Five 6AS7g type tube comparison

Five 6AS7g type tubes were measured for comparison.
Measurement methods and notes can be found in this post: xDuoo TA-66 technical measurements

Index:
Post 1: Introduction, output impedances, THD+N surfaces
Post 2: D2 surfaces
Post 3: D3 surfaces
Post 4: 4+HD+N surfaces
Post 5: frequency response, bandwidth (transient response)
Post 6: 1 KHz 0 dBu distortion suite, 1 KHz distortion vs level
Post 7: Residual Noise, Multitone
Post 8: Chatham 2399 noises, measurement suite report pdf files
Post 9: reserved for additional data and/or corrections

Amp used for tube evaluation and measurements: xDuoo TA-66

Output tubes evaluated:

Left to right
Shuguang 6n5pj
Sovtek 6AS7g
Chatham 2399
Tung-Sol 5998
GEC A1834

Measurements for each tube will be presented in the same order as listed above. Animations will also display in the order listed above.

Introduction

Tube rollers have long described different auditory experiences with different tubes in their respective amplifiers. Traditional measurements have not correlated well or only hinted at audible differences. An examination with a new measurement technique, distortion surfaces, may provide greater insight.

In this evaluation only the output tube in an single OTL amplifier will be changed.
Driver tube will remain constant, in this case a 6n2.

Listening was performed with a Holo Audio Cyan 2 and Spring 3 DAC, Sennheiser HD800S, HD800-JAR, HD6x0 headphones.

Brief listening descriptions:

Shuguang 6n5pj rendered music nicely, feeling well balanced, engaging
Sovtek 6AS7g added some refinement in transient response and a little more punch in the bass.
Chatham 2399 very similar to TS 5998 to my ears
Tung-Sol 5998 does some additional magic that is to my liking, but I don't think it is worth today's asking prices for what is adds over the Sovtek or Shuguang tubes.
GEC A1834 has a gorgeous midrange but gives the music a vintage presentation. Bass is slightly less punchy compared to the others and highs feel a little rolled. A great example to hear this vintage feeling is a track by the Charlie Haden Trio West called Haunted Heart. At 6:28 the song transitions from modern recording to a 1950 cut of Jo Stafford. The midrange remains wonderful but one can tell it is definitely a vintage recording. The A1834 is similar in rendering music but not so noticeable as the example recording. Ridiculous prices today and unobtainum.

Detailed listening descriptions may be found in this thread:
6080 comparisons
6AS7 comparisons

Output Impedance (nominal approximation):

Output impedance is the first traditional measurement demonstrating differences existing between the tubes. Note Shuguang and Sovtek to be similar at 124 Ω. Likewise 2399 and 5998 are similar at 98 and 93 Ω respectively. Possibly one of the factors in what I find attractive in the sound of these two tubes. A1834 is in a class by itself at 116 Ω.

Distortion Surfaces

This new 3-D presentation of distortion measured over frequency and level contains much more information than traditional numeric specifications. Surfaces for THD+N, 2nd harmonic (D2), 3rd harmonic (D3) and 4+HD+N (residual crap) appear below. THD+N does allude to differences but is a composite of too many measurement attributes. Decomposition into the other three will provide greater insight.

When comparing the surfaces presented pay attention to the overall shape of the surfaces. For example note all five THD+N share falling to a dip and then rising again. The overall surfaces vary in the x-axis (Dist dB) and the dip varies in the z-axis (dBu) but the overall shape is remarkably similar. Shape similarities exist in D2, D3 and 4+HD+N distortion surfaces but greater variance in shape appears. D2 and D3 are the most relevant for correlating to audible similarities and differences.

THD+N











Animation for enhanced visualization.

 

2nd harmonic distortion D2




Observe how Shuguang and Sovtek are remarkably similar




Observe out 2399 and 5998 are remarkably similar but distinctively different from Shuguang and Sovtek.


Observe how A1834 differs from the previous four.


Animation for enhanced visualization.

 

3rd harmonic distortion D3










Again the Shuguang and Sovtek are similar to each other while 2399 and 5998 are also similar to each other but somewhat different from the modern production tubes, though the difference is less pronounced. A1834 is more similar to the other vintage tubes than the modern production tubes yet still in a class by itself.


Animation for enhanced visualization.

 

4+HD+N distortion (crap factor)












Animation for enhanced visualization.


Shuguang 6n5pj is the cleanest but they are all remarkably similar.

 

Frequency Response comparisons










A1834 is slightly different from the rest, but if frequency response were the yardstick, nothing stands out as significantly different.

Square wave response comparisons

Shuguang 6n5pj

BW ~= 125.8 KHz

Sovtek 6AS7g

BW ~= 120.0 KHz

Chatham 2399

BW ~= 133.2 KHz

Tung-Sol 5998

BW ~= 130.9 KHz

GEC A1834

BW ~= 125.0 KHz

Chatham and Tung-Sol produce the fastest transient responses which may contribute to what I enjoy about them.

 

Amplifier 1 KHz Distortion vs Amplitude


6n5pj shows good matching between left and right sections
It also has the lowest distortion numbers


Sovtek isn’t as well matched between sections and shows higher distortion numbers


2399 is better balanced between sections while having higher numbers than 6n5pj


5998 is not so well matched between sections. Right channel has remarkably lower numbers.


A1834 isn’t particularly well matched and has higher distortion numbers.


Amplifier 1 KHz Distortion vs Amplitude











As these are all the 1 KHz slice from the distortion surfaces they allude to differences but do not provide as much insight as the distortion surfaces above.

 

Residual Noise











Greatest differences are AC mains at 60, 120, 180, 240 and 300 Hz


Multitones











6n5pj is the cleanest but otherwise these measurements are substantially similar.

 

Tube Noise

Buying vintage, NOS tubes is always like playing roulette. One tube may be quiet, well matched between sections and render music representative of the particular tube family. Another may have considerable section mismatch and/or suffer from audible hum, microphonics, and various tube noises during warmup and/or post warmup.

The Sovtek 6AS7g I acquired has a low level glassine shimmer in its left channel, near threshold of audibility. So far a Herbie’s Audio Lab Rx damper appears to have reduced the shimmer to below audible threshold.

Chatham 2399 warmup and residual noise
Tinks can be heard during first 10 minutes of warmup at approximately 1 minute intervals. After warmup there are residual glass like sounds in right channel, very close to audibility threshold. When no music is playing and the room is VERY QUIET I can hear them. Even the slightest room noise such as HVAC running softly masks these glass sounds.

Attached are recordings of the tinks and glass sounds. Each one has been recorded at 94 dBSPL = -6 dBFS and then digital gain applied to create a version 30 dB louder. To hear in normal playing context, play a typical music track and adjust sound level for normal perceived loudness. Then play the 94 dBSPL recording. It may sound faint or not be audible at all in your specific environment. Play the 64 dBSPL with the same amplifier sound level and the sounds will become apparent along with AC mains hum.

Standard measurement suites pdf files for the amps attached

Please note GEC A1834 shows a FAIL result. Several tests results failed due to the high 60Hz and 120Hz AC mains noise being above -72 dBu or 0.025%.

 

reserved for additional data and/or corrections

 

Fantastic stuff!

The similarities and differences between the 2399 and the 5998 are interesting, because they are the exact same tube. All 5998 and 2399 were were manufactured by Tung Sol, with different labels. The highly desirable Western Electric 421A is almost universally considered to also be the same tube, but cherry picked for high Gm and matched sections.

I've never really seen (or sought for) data on the same tube being measured for distortion across tubes and individual sections, so this is really very handy. And 'regular' tube measurements/tests likely tell us very little.

 

I always thought that as well, but have never seen them side by side before. The photo sure makes it look like the plates on the 2399 are shorter than the 5998... but that could just be an optical illusion...

 

There's lots on the Bottlehead forums about this. Differences in construction seems to be present across manufacturing dates, but both tube types manufactured at the same time are almost certainly the same.

 

The multitone charts are interesting to me. It's not just that the Shuguang 6n5pj is cleaner, it has a descending pattern on the grass that doesn't seem to level out within that frequency range, whereas the other ones all level off by 1k.

 

dual triode assembly is positioned higher in the 2399 than the other tubes. Look closely at the top neck on the 2399. Plate size is the same as 5998.

 

I’m thinking tubes are sometimes like expensive wines. There’s probably some bias because of a vintage or justification of the price spent on rarities, and I bet most people can’t tell the difference or would care if they didn’t know. That bloominess is a feature? Those recessed mids give out a fun v shaped sound? That warmpoo is a signature warmth? Ooh I like that, uh huh.

Joking aside, I think you guys are right…it’s the implementation, the caps, the OPTs, etc. If the amp doesn’t sound good with so-so tubes, putting WE, Mullard, Tung-Sol etc probably won’t help that much. And let’s not look down on a Shuguang, or a Gold Lion, or a JJ. These probably pair well with the right equipment, and without the NOS nervosa.

Anyway, pass me the Two Buck Chuck while having a nice dinner with good friends.

 

Honestly, the TA-66 these tubes are being compared in probably -is- the right equipment that the Shuguang pairs well with. I can't believe someone would design an amp like that for mass production if they didn't have access to a large stock of reasonably-priced tubes that they know for sure will allow the amp to sound good.

 

Updated first post to include links to very detailed comparisons in 2015 thread on Bottlehead Crack
Detailed listening descriptions may be found in this thread:
6080 comparisons
6AS7 comparisons

 

Thomson-CSF 6080WA

Thomson-CSF 6080WA (abbreviated T6080 for convenience) was measured to compare with previous 6AS7g tubes in posts above and can be seen in the picture found here

Measurement methods and notes can be found in this post: xDuoo TA-66 technical measurements

Amp used for tube evaluation and measurements: xDuoo TA-66
DAC used for listening evaluation: Holo Audio Cyan 2
Headphones: HD800S, HD800-JAR, HD800-Jmod

Brief listening description:
T6080 has a wonderful midrange similar to the A1834 while maintaining transient response and bass punch of the Sovtek 6AS7g yet having an airy quality like the Tung-Sol 5998. Each of those tubes performs the specific attribute better, however the T6080 is a very good balance of all attributes. This balance probably accounts for my current affinity towards the T6080 during listening sessions. As of this writing, it is still an available tube, though not at bargain prices anymore nor outrageous either, yet.

Output Impedance (nominal approximation): 117 Ω
Very similar to the A1834 at 116 Ω.

Distortion Surfaces









Frequency Response



Bandwidth (transient response)

BW ~= 128 KHz


1 KHz 0 dBu distortion suite



Residual Noise



Multitone



Measurement suite report pdf file attached below

 

I think Atomic Bob's work here is really cool. I've always wanted to see something like this with tubes. I've owned RCA 6as7g, Tung Sol 7236, Tung Sol 5998, Svetlana and some tubes just seem to work better than others - I wondered sometimes with 5998s if it was just the output-impedance-lowering-factor cleaning the sound up a bit on these OTL amps but there is just a lot going on, subtle quality differences that can ad up. I recognize that on a well-designed amp the tube shouldn't make or break it but there are obvious differences that can become strong preferences. As impractical and basically impossible as it may be - I'd love to see something like Atomic Bob's work on 6as7 equivalents with NOS 6sn7 tubes. It's such a crap shoot and hard to say if a lot of tube-specific-lore is just inherited subjective nonsense passed around to streamline sorting through the mess of it all like a bad game of telephone or sellers just hyping things up to profit on FOMO.