[Icom] Radio Shack 1:1 Isolation Transformers (long)
George, W5YR
[email protected]
Sun, 07 Jul 2002 18:17:16 -0500
I have posted this to all lists which I believe have subscribers who either
use the ubiquitous little RS isolation transformers or who have had them
recommended as a means for avoiding noise, hum, instability and other audio
system problems.
My intent is to present the results of frequency response studies which may
be useful in determining suitable applications for these units.
The subject transformer, RS Stock No. 273-1374, is a 1:1 isolation
transformer intended for use in circuits with an impedance level of from
600 to 900 ohms, source and/or load. The frequency response is given as
300-5000 Hz; presumably these limits are measured at the -3 dB points of
the response function, but
this is not stated. No spec is given for power rating or signal amplitude
rating, or for insertion loss.
These transformers are small, easily mounted, inexpensive and readily
available from almost any Radio Shack store. They are used in PSK31 circles
for isolating the computer soundcard from the radio. I have been using
these units for some time now with evident success. But, I have always
wondered how well the transformers actually performed in this application.
Besides soundcard/radio isolation for PSK31, another application for
isolation transformers in amateur radio is found in the interconnection of
various audio devices and components. The requirements for this
application can be much more demanding than the PSK31 usage.
For PSK31, and other digital audio modes, the transformers are required to
pass audio frequencies in the range of from a few hundred Hertz to as much
as 3000 - 4000 Hz. These are the frequencies found in the usual waterfall
display
on a modern receiver using current programs such as DigiPan, MixW, WinPSK,
etc.
Signal levels are intentionally kept low in this application to avoid
overdriving either the soundcard input stage(s) and/or the soundcard output
stage(s) and the transmitter audio input stages.
Looking at these requirements and the transformer specs, one concludes that
they are well matched to the application. Frequency range is adequate as
long as one doesn't go too low on the waterfall. Since limited signal
amplitudes are usually employed, core saturation and the resultant
distortion should not be a problem. Few receivers have IF filter responses
extending beyond 4000 Hz baseband so the high-frequency response is not a
problem. Although insertion loss is not specified, there is an abundance of
signal available from both the soundcard and the radio so loss is not a
consideration.
The one area where a potential problem rears its head in PSK31 is the use
of a
transmit or receive frequency near the low-frequency side of the waterfall,
say around
500 Hz. The problem is twofold: we are approaching the lower rated
frequency of the transformer and the output of many SSB transmitters starts
falling off with frequency below about 500 Hz.
However, this problem is self-limiting since we tend to avoid the lower
frequencies where our output power falls off and where any deficiencies in
the transformer could introduce distortion into the received signals. We
simply retune the radio to place these signals higher in the passband.
Conclusion: The 273-1374 is an adequate, inexpensive and effective means
for isolating audio circuits in the PSK31 application and probably most
other digital audio applications. Precautions are to refrain from using
frequencies below about 500 Hz and to use minimal signal levels.
Unfortunately, the other common application for isolation transformers -
isolation among audio devices - is less well served by these transformers.
For conventional SSB operation, where an audio passband of from about 350
Hz to about 2700 Hz is typical and effective, the frequency response of the
units appears to be adequate. For "hi-fi" SSB applications where a
transmitted bandwidth of from 60-80 Hz to 3000-4000 Hz may be used, we
immediately see a discrepancy between low-end frequency response and
desired performance. The high-end response appears to be adequate. Strike
One against the RS unit for hi-fi audio applications.
The other questionable performance area is their signal-handling
capability: how large can the signals be before the transformer cores are
driven into non-linear operation with attendant harmonic and
inter-modulation distortion products being developed? As one would expect,
this requirement is tied directly to operating frequency.
From test results that I will present later, it turns out that these
transformers can be used at frequencies lower than their specified range
if, and only if, signals are held to remarkably low levels which may not be
of practical use.
Although it is a requirement to use the transformers only in
low-signal-level applications, Radio Shack does not spec power or voltage
handling capabilities for these units.
It was the absence of this data and my own curiosity as to how well the
transformers were performing in isolating my soundcard from my outboard
audio equipment that I performed a series of tests to (a) measure the
response
function and (b) determine the maximum signal levels at spot frequencies
for
a minimum amount of distortion. Since I use my soundcard for a number of
audio analysis programs, I was concerned also as to the effect the
transformers might be having upon the results of these programs.
Anyone who has seen these units has noticed their small size, especially
the actual cores. One immediately suspects that core saturation will be an
immediate problem when attempting to operate with low frequencies at
significant signal levels.
Clearly, these are not pro-audio transformers which typically offer
frequency response from 10 Hz to 250 KHz with overall distortion of 0.001%
or less at normal pro-audio line levels. Any attempt to use the Radio Shack
units in place of a pro unit in "hi-fi" applications will likely yield
unsatisfactory results unless signal levels are kept exceptionally low and
low-frequency operation is severely limited.
The first examination I made of these units was simply to feed the primary
from a HP 201C audio oscillator with a 680 ohm series resistor to simulate
operation with a 600-900 ohm source. The secondary output was measured by
an HP 400L a-c rms electronic voltmeter with a 680 ohm resistor bridging
the input to terminate the transformer.
After running several response curves at various signal levels, I was
amazed at the excellent wideband performance of these little transformers.
I obtained almost flat response down to 20 Hz or less and as high as my
instrumentation would go: 20,000 Hz. The response function seemed
reasonably independent of signal level from a few millivolts up to a volt
or so. I checked several transformers and all seemed to have about the same
characteristics.
Encouraged by these findings, I next fed pink noise from my soundcard via
one of the Line output isolation transformers to another computer running
the
SpectroGram program where I measured the response spectrum and made a
third-octave plot. Again, the response was almost perfectly flat from near
d-c to over 20 KHz! A white noise test gave the same result.
Finally, I used spot frequency sine waves generated by a soundcard program
instead of the HP oscillator, to determine the response of the transformer
by measuring the output of the transformer with the HP 400L voltmeter. And
again I was rewarded with an almost flat response from below 20 Hz to above
20,000 Hz.
Clearly something was amiss or Radio Shack was selling one of the world's
best audio transformer for four bucks! Pro-audio transformers start around
$60 and cost well over $100 for top-of-the-line units.
As you might expect, there was only one plausible explanation for this
result: the
transformer had to be generating large levels of in-band distortion
products which would account for the apparently uniform response
functions. At
the low-frequency end, below about 500 Hz as it turns out, any signal of
significant amplitude generates substantial harmonic and IM distortion
products whose presence lends to the illusion of a broad, flat response
function.
Those tiny cores extract a price in return for their small size, price and
convenience.
To confirm this explanation, I obtained a set of frequency-response data
for a transformer connected directly from the Line output of the shack
computer soundcard to the input of the HP 400L voltmeter whose input
terminals were shunted by a 560 ohm resistor to terminate the transformer.
This also allowed the voltmeter readings to be conveniently converted to
dBu.
This same signal was fed into the Line input of another computer running
the SpectroGram program. Great care was taken to ensure that neither
soundcard was being driven into
distortion by constant scope monitoring of waveforms. Spot checks made on
several transformers had indicated remarkable similarity in their
characteristics, so only one was tested thoroughly.
Prior to actually testing the transformer, it was replaced in the system
with a connector and all proposed tests carried out without the transformer
present. The results gave confidence that the data would reflect the
performance of the transformer and not some arcane aspects of the
soundcards involved.
The following table shows for each frequency the maximum signal level
allowable at the transformer output to avoid generating harmonic or IM
distortion products a specified level below the fundamental signal level.
For each frequency, the fundamental signal was scaled to -15 dB on the
plot. Then the transformer input signal was reduced until the largest
distortion product (the third harmonic) was reduced to -90 dB or 75 dB
below the fundamental. The required signal reduction is given in dB.
Numerically, this value shows how much the signal level must be reduced at
a given frequency compared to the -15 dB 1000 Hz reference level to yield
no distortion products larger than -90 dB on the plot.
The signal to distortion ratio S/D after the reduction was computed by
measuring the remaining fundamental signal component magnitude and
comparing it to a level of -90 dB. While this is a stringent test for
distortion, it shows what is required to obtain essentially distortionless
performance. In the absence of better measuring equipment, the approach was
taken that "distortionless" would be taken as meaning no visible distortion
products on the plot above the -90 dB level.
Note that the reference frequency signal level of 415 mV (projected at a
level of -15 dB on the response plot) corresponds to a level of -5.4 dBu.
Signal levels in excess of this amount at any frequency will generate
distortion products greater than 75 dB below the fundamental frequency
component. At frequencies below 1000 Hz signal levels must be taken far
below -6 dBu in order to avoid distortion.
frequency signal level output power signal reduction S/D
--------- ------------ ------------ ----------------
----------
1000 Hz (ref.) 415.0 mV 308.0 microWatts 0 dB 75 dB
800 295.0 155.4 3 72
700 290.0 150.0 3
72
600 214.0 81.8 6 69
500 133.0 31.6 10
65
400 66.0 7.8 16
59
300 52.0 4.8 18 57
100 12.3 0.3 30 45
50 5.9 0.06 52 23
20 3.2 0.02 60 15
3000 475.0 403.0 0 75
5000 452.0 365.0 0 75
7000 415.0 308.0 0 75
10,000 468.0 391.0 0
75
15,000 405.0 293.0 0 75
20,000 330.0 195.0 0
68*
-----------------------
* This decrease in S/D is due to a falloff in the signal response of the
transformer with its input signal level held constant at the reference
level. It is not the result of distortion requiring an input signal
reduction.
These data are better placed into perspective when compared with the
performance specifications of a typical modestly priced "pro-audio"
transformer, the Furman Iso-Patch IP-2B which I use on the audio side of
the shack.
While the RS units appear to be of use only when operated below about -6
dBu from 1000 Hz to 20,000 Hz, and must be severely derated below 1000 Hz,
by as much as 60 dB at 20 Hz, the IP-2B is rated at 3% THD or less at a
level of +3 dBu at 20 Hz; +10 dBu at 40 Hz and +20 dBu above 100 Hz. A
comparison of the transformers reveals that the IP-2B has a relatively huge
core compared to the RS unit and that it is wound in a totally different
manner.
In summary, these data clearly point out the price to be paid for the low
cost, small size and overall convenience of these isolation transformers.
While they will perform reasonably well from 1000 Hz to 20,000 Hz at signal
power levels in the order of -6 dBu or less, operation below 1000 Hz and
below is compromised by excessive distortion unless signal levels are
significantly reduced by up to 60 dB.
While insertion loss was not measured, casual observations during the tests
did not indicate any undue losses in the frequency range where signal
distortion was low.
One could argue that for "quick and dirty" applications, the transformers
might be acceptable for operation down to a few hundred Hz but at
frequencies as low as 100 Hz the level reductions required to prevent
distortion are so large as to render the units practically unusable. Even
at 100 Hz with a 30 dB signal level reduction the signal-to-distortion
ratio is only 45 dB, compared with the average 75 dB performance obtained
from 1000 Hz and above.
Thus, we see why the RS transformers find reasonable application in
PSK31 systems where a frequency range of from 500 to 4000 Hz is adequate
and signal levels can be kept very low.
But, for general audio isolation among audio equipment components,
especially pro-audio equipment with its relatively high line-level signals,
the transformers would appear to have no feasible application.
These tests have not been intended to provide an exhaustive accounting of
the performance of these transformers, but rather to explain and confirm
the almost obvious conclusion upon examining them that they would not be
usable at low frequencies and/or at even modest power levels.
Since I have taken a great deal of space with this presentation, if there
are further questions or comments, perhaps it would be appropriate to
continue any discussion by private postings.
73/72/oo, George W5YR - the Yellow Rose of Texas
Fairview, TX 30 mi NE of Dallas in Collin county EM13qe
Amateur Radio W5YR, in the 56th year and it just keeps getting better!
QRP-L 1373 NETXQRP 6 SOC 262 COG 8 FPQRP 404 TEN-X 11771 I-LINK 11735
Icom IC-756PRO #02121 Kachina 505 DSP #91900556 Icom IC-765 #02437