The sibilance confused me initially, as there seemed to be more on the track that I ended up preferring. That comparison did convey to me a bit of what people perceive as tube 'warmth and character' vs the 'clarity' that I favoured (fyi I do get a bit of stress-related tinnitus in the 8 kHz frequency range, but don't recall it bothering me at the time).
-
WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
Take the blind challenge! 300B SET vs. Straight Wire with Gain
- Thread starter GXAlan
- Start date
The sibilance confused me initially, as there seemed to be more on the track that I ended up preferring. That comparison did convey to me a bit of what people perceive as tube 'warmth and character' vs the 'clarity' that I favoured (fyi I do get a bit of stress-related tinnitus in the 8 kHz frequency range, but don't recall it bothering me at the time).
solderdude
Grand Contributor
Yep... less sibilance can be preferred, certainly initially, but not when other aspects have to 'suffer' for the sake of less sibilance.
charleski
Major Contributor
That's a high output impedance interacting with the particular impedance of the speaker he used along with a lot of hf odd-order distortion products. This is common to a wide range of tube amps (and even some of the poorer solid-state designs) and is dependent on the actual speaker being used, though the XPL90's impedance graph is fairly typical. I don't know why anyone would think this was a special feature of the 300B. As I showed, this sort of overdrive effect is easy to simulate with DSP.
- Thread Starter
- #245
How did anyone know which numbers to put into the DSP? I have read a lot about sparkle, but I have never seen anyone make the comment that there is a dynamic treble boost that occurs with a variety of common bookshelf speakers. To my knowledge, my review at ASR is the first to articulate this effect. Maybe this is also a characteristic of other Chinese made SETs like the Willsenton that was popular among the popular YouTuber’s and absent from the very high end 300B SET designs.
So then you are getting into semantics.
1) There is a feature of the 300B that separates it from straight wire with gain. This makes it “special”
2) Through the measurements of this SET 300B, we have potentially translated 300B sparkle into this treble boost. We have found and measured on element of that difference.
Or stated differently, through these efforts of
A) acknowledging that subjectivists hearing the 300B SET’s “special sound”
B) investigating with available hardware at personal expense
You now have gained new knowledge about one secret of the sound signature of this particular amp.
solderdude
Grand Contributor
We still don't know.
All we gained in knowledge is about this particular amp into a resistive load.
What that taught us that it has an unexplained treble roll-off in a resistive load that is a-typical of SET.
The distortion seen and recorded is more 'typical' and could be simulated in DSP when the amp's transfer characteristic would have been known. It isn't because not enough measurement data.
A second effect, that will be harder to replicate, is the output impedance per frequency. This changes the FR load dependent so could only be replicated in DSP for a known speaker load.
Leaves the roll-off which, given the teardown, is most likely due to used line-transformer.
There is no dynamic treble boost. There is only a substantial high output impedance that is likely frequency dependent as well.
The used bookshelves, acc. to several SET specialists, are not intended to be used with these kind of amps.
You did make the first recording so that people could hear the difference such a type of amp (but not totally representative of expensive SET) that this type of amp does in a resistive load. Had it been a SET amp without that input transformer most likely the audible results would have given a smaller preference as there would not have been an audible treble roll-off which was a give-away.
The characteristic of such amps is just caused by tube distortion (easy to simulate that aspect), power output but above all the unusually high output Z that can cause tonal changes to the sound that is speaker dependent. This can be a positive effect or a negative effect.
As stated this is distortion profile and load dependent frequency response modification. The effect can be 'special' for certain and is not easy to predict when the speaker impedance and amp output Z/frequency is not known.
The only thing that was found was how the amp in your possession reacts to the 3 speakers you used and a resistive load.
It is quite likely, however, that other speakers will have a similar 'tonal modification' to similar speakers when connected to this particular amp and that indeed the amp will add 'sparkle'.
That is not what the blind test showed here as there were no files with the extra treble instead there were non representative comparisons with a tube amp that rolled off the treble.
When that is your goal then you should do a new recording and not use a resistive load but one of the speakers.
I think that would be far more representative of what you hear. After all you listened to speakers and not to the resistor.
That, however, was not shown. Besides this is acknowledged because 'we' know that the output Z is unsually high and harmonic and IM distortion is added which modifies the original (intended ?) sound.
I would recommend to invest a bit more time and also make available recordings with speakers as a load.
Recorded at the actual listening levels you listen at.
Make these available to @danadam who can level match the files to the recording of the D90 output (basically the input of the amp)
Don't get me wrong. This is not to make any less of your otherwise great effort to review this amp by adding this audible experiment. It would be more representative to what you heard as special IF you re-did the comparison with 1 or 2 speakers as actual load.
Then we can actually hear the modification the amp makes to the input signal of the speakers you intended us to hear.
- Joined
- Jan 27, 2019
- Messages
- 7,316
- Likes
- 12,265
It's worth pointing out on a science-minded forum: We should be cautious about making wide extrapolations from a single (or very few) test samples.
There's a reason why science operates by having numerous checks on the bias of the scientist: even the person seeking to be as objective as possible can fall
for cherry picking data to support their hypothesis. And why cautious scientists are so often saying "but we need more data/more studies." Likewise, it's just as easy for those here to seize upon very limited data to say "Ha! Suspicions confirmed!"
Seems to me there's a lot of different 300B amps out there and different speakers, and concluding from this test that no 300B/speaker combo can produce sound altered enough for some to find pleasant seems a bit rash.
(I don't have 300B amps, but I have posted about succesfully identifying my tube preamp in blind tests, on certain properties I find pleasant vs the solid state preamp).
Thank you all for elaborating.
In addition to the impedance-related FR (that varies with actual speaker load) I interpreted the second graph in the OP's post #68 (spoilered in the thread, and presented again in the review thread) to be illustrating a dynamic treble boost (or "the opposite of a loudness curve").
Note (contra @charleski) that I'm not arguing this is unique to any of the semantic signifiers I've used to identify the device under discussion (nor that this is desirable/ideal/preferred, just that it is different/potentially audible/interesting).
In addition to the impedance-related FR (that varies with actual speaker load) I interpreted the second graph in the OP's post #68 (spoilered in the thread, and presented again in the review thread) to be illustrating a dynamic treble boost (or "the opposite of a loudness curve").
Note (contra @charleski) that I'm not arguing this is unique to any of the semantic signifiers I've used to identify the device under discussion (nor that this is desirable/ideal/preferred, just that it is different/potentially audible/interesting).
- Thread Starter
- #249
Agree that the knowledge gained from the blind test is in a resistive load, not the published measurements of the stated amp used in the speaker which is what @charleski used to create his DSP which replicated that speaker/amp combo.
No disagreements there.
In my mind, the actual load would be a fixed FR change. In my mind, that would easily allow the resistive measurement to be transformed into a simulated speaker level by doing a REW sweep through the speaker load. The problem is that it is signal dependent since the FR changes if I send a REW sweep at different dBFS levels.
So what I record will only be representative of a very specific speaker at a very specific volume. Which makes the whole process even more daunting since I need to record through a speaker load at different levels and then digitally match the volumes, and then we additionally introduce the differences in THD+N on top of the FR changes (which themselves are partly due to THD).
What's the best bookshelf to use with a SET when I'm listening within the power envelope? I listen to at about 70 dB and 2.5 meters.
I can see bad speakers if there's a mismatch between sensitivity and available power, but after that, any speaker is a fair choice in my opinion.
Right. But in choosing the 3 speakers, I have simple 2-way speakers and each has a different high frequency transducer. I have shown a phenolic compression driver, a ribbon tweeter, and a dome tweeter. What other 2-way design bookshelf should I have added?
No real goal other than to share experiences and help decrease the tribalism in this world and encourage others to try making their own measurements and/or recordings.
It's not my worth spending more time and incurring RIAA copyright risk for just 43 other people to care enough to vote...
Time to move onto the next test.
solderdude
Grand Contributor
You would just have to record it at the normal listening level you would use normally.
Then record that output making sure you do not clip the input.
The level matching can be done afterwards. You can also create a separate file using a 400Hz (or 500Hz) sine and record those, then this can be be used as a guide for the level matching afterwards.
The ones you have don't seem to be a very dissimilar load. Just pick the one that sounds the best to you at the level you normally listen to.
Does not matter, no need to add more speakers. There are literally thousands of speakers and most will differ here and there as a load.
That's fair... it is just that you spent a lot of time but did not make us listen to SET sound at all making the time others spent, believing they would be hearing a SET amp a waste of time other than it was just comparing a degraded signal vs an original signal.It's not my worth spending more time and incurring RIAA copyright risk for just 43 other people to care enough to vote...
- Thread Starter
- #251
I’m sorry. Sometimes we all waste our time on Internet forums and participating in Internet arguments.
In the original post, I did point out that everything was going into a resistive load: “I then switched out the speakers for a E1DA Cosmos ADC with a resistive load and recorded real music at 32-bit 44.1 kHz.”
DanielT
Major Contributor
Typical. Tested again group B: 2 vs 5 and in connection with that disconnected the subwoofer. Okay 5 there her voice sounds brighter and on the contrary a little more muffled, subdued treble in 2. But I wrote typically because now I don't know if I'm hearing it (but I think so) or if I'm just imagining that I'm hearing it because you're "supposed" to hear it.
Best to stop testing and think about this before I'm admitted to the mental hospital.
Instead, today I will relax with a Friday beer and just listen to music without analyzing it.
Best to stop testing and think about this before I'm admitted to the mental hospital.
Instead, today I will relax with a Friday beer and just listen to music without analyzing it.
- Joined
- May 26, 2021
- Messages
- 510
- Likes
- 801
If you can find a 16 Ohm bookshelf (good luck with that...) that would be a good place to start. Rogers LS35/a??
An amp like this normally would be used with horns if the user is in a normal room. For nearfield a high efficiency driver in a bass reflex might work; the amp can't make the 'full range' driver do much excursion so it might not be too bad in the Doppler Effect dept....
IOW a bookshelf isn't the normal fare for an amp of this type.
charleski
Major Contributor
Let me recommend an important element of scientific philosophy: the principle of mediocrity. This evolved from the work of Copernicus (that we are not privileged observers) and evolved to become a foundation stone - the laws of physics that apply here on Earth also apply to the furthest galaxy, all humans share the same physiology, any sample taken at random is probably going to be highly similar to another random sample.
I’m sure there are many people who are happy listening to their 300B amps. The major finding here was that there was no statistical difference between signal chains. These people would probably be equally happy listening to something else, but only if they were willing to accept that there was nothing special happening. Many people struggle with the idea that they lack specialness, which is sad.
Anyone who wants to contend that the 300B (or any other directly-heated triode in SET configuration) has special properties that place it in a different class to other active elements will need to provide proper evidence of that. This little study rejected that hypothesis quite decisively. My suspicions are founded on the principle of mediocrity, and I’m not surprised to see them confirmed. The principle of mediocrity has proved to be an extremely powerful tool in science.
charleski
Major Contributor
It’s definitely interesting, and well done for demonstrating it (it would be good to see the way distortion varied with frequency at the different output voltages though). But, as I showed, this is the result of filtered odd-order distortion. A compressor/expander is simply a way of symmetrically distorting the waveform, producing odd-order distortion. It might be interesting to explore the causes of it in this case, but I suspect they will be rooted in the specifics of the plate load/transformer/load reactance leading to the tube getting overdriven at high frequencies.
Several posters—including the OP—demonstrated that the comparison files gave us part, but not all of the sonics of the tube amp used. You might be happy to over-stretch the conclusions, but apart from reinforcing a personal narrative (or "suspicions" if you prefer) there's no sound justification for doing so.
Invoking statistics that way is pretty dodgy (the OP does something similar, but draws conclusions non-egregiously). Out of the small sample, a number of participants differentiated consistently, via blind A/B or ABX. Others differentiate at all. That's a distribution of ability within a population (without knowing all the variables that led to same, don't launch into an anti-exceptionalist diatribe) not an empirical conclusion about sonic characteristics.
Last edited:
computer-audiophile
Major Contributor
@MattHooper
As always, a very well thought out comment - thanks.
I have already tried to say something similar. This amp is a random single sample and I don't even think it's among the best I've personally heard and know. Also, what is to be concluded from this single sample? Leaving aside the methodology of listening over the internet with the own system at home.
There is a lot written here about blind tests and I understand the concept, having participated in blind tests myself.
I can usually hear differences quite well with my eyes open.
I would recommend to listen to as many audio things as possible and if possible to compare A/B/C... side by side. I was very rarely able to directly adopt experiences from other audio enthusiasts 1:1, even though I know some legendary figures and installations, was deeply immersed in the scene and mixed in. We are all too different! That usually starts with the taste in music.
Last edited:
computer-audiophile
Major Contributor
When it comes to how something sounds, logically this can only be done by listening to it directly and not through a bottleneck. Quality of an amp e.g. can be substantiated in context by measurements. But good measurements are only one prerequisite for good sound. Inappropriate methodology can be misleading. In the end, sound can only be heard.
I too, usually listen to sound, strange as it may seem. When it comes to audio, I tend to listen to sound that has passed through many bottlenecks long before it got anywhere close to my ears. 100% of the time.
computer-audiophile
Major Contributor
Sure, no contradiction, on the meta-level you can see it that way.
But here we are talking about a single component, namely a specific Raphaelite 300B amplifier. It is about how this one sounds.
Similar threads
- Poll
- Poll
