Hi Marco
Actually my comment was partially in gest. Of course the situation is not black and white.
But we do know a great deal of what correlates and what doesn't. What bothers me is that people pick and choose what they want to believe. Like THD and IMD. Its been shown that it does not correlate with subjective impression and yet people still talk about it all the time like it matters. Olive at JBL has shown measurements that correlate to subjective impression nearly 100% and people still talk about frequency response issues like his results don't matter. Many here have not read all of the work out there and they can be excused for not knowing what matters and what doesn't. But once you know the "answers" it is inexcusable to simply reject them because they don't suite your position.
And there are areas where we don't have the answers and in those areas it is possible that contrary views are rational as there are still questions. But those areas are not that big. We know a great deal.
Last edited:
Do you mean something like that? This is my system and the presently discussed design strangely looks alike. This is a 15 " woofer JBL 2235, a 10" mid bass Delta 10-A (Rms 2.0
), Autotech JMLC 350 with a GPA 388-8 (ceramic magnet and aluminum diaphragm) and Autotech JMLC 1000 with Celestion CDX1425, digitally crossed and multi amplifiedChris
Attachments
g3dahl,
This is exactly what I found when comparing a Truextent Be 4" diaphragm to its ribbed Ti counterpart in A/B testing with the same JBL core.
The "silent" background could already be get with a Ti-SL (aquaplassed) diaphragm (works pretty well at damping those nasty breakups), but the Be bring something else to the table, related to tone as you describe.
I did my comparison with EQ to closely match responses because any variation in frequency response can impact the subjective impression a great deal.
Looking at your measurement it looks like a simple 1st order LP filter around 4 or 5kHz could add pretty well with the current rolloff to obtain something not too distant from a 2nd or 3rd order LP filter around 8kHz or so, depending on the actual impedance curve...
What crossover frequency are you aiming at for the ribbon?
An externally hosted image should be here but it was not working when we last tested it.
Pano,
I am using autoformers in my active setup, without any other passive component between my amp and compression driver.
I have not made A/B comparison with a simple lpad, or even a simple inline resistor (+ EQ), but I plan to do that soon.
The "impedance curve magnifying" thing is not a problem if you consider it is just multiplied: the load variation is still the same for the poweramp. Same would go for a passive crossover: it is not more difficult to do a passive crossover for a 16ohms unit than for a 4ohms one (and the components often turn out to be less expensive as Lynn pointed out
), and this is exactly the same situation (with bigger absolute impedance variation for the higher load). This does not change the way the passive filter will act.Let me list some technical arguments that could go in favor of an autotransformer over an lpad in an active design:
- higher load for the poweramp, which means less distortion
- the high load will let an AB poweramp work in class A for the same amount of Watt (received by the driver) than it would have without the autoformer, whereas it would not be the case with an lpad
- more damping for the compression driver, which could not bring any audible gain, but can protect the diaphragm from overexertion from external "signal", like a big pressure variation from a brutally closed door, or from a subwoofer.
This one sound anecdotal, but I read somewhere (can't remember where) the story of an installer that kept blowing his diaphragm all the time until he removed the protection cap and let the amplifier damp them again...
the cons:
- expensive, but it will let you use less expensive passive components. Even a cheap autoformer (like the ones found in parts-express inwall volume control) can be good enough for HF duties (with a bit of EQ up high...)
- autoformer add distortion, but it should not be a problem for HF devices (even when crossed over in the 500Hz range...). I could not measure any additional distortion in my setup anyway (whereas I could clearly measure the reduction of distortion.noise from the amp...)
Perhaps this is off topic but can someone link some of the tests and/or what was measured that corresponds to subjective impressions. I'd like to educate myself and do some reading. Thanks in advance!
Now this I completely do not understand. Isn't the "stated" reason for the autoformer to reduce the sensitivity? Why not just do that in the amp or active crossover if the system is active?
(But I would never connect and amp directly into a compression driver for safety reasons. One bad turn on transient could break it. Even in an active setup I use a single cap and a resistor and then just do the active part with these components in place.
The best source is clearly Toole's book (a countryman of yours.) But he hardly even mentions the THD issue. For that you should go to my website and read the papers. Olive's papers are also at the AES website or maybe even on his website.
It then becomes important to note the areas of Toole that are still controversial. To me those are almost exclusively having to do with the very early room reflections and whether or not these are desirable. There is not a consensus on those details. But the rest of it is all pretty solid.
I am sure that many here don't "buy" Toole. He is clearly negative on the audibility of electronics. I am sure that he doesn't use tube amps.
There is quite a bit of Olive citing Olive here: Harman Innovation Hub | The Science and Marketing of Sound Quality
Suppose I will venture an analogy, I dislike analogies but here goes.
If you build a house to be built of standard size bricks and tiles with standard fenestrations it al goes together if the geometry ans th maths is correct. However, if the bricks are all different sizes, and the windows are no multiple of brick dimensions, the house may cobble together with gaps filled with different cement fillets.
Hifi with component parts is like this with the true parameters i.e flaws of the parts can put any math of perfect parts into the long grass. It gets easier if you stop adding DAC /ADC 's and other bits like confetti. You cant get it wrong on ten fingers can you ? Well sounding minimalist amps speakers etc will actually measure well and sound well. This is also because a simple design usually done with the best or good parts does not have any S... to hide.
So the measurable sounds and looks good and so does the unmeasureable measurables sound good and the unmeasurable unmeasurables also sound good. Rumsfeld really was on the button.
To do that you would need an amp with a very low internal gain (10dB or so).
With the typicall 26dB gain amplifiers feeding directly a ~110dB/1w1m device the noise is quite noticeable, even with its input shorted, and no matter what the S/N spec says (especially considering these specs are either filtered or weighted in frequency, and modern amps tend to have more noise up high).
You also get any crossover distortion and similar effects right in your head.
So with those amps you have to use either an lpad or an autoformer...
An autoformer turns any amp into a "firstwatt" design
That is similar to what JBL did with the M2, but with a cap and a lpad. The lpad is used to reduce the noise from the amp (mega watt class D amp), and the cap is part of the filtering. They do not put a resistor parallel with the cap though: the rolloff is compensated actively.
LineSource,
A cursory look at the graph you printed shows something that I have seen many times with titanium, its tendency to have high frequency breakup modes. I see the same thing in POS graph of the aluminum vs Be diaphragms in the Radian drivers. It shows as a rise in output at high frequencies and also as that rather rough decay plot of the impulse response.
This is the major reason in my eyes not to use titanium in high frequency devices, what many will refer to as fatiguing when listening to music. I can imagine that many manufacturers would not want to work with magnesium due to the flammability issues during manufacture. It is not an easy material to work with and even in aerospace it is now seldom seen being replaced by carbon composites where weight is an issue.
A cursory look at the graph you printed shows something that I have seen many times with titanium, its tendency to have high frequency breakup modes. I see the same thing in POS graph of the aluminum vs Be diaphragms in the Radian drivers. It shows as a rise in output at high frequencies and also as that rather rough decay plot of the impulse response.
This is the major reason in my eyes not to use titanium in high frequency devices, what many will refer to as fatiguing when listening to music. I can imagine that many manufacturers would not want to work with magnesium due to the flammability issues during manufacture. It is not an easy material to work with and even in aerospace it is now seldom seen being replaced by carbon composites where weight is an issue.
Well, I think bricks are used more for cosmetic purposes here nowadays. But in ancient construction, there always were ways they could put together various pieces to get what they wanted.
I do agree you have to start from something simple, but when you want to improve in it, you either have to improve the component, use it in a smaller operation range, or design something to improve on it. From what I can tell, there is nothing wrong with the digital format as long as you increase the bit depth and sample rate. The main problem is proper design and layout of the circuits because the digital circuits do have some negative effects on the analog sector. The power supply is also important.
Old school records are fine, and they certainly audibly have some advantage, but there are problems that you will never solve.
Total tolerance control of the equalization curve,
Complicated matching from stylus to the equalization circuit.
Mechanical backlash.
Drag force inducted vibration
Etc.
All these considered, I just go with digital content and with the highest format possible.
I would be very interested to know if Toole ever did subjective tests of tube amps or other electronics as part of his research. It's possible this testing happened and the results got buried, either because Harman doesn't make tube amps or because they didn't see any benefit to challenging audiophiles with data that conflicts with their prejudices.
Sorry, must disagree with the bit depth and sample rate thing. Redbook standard is fine, is capable of superbly good sound - digital falls down because of poor implementation of the playback electronics, as in some of the things you mentioned. Get everything right, then digital needs no excuses, whatsoever ...
I think everyone has different experience regarding bit depth and sample rate. A friend of mine whom used to be into recording felt bit depth made the most difference, some swear that DSD is best, some tell me 88.2kHz is best, some hear a difference but prefers redbook standard. For me, 192khz 24bit hits the sweet spot, others are acceptable.
While lately many conspiracy theories turned out to be true, I don't think Tooles case is it. 'Objective' subjective test results boil down to differences in output impedance, so no one is interested to further study this issue, not that there are buried 'results'.
What can be hypothesized is that higher order nonlinearities are more audible, so that crossover distortion in electronics is where we should concentrate at. But mostly cheap chipamps get this correct without resorting to exotic designs. Well if someone has the time and finances, he could make a study involving badly made class AB amp for low anchor, a common design that still has rising distortion with decreased level and a 'clean' design. To make the test more sensitive use speakers with high voltage sensitivity. Wonder if even the low anchor can be distinguished reliably. There are bigger fish to fry in acoustics.
Has anyone in DIY even measured high-order nonlinearities apart from Geddes?
Last edited:
Yeah. If the recordings are produced with the best equipment available today right through to the reproduction at the venue it can be fabulous. And digital is so convenient. Problem is where audio designers go mad with device complexity, and numbers that what the digital can produce is noticeably inferior. Studio engineers have perhaps too much scope to add devices like mixers and all the other paraphernalia, with all the mixing and manipulating the recording, it only makes it worse. This is the norm.
Absolute equalisation is not really necessary as some recording are themselves poorly balanced.
Any uncompressed 24/96 format like FLAC ALAC or even 16/44.1 can do great sound as you infer. But MP3 irritates me but some of that may be down to poor recording.
Hard to beat a guitar or other live instrument played directly through a good analogue system. i.e no recording chain.
Vinyl can be good but my revealing analogue SS reproduction system is showing serious defects in many old record pressings. Only some exceptional vinyl cuts it, and that was proven true in the days of Koetsu phono cartridges.