May I respectfully remind you that calculating the value of a link capacitor for a cut-off frequency of 20 or 20,000Hz at -3dB is a bit far from what one can expect from a "High Fidelity" system.
The phase linearity is something that makes a difference.
Some people are interested to find out if a difference exists between two DUTs, other people are interested to find out if they prefer one DUT over the other (or are interested to find out if other people do prefer one of the..).
Sometimes they want to know if a specific difference can be detected.
Sometimes experimenters are interested what the population hears (or what subgroups hear).
And further it depends if they are exploring certain properties of the hearing sense (often related to the physiology reasons) and are therefore in most case unidimensional tests.
Otoh we are usually discussing multidimensional experiences and that is in reality a bit more complicated.
After settling the research question the experimenter has to design an experiment to approach that question in the best possible way.
If interested in population parameters he has to draw a sample from that population in the most random way possible.
If interested if a difference between two DUTs exists in principle it might be sufficient to find a single human that can differentiate between the DUTs.
If interested in the abilites of a specific person he would only test that person.
Furthermore it depends if he is searching for universial effects (under every condition in every reproduction system with every piece of music) or for subsets.
And within he choose one of the available test protocols; ABX is one of these and it means that the test participants listens to "A" then listens to "B" and finally listens to "X" (which can be either "A" or "B") and has to correctly identify it as "A" and "B" .
In the original version of this test there was no chance to repeatedly listen to the three options while it is at the discretion of the test participant how often he wants to listen to them.
But it is important to remember that it is only an ABX test if he listen first to "A" then to "B" and then finally to "X" to decide.
Another test protocol is the A/B test - often named as paired comparison - where only two items were presented. An A/B test can be a "same/different" test - when all four possible combinations "AB" "AA" BB" and "BA" are presented - and the participant has to answer if "same" or "different".
It can be a preference test - only "A" and "B" are presented in a random order" - and the participant has to express his preference.
It can be a discrimination test - again only "A" and "B" are presented in a random order - and the participant has either to identify "A" or "B" for a correct response or has to detect a certain difference like "more treble" or "more bass" .
There a several more (like ABC/HR often used in the assessment of lossy compression codecs) and all have specific advantages or drawbacks.
Beside the general interestests and quality criteria that should be met, there are several others that might be of interest, like efficiency for example, that could matter.
And the experimenter has to consider that every test situation and even specific test protocols can have an influence on the results.
Wrt the ABX test protocol it is known (since the ~1950s) that participants function within the ABX worse than in an A/B test.
In food tests there were some comparison done between different test protocols in the 1990s and the proportion of correct responses were higher in A/B tests as in ABX tests. (in short, links to the various publications i´ve posted in a "DAC listening test" thread).
Reasons could be (in our models of the internal judgement processes) that the ABX method is mentally more involving/stressing and that it triggers a different internal decision process than an A/B .
In any case an experimenter should not expect that every particpant in a controlled listening test will be function at his best and should therefore provide countermeasures like accomodation time, training and (mandatory) use positive controls and negative as well.
I hope that helps a bit.
Assuming that Jakob2 is right is a good choice as he most likely is....
I tend to agree, but what do I know?Assuming that Jakob2 is right is a good choice as he most likely is....
A lot more now thanks to a clear explanation 
I tend to agree, but what do I know?
You´re welcome.
Sorry for the numerous typos and a sentence needs clarification as the important part is missing:
"In the original version of this test there was no chance to repeatedly listen to the three options while it is at the discretion of the test participant how often he wants to listen to them."
It should be:
In the original version - invented around 1950 - there was no chance to repeatedly listen to the three options while in the modern/todays variant it is at the discretion of the test participant how often he want to listen repeatedly.
Did you mean a FM tuner ?
This thing with a 19kHz carrier that carry the stereo signal ?
It is nice to get some background music, or in a car stereo, but, in a hifi system ?
Who cares about ceramic caps distortions in a FM tuner ?
Ah, Ok, in europa, we call this DAB+, not FM. And it is quite new: very little coverage.
But tell-me, John would listen to a digital signal? ;-)
And likely used a software defined radio borrowed from a different product. This isn't a high budget product that worthy is drawing ire over suboptimal output coupling, just replace them and move on. It's more a seemingly great product out punching it's weight, which makes the rant even more confusing.
Very different to DAB as is in the same band as the FM and autoswitches. In a way a much better idea.
Tornesol, I hear/understand what you are saying......broadcast FM ain't HiFi but it can be pretty darn decent nonetheless, and can happily inspire new playlists/musical interests.
Upstream passive and active devices cause all manner of linear distortion and
non linear distortion behaviours, the final output stage active and passive device behaviours constitute a 'final filter' as does the input device constitute an 'input filter' for the system.
Broadcast FM can be pleasing/pleasant and nicely close to the source CD, but yes it will never be quite the same.
Dan.
Some of the best sound I ever heard were 15 ips uncompressed tapes played over FM radio. As we often say here the source quality is foremost.
The 19k is the pilot tone for generating the 38k subcarrier. IIRC the absolute best SNR you can get from the system was low 70's but I might not recall correctly. Typically the 19k is notched out if not a small percentage of the population find FM unlistenable.
The audio is cut at 15kHz with a very steep filter (100 dB/oct roll-off and 1000's of degree phase shifts) to avoid interference with the 19kHz pilot.
I once did DB listening tests on such a filter with a few guys; 3 out of 4 could reliably identify whether it was in the chain, and they preferred it with the filter in ...
Jan
I never though of that, the fuzz pedal guys hide their stashes of NOS 1N62 Ge diodes like gold.
On my system, FM radio is just awful (trebles). While it sound ok in little speakers.
It is just like the sound of TV: I was obliged to buy a little system for my TV, in my Hifi it was a nonsense: You just listen to hundred of defects, noises, etc...
And the "size" of the sound is ten time the one of the (quite big) screen ;-)
No way.
My FM tuner is somewhere in my cellar.
Hi Dan
I certainly would not use them for coupling audio signals and I am clearly of the best cap is no cap when it comes to signal coupling - but how often do you get away without that, not often. My phono stage design, no coupling caps at all, ultra LF servo though.
Power supply is a different matter, but only to power digital (switching) circuits. Here I use 0.33F/5.5V caps in parallel with largish ceramic layer caps, something I would never do for signal coupling, not in a million years. So 'horses for courses' apply. I only use this on DACs (usually 2 or more are needed, typically 3.3V and 5V) and would love others to try it and report back. Do it also on the 3.3V powered clocks.
I am sure you have something there you could try this on. I won't try do describe the sound, just find out for yourself.
Cheap as chips!
This I believe is about back-EMF noise suppression (switching), and with the clock supply, keeping ULF noise out of the clock.
Joe
Last edited:
Yes Joe just a comment on my observations, I sense a definite increase in discomfort when someone is in a situation where they realize that they are confronted with a decision and they simply "don't know" that is some process or clue or whatever you want to call it is absent. That is why IMO folks that are used to casual sighted comparisons have strong reactions to suddenly going "blind".
- Status
- Not open for further replies.