Interesting link, thanks. It's not entirely clear what "straight wire loop means" or how the noise pickup occurs in SY's measurements but, yeah, it'd take 50dB CMRR in that case to put the mains grass to measurement limit. The extra 10dB needed over 40dB CMRR would likely be worthwhile with typical amplifier gain structures.
There was absolutely NO SARCASM in what I said.
I spent more than 20 yrs designing stuff for commercial HiFi as head of R&D.
I used DBLTs extensively as I have this naive believe that making something sound better helps sales .. and this proved very successful for us. As I said, the Man (and Woman) in the street is actually quite perceptive .. much more perceptive than the Golden Pinnae.
But generating solid BS for Marketing was an important part of my job .. as was entertaining HiFi reviewers .. many (most?) of which were deaf.
But Marketing objected to some of my monikers for our advanced tech. eg Powered Integrated Super Sub technology which was my greatest contribution to the State of the Art and valuable today.
There are good audible, technical & measureable differences due to bi-wiring bla bla. But often these are NOT what the speaker designer wanted. They may make things better but also worse.
Some of this is addressed by my Powered Integrated Super Stuff .. which applied to the rest of the frequency range too.
The "noise" in SY's plots is all 60 Hz-related. My guess is that it's either related to 60 Hz (and harmonics) residuals present on the B+ and heater supplies or a ground loop in the test setup. The improvement from the input transformer is most obvious on the 120, 180, and 300 Hz components, which points to a ground loop issue.
By "straight wire loop", I'm guessing SY means a loop-back test. I.e. cable from generator output to analyzer input.
Twest820 can speak for himself if needed, but I think what he meant was that his sarcasm detector needed to be recalibrated, as he either misread your straight comments as being sarcastic or vise versa. No worries. With language barriers, lack of visual feedback, etc., these things happen. It's amazing internet forums work as well as they do given those deficiencies.
In the future, I'm sure international conflicts will arise because of the mistranslations of Google Translate. 🙂
~Tom
I think it means signal out on sound card looped back to the analog input. What would normally be done as a cal measurement.
The power supplies are pretty clean, so I think all the noise is on the PC soundcard output. A good demonstration why AP stuff is worth the asking price 🙂
Still there is a definite market for a small board with a THAT1200 on to add a balanced input to anything, pre or power...
That would be mine as well. Unless the grounding within the Audigy is horrible the loopback you and Bill are thinking of shouldn't have mains grass---it's my understanding the Audigys are pretty decent. More likely straight wire means connecting +in of the preamp to the out pin with a wire, effectively shorting the preamp circuitry out of circuit but leaving the ground structure more or less in place. Someone could always shoot SY a PM and ask.
Probably no need to go that far, actually; seems equivalent results can be obtained from a smaller trained cohort.
Which, depending on design intent, may or may not be a good thing. Attempting a biamp/triamp conversion without some amount of level management and EQ capability isn't something I'd recommend. With the low cost per channel of chipamps, inexpensive availability of calibrated measurement mics, and more or less infinitely configurable DSP+DAC to the power amp arrangements it's much easier than it used to be. The worst case is one ends up putting things back they way they were so, if one thinks ahead a little and takes a few snaps of the crossovers and such with the mobile before disassembly, the risk is well contained.
Yeah, we tried other bodily functions and marketing didn't go for those either. 😉
That's awesome. One of the groups at work used names of local rivers for their projects. This went well until the Nooksack project. For some reason corporate balked at that one... 🙂
~Tom
I agree. There're definitely good arguments for and performance to be gained by using differential signaling.
~Tom
Whenever I read these descriptions of "how to listen" I go into a frenzy of 🙄's - because they are so close to being completely useless as a means of assessing the important qualities of a playback system, or component within. They pay almost zero attention to learning how to recognise the markers of typical misbehaviour of audio systems - the attributes which most react to, that cause kgrlee's Man in the Street to give a thumbs down to what they're hearing ...
Really? Sean Olive's article (linked by twest820, quoted in Post #569) describes the various misbehaviors he trains his listeners to identify. This includes frequency bands, various distortion mechanisms, dynamic range limitations, etc. Which of Mr. Olive's characteristics do you find superfluous and which would you like to add in order for the list of characteristics to be more in line with your personal preferences?
~Tom
To break it up, firstly I wouldn't concern myself with frequency bands, this is a straightforward, solvable problem using realtime DSP, or via prior processing - FR is not an issue for me, but for those whom it is, these and similar methods "resolves" that. Of course, flattening the response may then emphasise problems elsewhere, non-linear distortion issues, but that is not FR.
Then, what Sean identifies as distortion is
These are very distinct and easily identifiable, or non-specified "gross distortion" qualities - again there are known, readily applied methods of resolving these.
The interesting areas are "categories of timbre, spatial, dynamic" - IME, these are elements all associated with distortion; they vary depending on the audible level of various distortion mechanisms. However, Sean wants the listener to rate these in subjective evaluation terms, similar to that of wine flavours; whereas to me these are faults, as in how vinegary or musty is the sample - the question should be, how little is the playback tainted by these faults.
Then, what Sean identifies as distortion is
These are very distinct and easily identifiable, or non-specified "gross distortion" qualities - again there are known, readily applied methods of resolving these.
The interesting areas are "categories of timbre, spatial, dynamic" - IME, these are elements all associated with distortion; they vary depending on the audible level of various distortion mechanisms. However, Sean wants the listener to rate these in subjective evaluation terms, similar to that of wine flavours; whereas to me these are faults, as in how vinegary or musty is the sample - the question should be, how little is the playback tainted by these faults.
While the best ears I've tested have been recording engineers & speaker designers (with one exception), there are deaf recording engineers & speaker designers too.
There is no need to go to the lengths of the false prophets Floyd & Olive. Much of their stuff is cribbed from Fryer AES E-Library Absolute Listening Tests-Further Progress & AES E-Library Intermodulation Distortion Listening Tests
There's stuff in the Olive et al method which hinder obtaining statistically significant results quickly. eg
- I never do ABX tests but ABC tests. Which also allows you to 'calibrate' your panel.
- Each Listener is tested in isolation
- He chooses his own music and the playback levels
- The Listener is NEVER told what he is listening too.
- I only ask the Listener to rate each presentation from 1 - 10 and comment. (You see examples of trying to 'train' listeners in the first AES paper. We abandoned that very early on.)
I don't want to add to his less than useful list but get rid of it completely.
As an important example, the speaker that has come out best in nearly 2 decades of DBLTs is a small ported box with a LF roll-off of 70Hz. But it consistently attracts comments of 'tuneful and articulate bass' from experts & novice alike. The experts might pick up that it has restricted LF but still like the bass and mark it highly on their favourite music.
Last edited:
Take note when the voice from the kitchen speaks.
Lowered bass driver distortion...harmonic/imd/doppler.
My Behringers can sound 'better' with switchable bass roll out...it's a choice of having the bottom end stuff, or missing out on the low bass and getting cleaner mid bass/mids.
Provided the porting tuning is musically right (chordant), the average pop music/tv listener will prefer this lower distortion option.
The key here is ''on their favourite music''.
Key for me also.
Dan.
well I have a pair of very marmite speakers (apogee centaur minors) which are pernickety enough to annoy me on occasion, so I know they would not win any DBLT. And no point doing DBLT on my current power amp vs the Mod-86 as I could tell which is which without a note being played.
So I will be sighted and biased.
So I will be sighted and biased.
I'm sure you will be very happy with your stuff as long as no notes are played.
But some of us use the gear to play music and prefer to enjoy rather than be annoyed. I confess. I'm very biased in favour of gear that allows me to do this. 🙂
you misunderstand. Current amp has a residual hum I have never been able to clear and have put down to a transformer that was borderline for 60Hz and doesn't like 50Hz (despite being an export model). So I could never ABX, so I won't bother. It was only a stop gap which has ended up being in place for 18 years. Cheaper to build something better than try and fail to fix something widely accepted as suboptimal.
Ref the annoyance, that is the sweet spot. But I can live with that for what it does well.
Ref the annoyance, that is the sweet spot. But I can live with that for what it does well.
There was great discussion awhile back about how to pinpoint what sonic attributes meant the most to good sound right here. Some of you posted in that thread so it won't be news to you but what did you think about that guys methodology?
I want my power AMP to be as neutral as can be. Nothing added or taken away as far as we can tell. Everything else I can do at line level or with room changes. But that's just my view (and why I like the Mod-86 concept).
That's what we all want in theory but how do we know we have it? Which measurements are the ones that matter the most? I could be wrong but I think that was the point that former member was trying to make. Of course, just about all the distortions are ridiculously low with this amp so maybe Tom has come up with the ultimate answer after all 🙂
Tom, have you done any time distortion analysis on your modulus? More specifically, I'd like to see a close up of the A to B signal crossover, a more thorough slew analysis (both up and down) and any asymmetrical signal crosstalk from using a common supply.
Tom, have you done any time distortion analysis on your modulus? More specifically, I'd like to see a close up of the A to B signal crossover, a more thorough slew analysis (both up and down) and any asymmetrical signal crosstalk from using a common supply.
Absolutely. Lots of time domain measurements.
The slew rate measures to 14 V/µs with the amp loaded by 8 Ω || 1 nF. The slew rate is symmetric - i.e. 14 V/µs up and 14 V/µs down.
I optimized the stability components in the time domain after getting close to optimum in the frequency domain (measuring the loop gain with a network analyzer). The compensation networks were tweaked for the best transient response and best overdrive/clipping response.
I'm not sure what you mean by "A to B signal crossover". Are you talking about the regular class AB crossover distortion? Or the distortion that arises when the amp leaves class A and enters class AB (i.e. the output current exceeds the quiescent current of the output stage)?
I have swept the THD+N vs Power from mW to full power. There were no surprises there. The harmonics are well below the noise floor at the low output powers, so there isn't any distortion to look at really...
At 0.00018 % THD, you won't be able to resolve any of the distortion on an oscilloscope. The APx525 is pushed to the limit to resolve it.
I have not done crosstalk yet as I don't have a stereo amp built. I have chosen to prioritize customer orders and inquiries higher than building a stereo amp.
My plan is actually to build a flea power amp using the Modulus-86 boards. Flea-86? 🙂 There are a couple of reasons for this.
1) I have some 12 V, 50 VA transformers lying around.
2) I need better audio for background noise than the built-in speaker in my Mac Mini can provide.
3) I want to try the Mark Audio Alpair drivers and Martin King's MathCad scripts out in practice.
So my first Modulus-86 build will be a 15-16ish W rig intended for driving a pair of Alpair 6P in a mMar-Ken6 enclosure (take the port spacers out for the P version of the driver).
My high-power build will be next. I'm thinking to use a composite amplifier with multiple LM3886'es (or channels of an LM4780) in parallel. Such an amp would be useful to me for multiple reasons.
Stay tuned. Remain patient. Good things will come to those who wait. 🙂
~Tom
PS: Re. listening tests: I prefer to use my own music collection for that as well.