Lundahl Shorty ( story or naughty ! )

Selecting an output transformer

by Bill May

Your average audiophile is lucky if he gets to hear two or three
output transformers before he makes his choice. It probably
takes him a lot of time and organization to hear that many.

So how does he make his choice? There are only three ways:
listening to enough transformers under controlled conditions to
make a choice, buying on measurements, and buying on
recommendation.

Listening to one or two or three isn't really enough. For the test to
be meaningful, the transformers must be installed in duplicate
amps, or at least in very similar amps. For the choice to be
meaningful, the transformers must be installed in an amp very
similar to what the user will build.

He can study manufacturer's spec sheets. Some manufacturers
lie, or take their measurements in such a way that they are not
truly relevant to any real-life amplifier. That's before the hype
even starts. Those who read the hype can easily be misled into
thinking there is a single figure of merit for a transformer. It can
be bandwidth, low bass extension, power handling. It isn't true.
Any transformer that will sound good has all its desirable
technical features in balance.

The hype includes the price and the reputation of the
transformer. That makes sighted tests suspect to any engineer.

He can ask for recommendations. The problem is that those who
recommend a particular transformer may have heard only that
transformer, or perhaps one other, and possibly not in suitable
amplifiers. Or the recommendation may be informed by nothing
more than spec sheets and hype.

In the end, only relevant, blind listening tests count.

This doesn't apply only to the average audiophile. Even
professionals can be taken in!

Here is an example of how measurements, hype and sighted
tests can produce different results These tests were conducted to
choose the output transformers to be used in an amplifier Real
McCoy Audio was designing for Japanese manufacture. The
manufacturer intended us to choose between the six most
expensive Japanese transformers. The cheapest Tango, the
Swedish Lundahl, American Magnequests and British Audio
Note UK transformers were added as statistical controls, and the
no-name potted Chinese and open-frame Russian ones as
placebos.

The electrical ratings are for weighted results of measurements
taken in circuits optimized to the transformers. In the sighted
tests knowledgeable listeners were permitted to see the
transformers and were told their relative prices, if they asked. I
have included a price scale, on which 1 is expensive and 5 is
cheap. In the blind tests no-one in contact with the listening
panel knew which amp was playing. The sighted tests were
conducted after the blind tests. Several of the tests were
repeated with a different listening group to confirm the result.

Table reads in 12 pt Courier.

RATING ELECTRICAL LISTENING TEST LISTENING TEST
- MEASUREMENT SIGHTED BLIND
1 Tango 1 Tango 1 Lundahl 3
2 Lundahl 3 Tango 2 Nature Snd 1
3 Tamura 1 Tamura 1 Tango 2
4 Nature Snd 1 Nature Snd 1 Tamura 1
5 Nature Snd 2 Magnequest 1 Tango 1
6 Tamura 2 Nature Snd 2 Nature Snd 2
7 Tango 2 Lundahl 3 Audio N UK 4
8 Magnequest 1 Tamura 2 Tamura 2
9 Audio N UK 4 Audio N UK 4 Russian 4
10 Magnequest 2 Magnequest 2 Magnequest 1
11 Tango 3 Tango 3 Tango 3
12 Chinese 5 Russian 4 Chinese 5
13 Russian 4 Chinese 5 Magnequest 2

The Measurements column holds no surprises. On
measurement, all of these except the Russian transformer are
good performers. Some audiophiles scoff at the idea that a set of
electrical measurements may predict the outcome of taste tests
with certainty. That is true. But if one lowers the expectation a
little, correctly weighted measurements will at least eliminate
components on which further time should not be wasted. That
proved the case here. Of the bottom five in the measurements
rating, only one transformer is not in the bottom five in the
blind listening tests.

The Blind Listening Tests in the last column hold several
surprises. A comparison with the Measurements table shows that
the best-measuring transformer does not always sound the best.
Price comparisons show that the most expensive transformer
does not always sound the best, even within the same brand. The
high rating of the Russian transformer shows that precision is not
as highly valued among the musically inclined as engineers
would prefer. The Russian transformer rose so far above its
measurements in the listening tests because its inaccuracies are
aurally pleasing in the "presence range". One of the listening
panel noted in his Comments block: "Who cares about accuracy
when you can have ecstacy."

The Sighted Listening column shows the effect of brand name,
hype, and possibly of appearance. Comparison with the Blind
Listening column shows that some reputations are earned and
some are not. The same comparison shows that an industrial
physical appearance can depress the sighted rating of a
transformer like the Lundahl which measures exceptionally well
and in the blind listening tests came first. Equally hype can
boost the sighted rating of a transformer, like both the Magnequests,
well beyond its rating in either measurements or on blind
listening. A comparison between blind and sighted ratings of the
Japanese transformers is interesting. It shows that even where
appearance is not a consideration, and the reputation is earned,
knowledge of relative price can distort the rating.

The chief facts to emerge from these tests are that with regard to
selecting output transformers for listening pleasure and fidelity:

*data sheet measurements and reputation must be
supplemented by other means to predict listening pleasure

*careful weighting of measurements on hand of experience goes
some way towards an initial rough selection

*sighted listening tests are too subject to the effects of reputation
and hype to be used as a predictor of listening pleasure

*blind listening tests prove that some reputations are earned

*blind listening tests prove that a high reputation achieved by
hype is no guarantee of listening pleasure

*blind listening tests prove that high price is no guarantee of
listening pleasure

*blind listening tests prove that an industrial appearance, if kept
out of sight, is no bar to listening pleasure

*blind listening tests prove that a modest price is no bar to
listening pleasure

*success in blind listening tests is not always a guarantee of
fidelity in reproduction

*blind listening tests should therefore always be used

*but only in conjunction with careful interpretation of the
measurements

Incidentally, after examining our results the Japanese
manufacturer chose the mid-price Tango for sonic, marketing
and financial reasons.

****The late Bill May spent forty years in high tension electricity
supply. A long-time tube hi-fi hobbyist, he was then technical
director of Real McCoy Audio, a design and prototyping shop.

(c) Copyright 1998 William May and Real McCoy Audio
This article may be reproduced free of charge for DIY and amateur
use as long as the entire article including the copyright notice and
this permision is rendered complete and unaltered.

for info email me at sales@diyparadiso.com