Solid state/class d vs tube amps

[-Asdfgh-]

I can't discuss specific "tricks" or approaches behind any of the solid state and/or hybrid technologies I might be doing research on for obvious reasons.

Of course!

The numerical modelling side is of interest to me, as I have worked as a research software engineer in the past, although mostly neural networks and related technologies. We did joke about modelling amps with neural nets about 20 years ago, but didn't do anything about it. That wouldn't have allowed for rule extraction that could have informed the analogue side, though.

 

[DavidEdenAria]

In fact, the glass to metal seal is mainly a ceramic problem; but, we can just generally lump it into being a Materials Science problem.

Regarding NOS vs new production tubes, I am rather more inclined to believe what I have heard from both Ron Wickersham and Andy Field (@agedhorse) regarding new production tube reliability than Mike at KCA NOS, who apparently makes his living by selling NOS tubes.

Nothing against NOS tubes. I’ve messed with a whole slew of 5751’s getting sounds I like. But, let’s be honest about where the differences are and aren’t.

You are not entirely correct regarding Mike at KCA......he sells by far mostly 'new production' as the supply of NOS and strong old production pulls dwindles down to extremely low levels.

He has always sold both old and new production....if you want to discount the experience of 'the gear page' with both types for the last near 20 years, thats fine.

I mentioned "Watford Valves" as well....and many others that are smaller operations than the corporate giants....Watford has sold 100,000's of NOS and new production for decades with direct feedback from techs and players.

Let me mention once again, great strides have been taken in overall tube quality in the last few years....when i started collection NOS tubes, much of new production was not trustworthy with out full screening and burn in, for power tubes mainly.

Ken Fischer himself used to wack RFT/Siemens EL34 with a rubber mallet while idling to weed out the flimsy ones for example.

 

[-Asdfgh-]

Ken Fischer himself used to wack RFT/Siemens EL34 with a rubber mallet while idling to weed out the flimsy ones for example.

Rubber is probably an important word there.

 

[Drgonzonm]

In fact, the glass to metal seal is mainly a ceramic problem; but, we can just generally lump it into being a Materials Science problem.

The technology to address thermal expansion coefficient differences is 70 years old. Raytheon. Rca's western electric and others have published the technology. If the current manufacturers chose not to use the technology is a different issue.
I'll make a bet mesa boogie and other high end amp builders test the tube lots to assure QC. I am also guessing they burn in tubes before shipping amps. I've been there working qc.

 

[agedhorse]

Let me mention once again, great strides have been taken in overall tube quality in the last few years....when i started collection NOS tubes, much of new production was not trustworthy with out full screening and burn in, for power tubes mainly.

Who in their right mind wouldn't screen and burn in ANY tube, at least until developing their own statistics on that particular tube? Then, at least a sampling screen-burn-screen to validate that the sample conforms to the statistics, if not 100% burn and test.

There's so much talk about things here without considering the compete data, without considering that some of the companies at least have very sophisticated, thorough programs for determining how a sample tube actually conforms to how the tube is SUPPOSED to perform. If one were to look at some of these arguments that tube resellers (in general) make as part of their marketing, you would think we were all back-woods, ignorant, dumb as a rock, gullible folks. Clearly their business model works, not sure what that says but it's not very good IMO.

There are in fact design/manufacturing engineers that do study in great depth these specific tube aspects, and know FAR more than most of the resellers claiming to have tubes with the most secret mojo, tubes with a pedigree that would fit into a mystery/spy novel, etc.

To let you guys into a little secret, some of us do in fact have a very nice selection of NOS or aged OS tubes that might make some of you jealous. Why you might ask? Because it's a good way to benchmark tubes against these old stock tubes to gather DATA on what the differences might be. The challenge is many of the NOS tubes vary as greatly as the new production tubes so the benchmarks become specific tubes rather than tubes with the same manufacturer/model. Marketing any product that is a moving target against another that is also a moving target makes it almost impossible to guarantee exact performance. When you add some of the flowery descriptors, much like the high end wine industry, it becomes a very slippery slope to draw any meaningful conclusions from.

 

[bucephylus]

The technology to address thermal expansion coefficient differences is 70 years old. Raytheon. Rca's western electric and others have published the technology. If the current manufacturers chose not to use the technology is a different issue.
I'll make a bet mesa boogie and other high end amp builders test the tube lots to assure QC. I am also guessing they burn in tubes before shipping amps. I've been there working qc.

All I said was that engineering the glass to metal interface is primarily a Materials Science issue as opposed to simply being metallurgical.

Are you wanting to argue that point? Not sure I understood what you are criticizing.

 

[Drgonzonm]

eeAll I said was that engineering the glass to metal interface is primarily a Materials Science issue as opposed to simply being metallurgical.

Are you wanting to argue that point? Not sure I understood what you are criticizing.

From my point of view given the technology the problem was metallurgical. Material science as subject is only a three or four decades old, unless you went to MIT or Berkley. Metallurgists have had to deal with ceramics, glass since the dawn of the bronze age as well the common and less common metals. Just look at the book written by Agricola.

 

[bucephylus]

From my point of view given the technology the problem was metallurgical. Material science as subject is only a three or four decades old, unless you went to MIT or Berkley. Metallurgists have had to deal with ceramics, glass since the dawn of the bronze age as well the common and less common metals. Just look at the book written by Agricola.

 

[DavidEdenAria]

Who in their right mind wouldn't screen and burn in ANY tube, at least until developing their own statistics on that particular tube? Then, at least a sampling screen-burn-screen to validate that the sample conforms to the statistics, if not 100% burn and test.

There's so much talk about things here without considering the compete data, without considering that some of the companies at least have very sophisticated, thorough programs for determining how a sample tube actually conforms to how the tube is SUPPOSED to perform. If one were to look at some of these arguments that tube resellers (in general) make as part of their marketing, you would think we were all back-woods, ignorant, dumb as a rock, gullible folks. Clearly their business model works, not sure what that says but it's not very good IMO.

There are in fact design/manufacturing engineers that do study in great depth these specific tube aspects, and know FAR more than most of the resellers claiming to have tubes with the most secret mojo, tubes with a pedigree that would fit into a mystery/spy novel, etc.

To let you guys into a little secret, some of us do in fact have a very nice selection of NOS or aged OS tubes that might make some of you jealous. Why you might ask? Because it's a good way to benchmark tubes against these old stock tubes to gather DATA on what the differences might be. The challenge is many of the NOS tubes vary as greatly as the new production tubes so the benchmarks become specific tubes rather than tubes with the same manufacturer/model. Marketing any product that is a moving target against another that is also a moving target makes it almost impossible to guarantee exact performance. When you add some of the flowery descriptors, much like the high end wine industry, it becomes a very slippery slope to draw any meaningful conclusions from.

Ive got dozens of NOS Mesa STR 6L6GC and 6CA7 all which are Sylvania/Phillips as well and yes they are a benchmark/standard....i am saying there are successful tube vendors for many many years that are not giant corporations...Angela Instruments is another.....OTOH, there are LOTS of vendors on Ebay misrepresenting tubes, let alone not doing any kind of thorough testing.

My power tubes go thru a Hickok 533a and a BK 707 first for mutual conductance, shorts, the 533a will test for noise as well.... and grid emission in a BK 600 and i do realize these testers are not at full operating volts but all were highly regarded in the field back in the day...i then use a Bias Rite to track current draw in a live amp for at least 30 minutes....this method has weeded out tubes with slowly (or occasionally fast) drifting current draw.

 

[bucephylus]

From my point of view given the technology the problem was metallurgical. Material science as subject is only a three or four decades old, unless you went to MIT or Berkley. Metallurgists have had to deal with ceramics, glass since the dawn of the bronze age as well the common and less common metals. Just look at the book written by Agricola.

Agricola? This guy?
Invalid Link Removed
Seriously? I hope not. Maybe I’m just missing humor here or something.

 

[Drgonzonm]

Agricola? This guy?
Invalid Link Removed
Seriously? I hope not. Maybe I’m just missing humor here or something.

No intent in humor, it was the best text on mining and metallurgy for 200 years. Basically, Old George describes that Metallurgy is a science and art, and should be left up to the professionals. Even describes making glass. He was a little early on making tube for amplifiers.
I have a few updated books, but it fun to reference the books that were produced to teach in the Jurassic Age.

 

[GreyBeard]

All this tube discussion makes me glad I can be satisfied with my class D pf-500. I plug it in, turn it on, and play.

 

[JimmyM]

All this tube discussion makes me glad I can be satisfied with my class D pf-500. I plug it in, turn it on, and play.

I do that with all my amps.

 

[Drgonzonm]

Agedhorse describes the current state of tube amp design very well. There are companies like the one he works for, that has better data than that published by say RCA, Western Electric, GE. The rest of us, have to rely on tech sheets that were developed 50-60 years ago. If we want a guitar amp, we push the tubes outside the center design values, but stay within published max values. Some tubes will take higher voltages than most of us are willing to design. We buy off the shelf. Bad mouthing vendors because the tube we got doesn't meet the bogey values published. (The big surprise, is the tubes we buy will average to the bogey tube, but from a statistical standpoint we will never find that mystical bogey. We rely on books like RDH4, and technical publications from the early tube developers. Maybe a few use spice, but from what I can tell, without lots of experience spice results are magic.
Does AgedHorse does a better job designing, no question. He also has the advantage of stating what performance he wants from some components, like transformers. We don't.
Agedhorse has a technology advantage, and I am not ashamed to admit it.

 

[agedhorse]

Ive got dozens of NOS Mesa STR 6L6GC and 6CA7 all which are Sylvania/Phillips as well and yes they are a benchmark/standard....i am saying there are successful tube vendors for many many years that are not giant corporations...Angela Instruments is another.....OTOH, there are LOTS of vendors on Ebay misrepresenting tubes, let alone not doing any kind of thorough testing.

My power tubes go thru a Hickok 533a and a BK 707 first for mutual conductance, shorts, the 533a will test for noise as well.... and grid emission in a BK 600 and i do realize these testers are not at full operating volts but all were highly regarded in the field back in the day...i then use a Bias Rite to track current draw in a live amp for at least 30 minutes....this method has weeded out tubes with slowly (or occasionally fast) drifting current draw.

Those tests miss all kinds of important operating parameters. For example, the tube might operate satisfactorily at one operating point that is tested for at low plate voltage but not track well to other operating points which would make the tube unstable or unsuitable for a specific application that the tube was designed for. Single point testing only tells you information about a single point. The true specifications are based on characteristic curves and it the tube doesn't conform to these curves then it may not be suitable in circuits where other points on the curve are used.

It's an example of quite possibly getting data that looks good but in fact may not be... a false sense of security. Anybody who walks a circuit up and down these curves would be well advised to have confidence that the curves are where they are expected to be.

 

[bucephylus]

No intent in humor, it was the best text on mining and metallurgy for 200 years. Basically, Old George describes that Metallurgy is a science and art, and should be left up to the professionals. Even describes making glass. He was a little early on making tube for amplifiers.
I have a few updated books, but it fun to reference the books that were produced to teach in the Jurassic Age.

Well, I’m certainly all for knowing where stuff came from. And, I do have a heck of a solid state science library; but it doesn’t go too much further back than J. Willard Gibbs, Timoshenko, and Darken and Gurry.

 

[agedhorse]

Agedhorse describes the current state of tube amp design very well. There are companies like the one he works for, that has better data than that published by say RCA, Western Electric, GE. The rest of us, have to rely on tech sheets that were developed 50-60 years ago. If we want a guitar amp, we push the tubes outside the center design values, but stay within published max values. Some tubes will take higher voltages than most of us are willing to design. We buy off the shelf. Bad mouthing vendors because the tube we got doesn't meet the bogey values published. (The big surprise, is the tubes we buy will average to the bogey tube, but from a statistical standpoint we will never find that mystical bogey. We rely on books like RDH4, and technical publications from the early tube developers. Maybe a few use spice, but from what I can tell, without lots of experience spice results are magic.
Does AgedHorse does a better job designing, no question. He also has the advantage of stating what performance he wants from some components, like transformers. We don't.
Agedhorse has a technology advantage, and I am not ashamed to admit it.

All good companies have (or should have) a technology advantage. I do use Spice, I also explore the bounds of a circuit to see if it behaves as expected when variables are intentionally changed. Some of this can be done in Spice (changing variables of the tube) and some are best done in circuit, but in circuit verification of Spice modeling is essential to verify that the model matches reality.

The earlier tube data published by the tube manufacturers is actually very good and complete, but it's important to verify new tubes to either their own published data or if applicable back to the original data. You must remember that these data sheets are complete ONLY when they are taken in context with the design formulas of the day... ie. doing the math like it was done back in the "good 'ol days". I actually was in the last class that was offered covering tube theory (combined with FET theory) while completing my EE degree. This class was probably 1979 or 1980. I also have my dad's tube text books as well as his design handbooks (kind of cool seeing his notes) which cover the math behind how these curves were derived and how they work. Now it's been years since I have gone back to this level of math, BUT I use the principles all the time... especially when designing outside the box and sometimes WAY outside the box. Understanding how the device works goes a long way to understanding what NOT to do and why.

Now I am much more math and engineering oriented in my approach, but Randy and John are much more intuitive in their approach. Part of this is experience, part of it is when Randy learned his craft (probably 12-15 years before me) and his instinct and gut feelings are VERY finely tuned. This is important because he knows where he needs to go in order to translate the sound in his head into circuitry. Since I am working more outside the box, circuits that are not very traditional, I find that the math and theory help come up with different approaches. There's lots of room for personality.

Something that I was not prepared for was the amount of statistics (or at least the principles behind statistics) that I use. Very important stuff, more important than calculus or differential equations, or linear algebra. The entire world revolves around probability and statistics, which was ironically one of my least favorite classes I ever took. Even more ironic was my other least favorite class was magnetics and electromagnetics... if I only knew then what I know now I might have paid better attention

Everything about a design comes down to specifying the right parts for the job, using them within their target statistical limits and understanding where the boundaries are.

 

[GreyBeard]

Everything about a design comes down to specifying the right parts for the job, using them within their target statistical limits and understanding where the boundaries are.

I'll go back to my previous statement in that sloppy QC from the manufacture is the issue with tubes. The manufactures are leaving it up to the tube distributors to sort out the good ones from the junk. Look at any tube sight and you will see all these ads about how they screen/burn in, look up their butt with a microscope, etc. etc. Back in the day, I could take all the tubes from my mom's 17" BW TV down to Belts Grill where they had a tube tester. I think all it did was tell you if the tube worked or didn't. Anyway, I would check all the tubes, identify the one that didn't work, and buy a new one. The new one work fine, no problem. Now it's a lot of voodoo, hocus pocus, BBBBSSSS. It's like if you play tubes, and the amp is an integral part of your instrument, you have to have a collection of tubes like a collector of fine wine. I don't mind replacing a tube now and then I just hate the BS.

 

[JimmyM]

I'll go back to my previous statement in that sloppy QC from the manufacture is the issue with tubes. The manufactures are leaving it up to the tube distributors to sort out the good ones from the junk. Look at any tube sight and you will see all these ads about how they screen/burn in, look up their butt with a microscope, etc. etc. Back in the day, I could take all the tubes from my mom's 17" BW TV down to Belts Grill where they had a tube tester. I think all it did was tell you if the tube worked or didn't. Anyway, I would check all the tubes, identify the one that didn't work, and buy a new one. The new one work fine, no problem. Now it's a lot of voodoo, hocus pocus, BBBBSSSS. It's like if you play tubes, and the amp is an integral part of your instrument, you have to have a collection of tubes like a collector of fine wine. I don't mind replacing a tube now and then I just hate the BS.

My tube collection not being used is two 6SL7's, four 6N1P's (lost my mind a little bit), two 6146b's, and maybe five 12ax7's. Most of my amps and devices with tubes have new manufacture tubes in them. There is no need to have a massive collection. There's not even much of a need to have spares anymore when you can get tubes overnighted if you really need them in a hurry.

 

[GreyBeard]

My tube collection not being used is two 6SL7's, four 6N1P's (lost my mind a little bit), two 6146b's, and maybe five 12ax7's. Most of my amps and devices with tubes have new manufacture tubes in them. There is no need to have a massive collection. There's not even much of a need to have spares anymore when you can get tubes overnighted if you really need them in a hurry.

Where do you order your tubes from?