WA's got the most playing time.
Attachments
A mosfet (a 'la Tubelabs Powerdrive) would do a better job than a tube but I like the delayed startup from IDH tubes, plus the fact that they don't fail as shorts like Mosfets do.
With a little trickery, cathode followers can be used to bias the output tubes into complete cutoff until they are warm and ready to draw plate current.
Probably not much of an issue with 6L6-like tubes, but handy for DHTs.