Method for a double blind test with details, please!

Archived from groups: rec.audio.tech (More info?)

"Per Stromgren" wrote ...
>I would like to test interconnect cables and other line level devices
> on a hobby basis. I do know what a double blind test is, in theory,
> but I how do I do this in practice? What kind of devices do I need to
> perform a double blind test, if I want to do this more automatically
> than to walk into the room, disconnect the device under test, throw a
> dice and reconnect it again? Do we need to be three persons to make it
> double blind?

I would use one of those "dice" or "roulette" circuits to generate
a random state (0 or 1) and then use it to flip the relay(s) one way
or the other. But then I wouldn't expect to hear any difference between
any decent "interconnect cables".

> Also, I don't have a good millivoltmeter, only a very standard DVM.
> Will this do when calibrating levels between two devices? Could I
> perhaps use some piece of software plus an descent soundcard on my
> laptop? I figure that what is important is that the meter reads the
> same value for identical signals, not that it has flat frequency
> response.

That would be my assumption. Have you looked at any of Arny's
info about ABX testing?

> Until, and if, I find this very exciting, I am on a strict budget.

Good luck. Don't waste to much time on cables, though.

Archived from groups: rec.audio.tech (More info?)

In <n5jk115eejt4a7qtv2ga9n3f5p1qb8pmnf@4ax.com>, on 02/21/05
at 10:12 PM, Per Stromgren <per.stromgren@telia.com> said:

>I would like to test interconnect cables and other line level devices
>on a hobby basis. I do know what a double blind test is, in theory,
>but I how do I do this in practice? What kind of devices do I need to
>perform a double blind test, if I want to do this more automatically
>than to walk into the room, disconnect the device under test, throw a
>dice and reconnect it again? Do we need to be three persons to make it
>double blind?

[ ... ]

From an experimental design standpoint, an ABX box makes a lot of
sense. I do wonder, however, why the ABX box is the only trusted
"neutral" device available.

In my travels, I've found that the ABX technology masks very subtile
differences that might otherwise be perceptible.

Our cable testing experimental design was not one that could lead to
publication, but it was illuminating. We had one set of "tricky" cables
that could migrate randomly in a store full of systems. Any staff
person could move the cable at any time, without telling anyone. (and
the cable was never visible -- one had to crawl over the equipment to
find it).

We all had the same reaction. We get to know how each system sounds. We
would all notice that a particular system would perk up a bit, check,
and find that the cable had landed there. Some days later the system
would go back to its former sound and on checking we'd find that the
cable had been removed. Before long we'd notice another system had
perked up, and verified that the cable had landed. Only the staff
person who last moved the cable knew where it was. (the cable was moved
on a whim)

We fussed with the cable for over a year in this fashion. We are a very
tough lot and manufacturers really hate to bring stuff around for us to
evaluate, because we can be pretty non impressed with the latest and
greatest. Our cable verdict was unanimous. We liked it and we could
always determine where it was strictly by blind listening.

Since then we've expanded to more "tricky" cables and we don't find
that they all have easily noticeable benefits, but some definitely do.

--

Be careful in your travels because at least half of the magic of
"tricky" cables happens when you accidentally clean the connections
while replacing the cables.

--

It's a sad state of affairs, but we find that the garden variety of
$5.00 interconnect has a much better long term reliability record than
the "tricky" cables. By reliability, we mean less broken internal
connections. The "tricky" cables are almost always better in many other
ways, particularly in preventing interference, but they do break more
often.

-----------------------------------------------------------
spam: uce@ftc.gov
wordgame:123(abc):<14 9 20 5 2 9 18 4 at 22 15 9 3 5 14 5 20 dot 3 15
13> (Barry Mann)
[sorry about the puzzle, spammers are ruining my mailbox]
-----------------------------------------------------------

Archived from groups: rec.audio.tech (More info?)

On Mon, 21 Feb 2005 21:28:37 -0800, "Richard Crowley"
<rcrowley7@xprt.net> wrote:

>I would use one of those "dice" or "roulette" circuits to generate
>a random state (0 or 1) and then use it to flip the relay(s) one way
>or the other. But then I wouldn't expect to hear any difference between
>any decent "interconnect cables".

I know. But my friend doesn't...

>That would be my assumption. Have you looked at any of Arny's
>info about ABX testing?

Yep.

>
>> Until, and if, I find this very exciting, I am on a strict budget.
>
>Good luck. Don't waste to much time on cables, though.

Don't worry. As I said, I'm trying to convince a friend to save his
money for more important parts of his system, i.e. the loudspeakers.

Per.

Archived from groups: rec.audio.tech (More info?)

Per,

Rod Elliot has instructions for building an ABX comparator here:

http://sound.westhost.com/projects-9.htm

Archived from groups: rec.audio.tech (More info?)

"Per Stromgren" <per.stromgren@telia.com> wrote in message
news:n5jk115eejt4a7qtv2ga9n3f5p1qb8pmnf@4ax.com...
>I would like to test interconnect cables and other line level devices
> on a hobby basis. I do know what a double blind test is, in theory,
> but I how do I do this in practice? What kind of devices do I need to
> perform a double blind test, if I want to do this more automatically
> than to walk into the room, disconnect the device under test, throw a
> dice and reconnect it again? Do we need to be three persons to make it
> double blind?
>
> Also, I don't have a good millivoltmeter, only a very standard DVM.
> Will this do when calibrating levels between two devices? Could I
> perhaps use some piece of software plus an descent soundcard on my
> laptop? I figure that what is important is that the meter reads the
> same value for identical signals, not that it has flat frequency
> response.

I've done this before. Interconnects are easy, since there is no need to
check levels. If you want to do a perfect job, it takes 4 people: The
testee, someone to proctor the testee, the testor, and someone to watch over
the testor to make certain he did what he's supposed to do. If the DBT is
strictly for your own information, you don't need to be proctored, so it's
down to 3 people. If you trust the person running the test, you can get
away with a SBT. Since I've never run across anyone that can identify
interconnects in a SBT, I haven't needed to go to DBT. IOW, if you can't
pass a SBT, you surely can't pass a DBT! So here's how you do it:

Arrange your system so that you can't see the interconnect.
Listen to both interconnects until you think you can tell which one is
connected by sound alone.
Now run the test in this fashion:
a. You leave the room. There must be no contact between the testor and
the testee during the entire period of the test.
b. Your assistant enters the room and disconnects the interconnect.
c. Assistant flips coin, and connects the appropriate interconnect.
d. Assistant leaves room and you enter room.
e. You listen and write down which interconnect is in use.
f. Repeat 16 times.
g. If you're right (or wrong) 12 times, there may be something there.
Run the test over again. Note that guessing wrong is just as statistically
significant as guessing right.

It's important to disconnect the interconnect BEFORE flipping the coin.
IOW, the assistant has to go through exactly the same procedure every time.
The first time I ran one of these tests I made the mistake of flipping
first, and just leaving the cable connected if there was no change. This
was a crucial error.

Norm Strong

Archived from groups: rec.audio.tech (More info?)

"Per Stromgren" <per.stromgren@telia.com> wrote in message
news:seom11l2dhnguuc7m28h40c7ifplbcn0uv@4ax.com...
> >Probably yes -at least in terms of level matching. Beware that standard
DVMs
> >often only have frequency response up to 1000 Hz or even less.
>
> Even deaf above? Could it be that bad? I thought they just could not
> guarantee the tolerance above some upper frequency.

Both! Even meters like the Fluke 75 have very poor response above a few kHz.
High frequency response was simply not a design requirement.
OTOH some meters are designed to go to VHF or even UHF frequencies.
It's all covered in the specs from any reputable manufacturer.

MrT.