What makes a speaker "tube-friendly"?

[Wally]

I purchased one of the K-12G tube amp kits from S5 and am now looking for a set of speakers to build to go along with it. I've seen on a few pages that people describe a speaker as "tube-friendly" or not./ What characteristics make a speaker tubefriendly, besides high sensitivity?

[Brian Beck]

Relatively smooth and constant (unchanging) impedance values over the whole audio band. Higher impedances generally are best (avoiding dips to 2 ohms, etc.)

[SY]

I'll half-agree. Low impedance dips are generally bad, but... two of the most tube-friendly speakers are the LS3/5a and the ESL57, neither of which has a particularly flat impedance curve. OTOH, Magneplanars do, and they don't work particularly well with tubes.

[Brian Beck]

If the nominal absolute impedance is high enough, then variations are less of a problem. I too enjoyed my 11-ohm-nominal LS3/5As with small tube amps. If the impedance variations are relatively smooth (I'm thinking of many ESLs) then the response effect will be a smooth tilt, not too terrible a situation. But speakers, like many from B&W for example, with several high-order crossovers, and many highly reactive impedance regions, will be hard on high-output Z tube amps.

[rdf]

The S-5 site isn't particularly specific and the rated output impedance of '8 ohms' may or may not be a classic error. However the 10GV8 is a combination hi-mu triode/pentode so it's safe to assume the K-12G doesn't have high damping factor. Speakers with complex bass alignments, especially those that already tend towards bass emphasis as many commercial seakers do, should be auditioned with your amp if at all possible before purchase. Personally I would probably stick with acoustic suspension.

[tubee]

High impedance due to X-over filters, is not tube-friendly either. Gives a higher mid output on speaker with tubed amps, especially when this amp has a low feedback in the circuit. This can be compensated with a calculated extra filter consisting of a L, C and R in series over speaker connection. Frequency in calculation is the X-over frequency, wanted impedance taken from impedance speaker, eg 8 ohm or whatever.
High impedance of speaker does'nt "thread" amp though.

[Brian Beck]

High impedance due to X-over filters, is not tube-friendly either.

To quote SY, I half-agree. But higher impedance deviations from the nominal impedance will be less troublesome than lower impedance deviations. As a simple example, suppose you have an amp with 1 ohm output resistance driving an 8 ohm nominal speaker. You already have a nominal divider loss of 8/(1+8) = -1dB "across the board". If the impedance doubles at some frequency to 16 ohms (resistive, to keep it simple), the loss is reduced: 16/(1+16) = about -0.5. If the impedance drops to half, 4 ohms, then the divider loss is 4/(1+4) = about -2dB. So the impedance peak caused a RELATIVE response change of +0.5 dB, while the impedance dip caused a -1dB change in response.

[SY]

Excellent point. The one-sided nature of speaker impedance forces an asymmetry in response deviations. I hadn't really thought of that, but it's true. The upside is reduced distortion from pentode output stages, albeit with diminished headroom.

[Brian Beck]

I’ve always considered the impedance peak of the second-order resonance of a sealed box bass alignment as a gift, if it’s used appropriately. Often, the impedance peak covers a broad range and can be pretty high. Since this is where a lot of musical energy exists (not to mention movie soundtracks), the resonance reduces the need for current/power and makes life easier for all amps, but especially higher impedance tube amps. The impedance peak causes a response peak too of course, but in the prior example the impedance peak could cause no more than a 1dB response peak no matter how high the impedance peak, whether 20 ohms, 40 ohms, whatever. Sometimes there is a free lunch…

[tubelab.com]

This impedance rise in the lower ranges can be used to your advantage when designing the amp. I have recently been experimenting with a triode connected SE EL-34 amp. It uses no global feedback, but cathode feedback is applied in the output stage.

I have tried several different operating points for this amplifier. The point that gives great sound is 420 volts, 60 mA, and a 3 Kohm load. Measuring the output power with this load yields barely 5 watts. A 5 K load with this voltage and current, or a more conventional lower voltage, higher current operating point with the 3 K load makes more power, but distorts with heavy bass.

Why is this? It seems that my speakers have a gentle impedance bump at 70 Hz. The higher supply voltage allows the amp to work into this higher reflected load impedance without clipping.

The picture below was taken with the scope directly across the speaker while a CD with a bass guitar solo was playing. The scope is set on 5 volts per division. Here is about 35 volts peak to peak across my speakers without any clipping. I can't get above 17 volts with an 8 ohm load resistor.

This is a case of tweaking the amp (and its operating point) to match the speaker. It required the use of a different output transformer than I had origionally chosen.