Gentlemen,

Could anybody tell me the ramifications of running an amp (valve, of course)
which has only one set of tappings for the speakers (8 ohm, presumably) on a
pair of speakers that are nominally rated at only 4 ohms? Is there any
damage likely to result to the amplifier (trannies, I'm thinking)?

Keith G wrote:

Gentlemen,

Could anybody tell me the ramifications of running an amp (valve, of course)
which has only one set of tappings for the speakers (8 ohm, presumably) on a
pair of speakers that are nominally rated at only 4 ohms? Is there any
damage likely to result to the amplifier (trannies, I'm thinking)?

It depends how hard you run the amp, and what sort of amp
it is.

An amp designed for 8 ohms so that full power, or over 2/3 full power
can be only just maintained into 8 ohms, where plate and/or screen dissipation
maximums are nearly exceeded, may over heat when 4 ohms is used,
because as power rises above 2/3 max for 4 ohms, the tube
dissipation rises above the rated max.

This can be demonstrated most easily with a high power class AB amp,
using a sine wave test signal.
Fortunately, most designers arrange the load value to allow some reduction
of load, and some lower load tolerance, since this allows a good balance between

distortion and power output.
The duty cycle with music is less than that of the sine wave.
But a guitar amp might over heat if a dude plugs in a second 8 ohm speaker,
in parallel, and turns up the wick. Second speakers could be seriesed,
which is safer for the amp, but it won't make the sound any louder.

In most amps, SE, and PP, the thd increases about 3 times for the same
output voltage level when the load is reduced from the optimum 8 ohms
to 4 ohms, or if you like, about twice for the same power level.
This is the case with guitar amps, and PA amps.

But with most hi-fi amps, operating so even with 4 instead of 8 ohms,
the first 5 watts covers most of the sound levels, and
the thd may never exceed 0.1%. This may be the case with an amp
with a max of 30 watts.

I like to set up PP amps to give class A into 8, and AB into
lower loads, which usually means that the amp does OK with
4 ohms, but only really struggles with loads below 2.5 ohms.
Hence a pair of KT88/6550 give 35 watts class A to 8ohms,
and 53 watts AB to 4 ohms, and perhaps an absolute maximum
of 70 watts into 2.5 ohms where thd is getting awful,
and then clipping prevents the power from getting any more
with lower loads.

In such an amp, if the power needed is only 3 watts, any load from
2.5 to 25 ohms works fine.
This indeed covers most values of loads for many of today's speakers,
which have this range of impedance variation across the band.

Impedances are often low at the crossover points in the speakers because
there may be two LC filters, one for a bass speaker,
one for the midrange speaker, and these 2nd order filters
present a lower than nominal impedance at the xover F,
so a nominal 4 ohm speaker can dip to 2.0 ohms,
and often at an F which is at say 300 Hz, where a lot of music
energy is centered.
To ameliorate this rather obnxious trait in 2nd order filters,
I prefer to have over damped filters, so that the
initial roll off of the drivers is first order, then ultimately
becoming second order, and this reduces the current passing wastefully
through the LC filter at the crossover F, in cases where the speaker R
is that of the critically damped maximally flat response,
or a value over this.
If the speaker Z is too high for the values of L&C chosen
the reponse will have a peak at the xover F, and a lot of
current flows through the LC at this F, and if there were no speaker,
the LC would appear to the amp as a short circuit at the xover F,
becaue its a series resonant LC circuit.

Some study on basic LCR behaviour as writ in the text books
might through more light on all this.

Patrick Turner.