Speaker impedance choice

modulator

I don’t care much for that place with the soft walls, so I am good at escaping by now… Sitting in my hidey-hole, playing my guitars, I decided last year to buy speakers, to
• upgrade my 15 Watt bedroom Blaster from eight inch to ten inch
• have one 10 inch available for a planned 5F1 build
• Have one 12 inch for a 15 Watt tube head cabinet.
So I bought some speakers of relevant impedances in 10” and 12” sizes. Recently, I became aware that speaker size impacts sound quality in my little hidey-hole. I manage to get acceptable sounds from the low-quality bedroom amp, but the nice, fat 12” speaker’ed amp does not fall in line so easily. I have thus decided to alter my original plans, and think that building a smaller-speaker cabinet might work better than a larger-speaker cabinet, in my current situation. Seems like a high-quality eight inch speaker would be ideal.

The smaller valve amp has several output choices, allowing the owner to use 4 Ohm , 8 Ohm or 16 Ohm speakers. The big monster has a choice of 8 Ohm or 16 Ohm.

My question in this thread is:
Neglecting resistive, capacitive, mechanical losses, and speaker sensitivity, simply assuming a 16 Watt amp, using a 16 Ohm speaker will need 1 Volt and 1 Amp to drive at full power. Using an 8 Ohm speaker requires either 1 Volt and 2 Amps, or 2 Volts and 1 Amp. (Which is correct?)

Please, amp techs and engineers, which would be the best route for the amp in the long run? And why? I know, it was built to handle all those impedance choices, but, surely, one of them is the “preferred” way?

klaasvakie

Hey Modulator

Dislaimer: I'm an electronic engineer, but know very little about guitar amps.

So I quickly had a look at some amp schematics and it looks like they all have an output transformer to match the speaker impedance (see this nice hand drawn schematic of a 1959 marshall: https://drtube.com/schematics/marshall/1959t-66.gif)

What this means in practice is that there is no "preferred" way, the tubes always see the same load, i.e. the primary side of the output transformer. The tap on the secondary side of the output transformer then generates the correct level to drive the chosen speaker.

If you have a long cable from the amp to the speaker, choosing a higher impedance speaker will mean you have less loss in the cable. Power loss in cable is I² *R, so halving the current means losing 4 times less in the cable.

So onto this question:

modulator Neglecting resistive, capacitive, mechanical losses, and speaker sensitivity, simply assuming a 16 Watt amp, using a 16 Ohm speaker will need 1 Volt and 1 Amp to drive at full power. Using an 8 Ohm speaker requires either 1 Volt and 2 Amps, or 2 Volts and 1 Amp. (Which is correct?)

We start by calculating the current neede, starting with the formula for electrical power:
P=V*I , then replace V with I*R, from Ohms law
P=I*R*I
P=I² *R
rework:
I=sqrt(P/R)

So for a 16Ohm speaker we have
I=sqrt(16Watt/16Ohm) = 1Amp
This gets us the current. To get the voltage we can just use Ohms law (V=I*R)
V=I*R
V=1Amp*16Ohm
V=16Volt

For a check we can see if V*I gives us P which it does (16Volt*1Amp=16Watt)

For a 8Ohm speaker we have:
I=sqrt(16Watt/8Ohm) = 1.414Amp
V=1.414Amp*8Ohm
V=11.314Volt

Check: 11.314Volt*1.414Amp=16Watt

Hope this helps,

modulator

Klaasvakie, thanks. I should have worked it that way. Going from here, I would guess less amps will imply less heat dissipation (and other losses), and using a long speaker cable, I should go up in cross section. Which is limited by the puny connection contact size at the driver, and the lack of space in the average jack. Oh, and the small contact area between jack and socket. Sort of, small cable out of jack, onto huge Litz cable, onto small spade connectors again. Oh well, low Amps and moderate Volts, so should not be much of a factor?

But, your opinion of the small connection surface typically available on the speaker spades?

klaasvakie

Yeah, I think you are approaching this correctly. A couple comments:

Choosing the highest impedance speaker your amp can support is probably always the right choice *.
In my opinion the single most important choice to make is wire guage. Wikipedia says "most people" can't hear the difference when less than 5% of the power is lost in the cable/connections. This means that with a 16Ohm speaker you have 0.8Ohm to play with, with a 8Ohm you only have 0.4. If you believe your ears are especially golden you might want to stick to <2% or <1% cable loss. This datasheet (http://www.mantech.co.za/Datasheets/Products/RIPCORD_ABED.pdf) has resistance data for some run-of-the-mill twinflex. You are looking at 0.1Ohm/m for the really thin stuff, to 0.01Ohm/m for 1.5mm² (16AWG). Remember you have to calculate the resistance to the speaker AND back.
Speaker jacks can be fairly sucky, work on 0.1Ohm per connection, halve it if you use your contact cleaner religiously. Speaker spades probably have enough area (compared to the wire area), but the quality of the connection depends on so many things, its hard to say what it will actually end up as. I would use maybe 0.02Ohm-0.03Ohm for a good joint. These resistance values are completely thumb sucked, google will surely turn up someone that has measured this.
For guitar work, the rest of the audiophily stuff, namely cable capacitance, cable inductance and skin effect don't really matter. The practical manifestation of all these effects is that higher frequencies get suppressed --- but since this is a guitar amp you can trivially work around this by nudging the treble knob on the amp slightly higher.

So in short:

Go for higher impedance speakers.
Choose a thick enough cable so that the losses are < 5%.
Don't worry about audiophily stuff like low ESL, or low capacitance or Litz, you can fix all these with your treble knob.
Gold plating connector ends is usually worth the money, it probably has the same effect of going up a wire guage and will likely be cheaper.

* There is one scenario I can think of where a lower impedance speaker might have an advantage and that is where there is a long run in an area with very high electrical noise. i.e. sharing a cable tray with a large motor or something, It will be more difficult to couple noise into the lower impedance system.

modulator

Thanks, good information. Always looked at wire gauge in terms of amperage rating, knowing thicker is less resistance, but never bothering with the actual resistive loss along the way.

Aside, being in the habit to over-engineer things to avoid problems that will never occur, I look at my Hi-Fi, with the nice Litz cables, connected to crummy plated babana jacks to plug into the amps, and I always wondered about the puny wires in the speaker cabinets. Well, short runs, little effect. But the connectors bother me. Does not look right. Works OK. Looks wrong. Sort of, "really, ZAR Kxxx for the cables and amps and speakers, but this is the only way to connect?"

klaasvakie

So on a Hi-Fi, the fancy shmancy Litz cables might actually be worthwhile (depends on your ears and actual setup), because on a Hi-Fi the goal is to reproduce the original signal as accurately as possible --- if your cabling causes higher frequencies to be suppressed then the Hi-Fi is failing at its job. That said, in most cases it should be possible to compensate for poor cabling (in terms capacitance, inductance and skin-effect, not resistance) using EQ at the amp.

With guitar equipment however, everything is designed to colour your signal (Just look at the garbage frequency response of the well-regarded Celestion Creamback: https://celestion.com/product/151/celestion_cream/), and in addition you usually have many adjustments available to manipulate the colour --- everything from the tone knob on your guitar to your pedal chain to eq adjustments on the amp itself. A celestion creamback is already cutting a massive 40dB off everything above 7500HZ, so having a bad capacitive cable cut another 0.5dB above 15kHz isn't going to make a difference, especially if you can compensate by adjusting one of the many tone knobs available in your signal chain.

V8

klaasvakie A celestion creamback is already cutting a massive 40dB off everything above 7500HZ, so having a bad capacitive cable cut another 0.5dB above 15kHz isn't going to make a difference, especially if you can compensate by adjusting one of the many tone knobs available in your signal chain.

Good stuff, really useful info you've shared here - thanks Klaasvakie!

modulator

You touch on points I hope to discuss in another thread, related to this one.