Resistive load on interstage audio transformers in tube type amplifiers

[k4kyv]

I have always loaded down the secondary windings of audio transformers to the manufacturer's specifications.  For example, I would put a 50k resistor across the secondary winding of a 500/600 ohm to 50k input transformer.  But I have consistently noticed schematics of broadcast and other high quality audio equipment, that most circuits use a much higher load resistance, typically 200k or more across a 50k secondary winding.  Often, there is no loading resistor at all. This is  for class-A amplifiers, so grid current loading is not a factor.

I have quite a comprehensive collection of radio engineering books, plus engineering data in some of the transformer catalogues, but none carry much discussion of this topic.  The RCA Radiotron designer's handbook says loading a transformer down gives smoother and wider frequency response, but increases both the tube and transformer distortion, implying that a resistive load should be avoided if possible.

Has anyone seen any data on this subject that goes into any more detail?  Those who have designed and built high quality audio amplifiers, how much loading did you use on your interstage transformers, if any?

[AB2EZ]

Don:

An interesting question!

I don't have an authoritative answer... but my guess would be the following:

The transformer always has some equivalent capacitance (transformer + the circuit it is connected to) across its output; and the driving source (reflected to the secondary) has some equivalent series resistance.

E.g., if the source resistance on the primary side is 600 ohms, then the equivalent source resistance (reflected to the secondary side) will be 50,000 ohms.

At high frequencies (maybe beyond the frequencies you care about in the application), the capacitance will form a voltage divider with the equivalent series resistance... leading to a 6dB per octave roll off. If you put a resistive load on the transformer, which is 50k ohms (or less) you will push out the frequency at which the roll off (caused by the capacitance at the output of the transformer) begins by a factor of 2 (or more).

At the low frequency end, the magnetizing inductance of the transformer will begin to roll off the response at a frequency where the impedance of the magnetizing inductance drops below the equivalent series resistance of the source at the output of the transformer. You can move that roll off point to a lower frequency by a factor of two (or more) by putting a resistive load on the transformer whose resistance is equal to the equivalent source resistance (or less).

Stu

[k4kyv]

...if the source resistance on the primary side is 600 ohms, then the equivalent source resistance (reflected to the secondary side) will be 50,000 ohms.

If the source resistance is 600 ohms, for example through an attenuator network, wouldn't that reflect to the secondary side as a 50K load, so that the transformer is effectively loaded to 25K if an additional 50K resistor were placed across the secondary?

[AB2EZ]

Don

I was asking myself the same question when I first replied.

I think the attached JPEG file helps to clarify this issue. The load resistance of the source (reflected to the output side) is in series with the parallel combination of: the magnetizing inductance (referred to the output side), the total capacitance (referred to the output side), and the resistive output load.

If you look back from the secondary toward the primary... you will see the source's series resistance as an equivalent load; but if you look from the source (including the source's own series resistance) toward the secondary (output)... then you will see only the magnetizing inductance, the capacitance and the output load resistor as parallel loads.

Stu

[AB2EZ]

Don

The more I think about this, the more I agree with your observation.

If the source (reflected to the output side of the transformer) is represented by a current source in parallel with a 50,000 ohm resistor... instead of the alternative representation as a voltage source in series with a 50,000 ohm resistor... then it is clear that any resistive load just adds, in parallel, with the source resistance.

Stu

[W2XR]

I think that in some cases, the secondary of an audio transformer is loaded with a resistance to dampen ringing that may occur if the transformer is looking into a non-linear impedance, such as the push-pull grids of a class AB2 or class B modulator.

In the driver stage to my push-pull class B 833A modulator, I load the secondary of the driver transformer with a 15K ohm, 50 watt resistor. This eliminates the slight ringing I would otherwise see if the secondary of this transformer was unloaded.

It also presents a somewhat more constant load impedance for the driver tubes to work into, which probably improves the distortion of the driver stage as well.

The loading resistor does waste some audio power, but this is relatively insignificant, as the bulk of the audio power is consumed within the grids of the modulator tubes.

With regard to small-signal transformers of the line-to-grid or microphone input variety, some of these transformers are specifically designed so that the secondary is to remain unloaded. This is due to the fact that ribbon microphones are designed to work into transformers exhibiting an unloaded secondary for best performance, particularly with respect to maintaining their rated  frequency response characteristics.

This probably also maintains the optimum SNR, as the gain of the circuit is not reduced as would be the case if the secondary of the input transformer was loaded with a resistance.

Just my two cents worth of contribution concerning this topic.

73,

Bruce

[k4kyv]

My particular application is the audio input to the HF-300 homebrew
transmitter.  The setup is a follows:  The nominal 600-ohm output of a UREI
BL-40 Modulimiter (solid state compressor/limiter designed for AM broadcast
service), feeds through a variable attenuator to a line amplifier, which
drives the transmitter audio section.  The transmitter uses four 2A3's in
pushpull-parallel (cry your eyes out, audiophools!) to drive the class-B
modulator, presently a pair of 810's.  The broadcast-quality input
transformer is a UTC LS-18, 500-ohm primary to 50,000 ohm  split secondary.
I have a 6 dB fixed resistive H-pad inserted in the balanced line between
the line amp and the transmitter.  A General Radio audio level indicator is
permanently connected to the output of the H-pad to monitor the audio level
going into the transmitter. I  originally had a 25K carbon resistor shunted
across each section of the transformer secondary, to load it down to 50k.

Last night, I removed the 25K resistors and replaced them with a pair of
100k resistors, placing a 200k  load across the secondary instead of 50k.  I
immediately noticed about 6 dB more audio drive to the modulator tubes.  I
reduced the signal level to the line amplifier until I once again had
approximately 100% negative peak modulation.  I noticed the G-R level meter
reading exactly the same as it did before I changed resistors, so the
increase in output shows up as more signal going into the transmitter, as
the 500-ohm line feeding the transmitter is seeing a higher load impedance,
as reflected back through the input transformer.  I listened on a receiver
to a recording played to modulate the transmitter while working into a dummy
load, and noticed little, if any change in quality, although I haven't tried
running a frequency response test.

My take on the situation is as follows.  The nominal 500/600-ohm signal
source feeding  the transmitter could be thought of as an a.c. generator
with zero internal  resistance, wired in series with a 500 to 600 ohm
resistor.  That effective series  resistor loads down the transformer
secondary at 50k as reflected through from the primary, and the 50 kilohms
of resistance I had shunted across the transformer effectively loaded the
secondary of the transformer to 25k. instead of 50k.  Although this might
have smoothed out variations and slightly extended the frequency response of
the transformer, the additional load soaked up 6 dB's of signal level, and
may have increased distortion, since both the limiter output stage and the input transformer were having to handle substantially more audio signal current to maintain the required signal level.  The 200k's of resistance will still tend to damp out any ringing effect of a totally unloaded secondary, but the effective load on the transformer is now more nearly correct, supplied mostly by the 500/600 ohm internal resistance of the driving source. I could load the transformer to exact manufacturer's specs by adding a 50-ohm  resistor in series with each leg of the balanced audio line feeding the transmitter, at a slight expense of signal level.

[AB2EZ]

Don

I agree completely with your assessment.

I wonder:

When the manufacturer of the transformer specified the lower 3dB cutoff frequency (which depends upon the magnetizing inductance) and the upper 3dB cutoff frequency (which depends, among other things, on the equivalent parallel capacitance) did the manufacturer assume that both the input and the output had resistive loads (600 ohms and 50,000 ohms, respectively)... which would make the value of each of these specified cutoff frequencies twice as good?

Stu