Tube Rolling

[L0rdGwyn]

I certainly will.

[L0rdGwyn]

So, almost four months later, I finally found some time to sit down and do a comparison of the 6AQ5/6005 tubes I have on hand. 

My Audio chain:
Lossless FLAC via Foobar2000 > Topping D50 DAC > customized Crackatwoa (Sylvania 6SN7W short bottle driver, Tung Sol 421A power tube) > ZMF Auteur > my head.  Everything is plugged into a Furman PST-8D power conditioner and my DAC is linear power supplied, because, unfortunately, I am insane.

Here are the shunt regulator tube players:
Sylvania 6AQ5 (year unknown, stock tubes)
Sylvania Gold Brand 6005 (no date codes on boxes or tubes, I believe they are mid 60’s)
GE Five-Star 6005 ‘57 (white print)
GE Five-Star 6005 ‘62 (red print)

So I should start out by saying that when I first received the tubes, I did a quick AB comparison between the Sylvania Gold Brand and the ’57 GE Five-Star, nothing comprehensive at all, but at the time, I kind of thought the GE’s sounded better, so that is what I have been running since finishing my Crackatwoa.  I did not realize how right I was at the time.
I used the same song for all of my listening tests, I find that acoustic singer/songwriter music works well for me when critical listening.  Chose a song I am very familiar with, “Between the Bars” by Elliot Smith.  God only know what kind of psychological trauma I have inflicted on myself listening to this song one-hundred times.

My comparison started with what I thought would be the two highest performers, the Sylvania Gold Brand and the ’57 GE Five-Star.  After switching back and forth five or six times, listening to the first verse of the song, what I thought were subtle differences became more glaringly obvious.  The Sylvania Gold Brand sounded very much more claustrophobic, dull, and lacked texture relative to the GE’s (all relative, of course).  By comparison, the GE’s really increased the width of the track, adding far more texture to Smith’s double-tracked vocals, and more body to the music.  I continued going back and forth between the tubes to try and confirm what I was hearing, doing my best to stave off confirmation bias.  No doubt in my mind what I heard was legit, the GE’s were much better.

From there, I decided to compare the ’62 GE Five-Star and ’57 GE Five-Star, figuring they would most likely sound identical.  Well, that wasn’t true either.  While the differences were not as glaring as they were between the Sylvania Gold Brand and the ’57 GE’s, the ’57 GE’s still bested the ’62 GE’s.  The width and texture on the ’62 GE’s was better than on the Gold Brands, but the ’57 GE’s still managed the largest sound stage and detail retrieval out of the three sets of tubes.
Next I compared the Sylvania 6AQ5 stock tubes to the Sylvania Gold Brand.  To my surprise, the stock tubes sounded better than the Gold Brand, to me, despite them not being the more premium “ruggedized” 6005 model.  Differences between these two tubes were, again, smaller than between the ’57 GE’s and the Gold Brand, but appreciable.  My hierarchy was starting to form.

Lastly, I compared the ’62 GE’s to the stock tubes.  The GE’s did better on width and texture than the stock tubes, but not an enormous difference there.  I did a few more back-and-forth listens to the ’57 GE’s to reconfirm my initial thoughts, and they did that.
So, the ranking goes as follows:

GE Five-Star 6005 ‘57 > GE Five-Star 6005 ‘62 > Sylvania 6AQ5 (stock) > Sylvania Gold Brand 6005

It is somewhat fortunate I compared the best and worst tubes right off the bat.  Had I not heard the stark differences, maybe I would not have continued.  Now, these differences are not night-and-day (not like switching between stock Crack and Crack+SB, or 6080 to 5998 power tube, for example), but they are definitely appreciable and significant when you are listening for them.  This comparison got me REALLY excited, as I had not considered rolling the shunt regulator tubes a viable option to increase the sound quality of the Crackatwoa, and I’ve just about maxed out its performance in every other area.  Of course, this was a sighted test, nothing scientific here, but I feel comfortable enough with my observations that I am going to roll a number of other shunt tubes and report back.  Here is what I have coming so far:

-Siemens EL90 ‘65
-RCA EL90 ‘70 (commissioned Telefunken to manufacture these tube in Ulm, Germany, has Telefunken manufacture information on one side, RCA branding on the other)
-Tung Sol 6AQ5A smoked glass
-Tesla 6L31
-Tungsram 6AQ5
-Philips EL90

Some of these tubes are stateside, some are coming internationally.  Unlikely anyone cares that much about this, but I’ll update here with impressions anyway 

[L0rdGwyn]

So, almost four months later, I finally found some time to sit down and do a comparison of the 6AQ5/6005 tubes I have on hand. 

My Audio chain:
Lossless FLAC via Foobar2000 > Topping D50 DAC > customized Crackatwoa (Sylvania 6SN7W short bottle driver, Tung Sol 421A power tube) > ZMF Auteur > my head.  Everything is plugged into a Furman PST-8D power conditioner and my DAC is linear power supplied, because, unfortunately, I am insane.

Here are the shunt regulator tube players:
Sylvania 6AQ5 (year unknown, stock tubes)
Sylvania Gold Brand 6005 (no date codes on boxes or tubes, I believe they are mid 60’s)
GE Five-Star 6005 ‘57 (white print)
GE Five-Star 6005 ‘62 (red print)

So I should start out by saying that when I first received the tubes, I did a quick AB comparison between the Sylvania Gold Brand and the ’57 GE Five-Star, nothing comprehensive at all, but at the time, I kind of thought the GE’s sounded better, so that is what I have been running since finishing my Crackatwoa.  I did not realize how right I was at the time.
I used the same song for all of my listening tests, I find that acoustic singer/songwriter music works well for me when critical listening.  Chose a song I am very familiar with, “Between the Bars” by Elliot Smith.  God only know what kind of psychological trauma I have inflicted on myself listening to this song one-hundred times.

My comparison started with what I thought would be the two highest performers, the Sylvania Gold Brand and the ’57 GE Five-Star.  After switching back and forth five or six times, listening to the first verse of the song, what I thought were subtle differences became more glaringly obvious.  The Sylvania Gold Brand sounded very much more claustrophobic, dull, and lacked texture relative to the GE’s (all relative, of course).  By comparison, the GE’s really increased the width of the track, adding far more texture to Smith’s double-tracked vocals, and more body to the music.  I continued going back and forth between the tubes to try and confirm what I was hearing, doing my best to stave off confirmation bias.  No doubt in my mind what I heard was legit, the GE’s were much better.

From there, I decided to compare the ’62 GE Five-Star and ’57 GE Five-Star, figuring they would most likely sound identical.  Well, that wasn’t true either.  While the differences were not as glaring as they were between the Sylvania Gold Brand and the ’57 GE’s, the ’57 GE’s still bested the ’62 GE’s.  The width and texture on the ’62 GE’s was better than on the Gold Brands, but the ’57 GE’s still managed the largest sound stage and detail retrieval out of the three sets of tubes.
Next I compared the Sylvania 6AQ5 stock tubes to the Sylvania Gold Brand.  To my surprise, the stock tubes sounded better than the Gold Brand, to me, despite them not being the more premium “ruggedized” 6005 model.  Differences between these two tubes were, again, smaller than between the ’57 GE’s and the Gold Brand, but appreciable.  My hierarchy was starting to form.

Lastly, I compared the ’62 GE’s to the stock tubes.  The GE’s did better on width and texture than the stock tubes, but not an enormous difference there.  I did a few more back-and-forth listens to the ’57 GE’s to reconfirm my initial thoughts, and they did that.
So, the ranking goes as follows:

GE Five-Star 6005 ‘57 > GE Five-Star 6005 ‘62 > Sylvania 6AQ5 (stock) > Sylvania Gold Brand 6005

It is somewhat fortunate I compared the best and worst tubes right off the bat.  Had I not heard the stark differences, maybe I would not have continued.  Now, these differences are not night-and-day (not like switching between stock Crack and Crack+SB, or 6080 to 5998 power tube, for example), but they are definitely appreciable and significant when you are listening for them.  This comparison got me REALLY excited, as I had not considered rolling the shunt regulator tubes a viable option to increase the sound quality of the Crackatwoa, and I’ve just about maxed out its performance in every other area.  Of course, this was a sighted test, nothing scientific here, but I feel comfortable enough with my observations that I am going to roll a number of other shunt tubes and report back.  Here is what I have coming so far:

-Siemens EL90 ‘65
-RCA EL90 ‘70 (commissioned Telefunken to manufacture these tube in Ulm, Germany, has Telefunken manufacture information on one side, RCA branding on the other)
-Tung Sol 6AQ5A smoked glass
-Tesla 6L31
-Tungsram 6AQ5
-Philips EL90

Some of these tubes are stateside, some are coming internationally.  Unlikely anyone cares that much about this, but I’ll update here with impressions anyway 

Well, it's been nine months since I originally posted on the shunt regulator tube performance.  While I did briefly compare the other tubes I had on the way months ago, since no one ever showed any interest, I never finished this post.

Today, I decided to do a re-hash of those comparisons and report on them (albeit, in a much more abbreviated format) in case some Bottlehead forum lurker is waiting with bated breath for my final verdict.

I compared the best set from the first round, the GE five-star 6005, with the remaining tubes.

Right to the point, here are my final thoughts.

GE five-star 6005 > RCA/Telefunken EL90 = Tungsram 6AQ5 = Philips EL90 > Tung Sol 6AQ5A >>> Siemens EL90

I know "better" is subjective, so I should clarify, what I am looking for in the comparison is detail retrieval, soundstage, clarity.

The GE's outclass the rest in these departments, although I would say they make the sound slightly more dry.

The big shocker for me was the Siemens.  Holy distortion!  Definitely make the sound more "tubey" but at the cost of detail and soundstage.  I may give them another shot, since that could be appealing in some cases, but for pure technicality, not so much.

The Telefunken's, Tungsrams, and Philips tubes were too close to call, I could not tell a difference after a few listens.  Maybe nuances could be gleaned with continued listening, but then I think that is trying too hard.

Tung Sol's were pretty disappointing, lacking technical prowess, but not so tubey (like the Siemens) that they are memorable.

So, the GE five-star 6005 are THE BEST performing shunt regulator tubes in the Crackatwoa.  I have no idea why from an engineering perspective why that would be, but it is easily picked out.  These are dirt cheap on Ebay, looks like GE made a lot of them, so I would suggest anyone with a Crackatwoa give them a shot.

Once again, this is not a Speedball-level upgrade in sound, but it is an appreciable difference.

[Paul Birkeland]

What Kreg voltages do you get with the Siemens tubes?

[Paul Joppa]

Good work - thanks for posting!

I will however make a couple points - not to disagree, just to say that these explorations can be complicated. But first let me reiterate - I do not for a minute question that you heard what you heard. I am always grateful when listening tests are reported. It's one of the great things about the internet, we all get to benefit from vastly greater varieties of experience.

We all know that tubes lose performance as they age. My main point is that tubes also change as they break in. It usually takes 50-200 hours of conducting significant current to fully "form" the cathode. Until that happens, the emission capability is limited, and there is often some noise as well - sometimes just a subtle hiss, occasionally quite dramatic pops and crackles. The cathode is formed at the factory, but it's expensive and time consuming, so only the minimum is done. And it appears that the cathode loses its capability when unused for a long time. From your description, I speculate that the '57 GE was burned in before the serious comparisons started, and none of the others have been.

Second, a subsidiary point is that higher quality cathodes take longer to break in. This at least was understood in the "golden age" literature. Small amounts of silicon in the cathode base metal (nickel) speed the formation of the cathode, but also make it age and lose emission faster. So only the premium tubes have the high-purity cathode base material (expensive), and they take much longer to form before leaving the factory (also expensive).

I have no idea whether these effects influenced your comparisons, of course. But it's possible. Always keep an open mind!

On the subject of the regulator tube affecting the sound, this is much more likely when a cathode follower is involved. That's because the signal current loop includes the power supply (regulator in the C2A case). Back when we were developing the hybrid shunt regulator, I build a cathode follower with several candidate regulator topologies which could be switched for comparisons. We definitely heard differences! For comparison, the Mainline uses a current-source plate load which provides something like 50dB isolation from its (regulated) power supply. So does the BeePre.

[L0rdGwyn]

What Kreg voltages do you get with the Siemens tubes?

I'll check some point today or tomorrow, PB!

Good work - thanks for posting!

I will however make a couple points - not to disagree, just to say that these explorations can be complicated. But first let me reiterate - I do not for a minute question that you heard what you heard. I am always grateful when listening tests are reported. It's one of the great things about the internet, we all get to benefit from vastly greater varieties of experience.

We all know that tubes lose performance as they age. My main point is that tubes also change as they break in. It usually takes 50-200 hours of conducting significant current to fully "form" the cathode. Until that happens, the emission capability is limited, and there is often some noise as well - sometimes just a subtle hiss, occasionally quite dramatic pops and crackles. The cathode is formed at the factory, but it's expensive and time consuming, so only the minimum is done. And it appears that the cathode loses its capability when unused for a long time. From your description, I speculate that the '57 GE was burned in before the serious comparisons started, and none of the others have been.

Second, a subsidiary point is that higher quality cathodes take longer to break in. This at least was understood in the "golden age" literature. Small amounts of silicon in the cathode base metal (nickel) speed the formation of the cathode, but also make it age and lose emission faster. So only the premium tubes have the high-purity cathode base material (expensive), and they take much longer to form before leaving the factory (also expensive).

I have no idea whether these effects influenced your comparisons, of course. But it's possible. Always keep an open mind!

On the subject of the regulator tube affecting the sound, this is much more likely when a cathode follower is involved. That's because the signal current loop includes the power supply (regulator in the C2A case). Back when we were developing the hybrid shunt regulator, I build a cathode follower with several candidate regulator topologies which could be switched for comparisons. We definitely heard differences! For comparison, the Mainline uses a current-source plate load which provides something like 50dB isolation from its (regulated) power supply. So does the BeePre.

Thanks for your input Paul, a completely valid point, it's something I had briefly considered myself.  Unfortunately, burning in each set for 100 or so hours would be a process, but you've piqued my interest.  You are right, the GE's have definitely spent the most time in my amplifier, especially compared to this second round of tubes.  If our esteemed tube manufacturer's reputations are to be believed, I would expect high quality from the likes of Telefunken, Siemens, and Philips (Holland), which didn't necessarily show up in my listening.

I'll give it a shot and report back, 100 or so hours for each set and see if my impressions change.  Heck, I'll be listening anyway, might as well pretend I am being productive!

Really interesting regarding the cathode follower topology, I guess my ears can be trusted after all (for now).  We'll see what some burn time yields.

[L0rdGwyn]

I think I have completed my shunt regulator tube collection for the Crackatwoa, or at least, I don't want for any others.

Introducing the Telefunken 6005, manufactured in Ulm, Deutschland, based on the General Electric 6005.  Internal construction is very similar, although the getter is raised and parallel to the top mica, unlike the GE 6005.  Also, the plate is sectioned only in one place, rolled like a hard-candy wrapper, whereas the GE 6005 plate is sectioned in two places, like two half-circles meeting.

These tubes offer the greatest clarity, space, and bass definition of any shunt regulator tubes I have used in the Crackatwoa.  Listening right now with a Tung Sol 6SN7GT round plate input, MOV A1834 output.  Incredible.

[Love Rhino]

Has anyone ever tried a 6v6 tube with an adapter for the 6005?

[TheRedMantra]

I have been rolling tubes in my C2A and find that I love the GEC A1834/6AS7G. I came across an adapter to run two GEC A2293/CV4079 tubes and got it as a cheap experiment to see how similar it might sound to my all-star favorite.

I measured the voltages before running these tubes and everything is happy except the “OB” measurements on the high current boards are 79.4V and 82.2V; below the recommended range of 90-110V. Is it safe to run these out of range?

[Paul Birkeland]

You would need to post the OA and OB on the center C4S board and the OB voltages on the outer high current C4S boards. 

[TheRedMantra]

Center OA 67.6V
Center OB 71.2V

OB HC 147.0V & 147.9V

[Paul Birkeland]

OB HC 147.0V & 147.9V

Those voltages are what I would expect for OA on those boards.

[TheRedMantra]

Sorry, I got confused there. Yes, those are the OA measurements. Here are the measurements:

Center OA 67.6V
Center OB 71.2V

OB HC 82.2V & 79.4V

[Paul Birkeland]

Yeah so you have about 10V of bias there which isn't so bothersome.  If you had something like 67V coming out of one of the small board outputs and 70V on one of the high current OB terminals, that could have some wonky effects on the operation of the amp.