High impedance out amps VS as low as possible.

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I've been thinking about the output impedance of audio amps.

Typically, the goal is to make it as low as possible, but what would happen to the sound if I would double the voltage capability of the amp, but have an output impedance of 8 Ohm. Or, let’s go a little more extreme. Multiply the amp voltage by 10, have an output series resistance of 72 Ohm. With an 80 Ohm load, I can make a super low distortion amp is the 50 watt range.

As the series resistance increases, to the speaker, the signal should look more like a current regulated feed VS a voltage regulated feed.

I got this idea from the way video is driven. The driver is usually 2Vp-p, but with a 75 ohm resistor in series. With the 75 ohm load on the cable, the resulting video is 1Vp-p.

Brian.
 
Output impedance

The two major reasons to have a low output impedance is to improve damping of the woofer and to make sure that the electrical response at the speaker terminals is flat over the frequency band.
The impedance of a speaker is not uniform over the frequency band barring a few exceptions. So if you put a large impedance in series with it ( the output impedance ) the voltage from the amplifier will divide itself depending on the impedance at various frequencies. So a flat output at the amp will no longer be flat at the speaker terminals which will in turn lead to alteration of the frequency response of the speaker.
So identical amplifiers with different ( significant) output impedance should sound different with the same speakers. We are assuming that the speaker is like most systems with a varying impedance over the frequency range.

As far as low output impedance is concerned for damping purposes: The speaker's internal resistance itself is a limiting factor. So for the output impedance to have practically no significant effect it should be about less than 1/10 of the speakers resistance. At least that is what I have heard many people say. I myself think that if it is less than 0. 2 ohms it should be fine. I could make out a very slight difference with 0.5 and 0.1 ohms . It probably depends on other factors also. A 0.2 ohms output impedance with a Dynaudio Audience 50 speaker shows significant ups and downs in the frequency response at the speaker terminals.

You cannot compare this with RF. I am not a RF person but I thought that the equal impedance there is for maximum power transfer.
Cheers.
 
Re: Output impedance

ashok said:

You cannot compare this with RF. I am not a RF person but I thought that the equal impedance there is for maximum power transfer.
Cheers.

When working with RF, or any high enough frequency, matching impedances is a way to minimize reflection on the line. I say "high enough frequency", because it's the frequency of the speaker relative to how long the cable is, so it would be entirely possible to have this become an issue at audio frequencies.

When you're dealing with a long cable, and a high frequency, you run into the transmission line effect. Basically, if you send out a signal, you will have some reflected signal coming back if the impedances aren't perfectly matched. The magnitude of the reflected signal is directly related to the impedance of the line and the impedance at the end of the cable that it reaches. This is bad because you can have a lot of stuff on the line interfering with your signal, but it's also bad because it can require your supply to do wierd things, like drive your full output into a short, even though you have the cable and a load connected.
 
Optimum Z out ?

Hi Brian,
I had the same problem when I built my hybrid valve amp with no feedback. The output impedance was very high at around 1 ohm and it sounded far inferior to the Audiolab 8000 amp with an output impedance of much less than 0.1 ohm.
Not only was the bass not tight enough but the balance in the upper frequencies was wrong. I added feedback and dropped this to about 0.2 ohms and things improved . With an additional pair of output transistors and some more feedback this came down to 0.1 ohms or so and the system now sounded very nice. But I must add in spite of the problems, in the no feedback mode
the midrange was really nice . With feedback it was still nice but something had changed and it was not as nice as the no feedback mode.

So low feedback with plenty of output transistors ( low output impedance) should sound good !
I would suggest an output impedance that is less than 1/20th of the lowest speaker impedance that you plan to use. For a 4 ohm driver that would be less than 0.2 ohms.
I have also found that the equivalent amount of series or shunt feedback sounds different. So go ahead and test your design. It is the only way you can find out how it will sound.

Cheers.

In commercial designs you will find output impedances that are far less than 0.1 ohms. They could be as low as 0.01 ohms. In valve based systems this could be much higher. I am not sure but 0.5 ohms may be quite common. Some valve amps beat the pants off solid state designs !!
 
Transmission lines

Yes I forgot about the transmission line and reflections. I thought that this was only at very high frequencies and and not at audio frequencies. In fact this is a big issue in the 'cable' arena. Some say it is silly to talk about this in the audio frequency range and others who make expensive cable say it is not. Making exotic cable produces cables with different inductance and capacitance which react differently with different systems. In addition I have heard of a high-end amp break down due to rf oscillation when it was used with an 'exotic cable'. If you hear differences it could be just 'different and possibly likeable' but not necessarily accurate like an ideal wire. This difference would most likely be due to the same effects as 'output impedance'. Maybe someone should make identical systems in which one has the amp mounted on the speaker with not more than 1 foot of normal heavy guage wire. This test should be revealing.
Can someone work out the reflection in a speaker cable
(3 meters long ) with a 1 or 5 Khz square wave and confirm if it will affect the audio signal?
Cheers.
 
Re: Output impedance

ashok said:
So for the output impedance to have practically no significant effect it should be about less than 1/10 of the speakers resistance. At least that is what I have heard many people say.

For some tube (solid state also) amps this can be true, rather high output impedance. This can colouring the sound with certian speakers, bass reflex in particular (Am I right here?).

(My answer got a little time delayed...)
 
Re: Transmission lines

ashok said:
Can someone work out the reflection in a speaker cable
(3 meters long ) with a 1 or 5 Khz square wave and confirm if it will affect the audio signal?
Cheers.

This tranmission line talk is correct in theory but in real life? Minor importance I would say.

The speed of the signal in the speaker cable is 0.6-0.7 times the speed of light. How long time does it take for a reflexion to get back to the amp in a 10 meter (33 ft) cable? 55 ns! Which frequency does this correspond to? 18 MHz.

Can this tranmission line talk make any sence for audio signals?
 
I agree with most of what's been said. I'd say <0.5-ohms is quite adequate for most speakers.
Line reflections are not important for at least two reasons. Firstly the wavelengths at audio frequencies are massive. Secondly the output impedances and input impedances of equipment and speakers are radically mismatched anyhow. Cables sound different because of distortion effects and the effects of reactive loading on the circuits driving them. Some power amps, like Naim's, get very upset with high C low L speaker cables.
 
Re: Re: Transmission lines

peranders said:


This tranmission line talk is correct in theory but in real life? Minor importance I would say.

The speed of the signal in the speaker cable is 0.6-0.7 times the speed of light. How long time does it take for a reflexion to get back to the amp in a 10 meter (33 ft) cable? 55 ns! Which frequency does this correspond to? 18 MHz.

Can this tranmission line talk make any sence for audio signals?

The speed has nothing to do with it, it's the wavelength. I haven't done any math with this stuff in a while (I'm used to 50-ohm outputs, 50-ohm cables, and 50-ohm loads), but I think you follow the 1/10th rule. So for RF frequencies, you're talking like a meter. For audio frequencies, you're talking more like 1500 meters.

So if your audio cable is more than a mile, it becomes an issue.

(somebody check my math, like I said, been a while)
 
Re: Re: Re: Transmission lines

Narcisse91 said:


So if your audio cable is more than a mile, it becomes an issue.

(somebody check my math, like I said, been a while)


To do that, your speaker cable would need to have such an incredible low impedance, it would probably need to be something like 1 gauge, or, even 0.1 gauge.

As for reflections, I think we all havn't been taking into acount the inductance of the speaker wire. Granted, it's low, but, with a square wave at 20Khz & an 8 foot cable & the small capacitive load in the speaker's crossover, I believe it would be possible to see reflection bumps in somewhere between 500Khz & 5 Mhz. I wonder what these spikes would look like after the crossover, at the tweater's inputs.
 
test frequency

We can pretty much see that most people agree that we may not see any effects of reflections in the audio range. I do not think that any music source will produce 20 KHz square waves and its associated harmonics. With a ( system) roll off at 20 or 50 KHz we would not be able to get a square wave at these frequencies. Hence bumps in the response should not be present in a real world scenario.

It is good practice to keep speaker cables short. For those who can manage it it might be better to keep monoblocks next to the speakers and drive the monoblocks with good quality shielded cable and a preamp or a buffer amp. In engineering everything is a compromise - you pick the one that suits you best!
 
Re: Re: Re: Transmission lines

Narcisse91 said:


The speed has nothing to do with it, it's the wavelength...
(somebody check my math, like I said, been a while)

Speed and wavelenght is more or less the same thing. It's tightly connected. In "slow" cables (media) the wavelenght becomes shorter. Compare real waves on deep water and shallow water.
 
Re: Re: Re: Re: Transmission lines

peranders said:


Speed and wavelenght is more or less the same thing. It's tightly connected. In "slow" cables (media) the wavelenght becomes shorter. Compare real waves on deep water and shallow water.

That's correct, but has no bearing on this discussion, in terms of impedance matching and now transmission line effects on audio signals. Yes, at a given frequency, the wavelength and speed can both change, and yes, they are directly related but cables have rated coefficients for their speed (some number times the speed of light). This rated is done per some unit (meters, usually but may change from country to country), and is constant (ideally) for the length of the cable. Therefore, the speed of the signal doesn't change for the entire length of the cable (again, ideally). However, the length of the cable in relationship to the wavelength can change, and that's when you worry about reflection, if that wavelength is too short relative to the length of the cable. If the cable is "slow", no big deal, just takes long to get there.
 
Re: Re: Re: Re: Transmission lines

peranders said:
Speed and wavelenght is more or less the same thing. It's tightly connected. In "slow" cables (media) the wavelenght becomes shorter. Compare real waves on deep water and shallow water.

Speed, wavelength and frequency are all interrelated by the following:

Wavelength = speed/frequency

Frequency = speed/wavelength

se
 
The one and only
Joined 2001
Paid Member
I can hardly avoid referring you to my article on speaker
cables, found on www.passlabs.com which focuses on
the reflection effects.

High capacitance (slow) cables don't much affect the audio band,
as the typical first resonance mode is up around 1 MHz or
higher, but when an amplifier's open loop gain approaches
this frequency, the resonance can de-stabilize the loop,
resulting in oscillation, noise, and worse.

All easily fixed by an RC network on the speaker end, which
terminates the cable in something resembling it's characteristic
impedance.
 
Nelson Pass said:
I can hardly avoid referring you to my article on speaker
cables, found on www.passlabs.com which focuses on
the reflection effects.

High capacitance (slow) cables don't much affect the audio band,
as the typical first resonance mode is up around 1 MHz or
higher, but when an amplifier's open loop gain approaches
this frequency, the resonance can de-stabilize the loop,
resulting in oscillation, noise, and worse.

All easily fixed by an RC network on the speaker end, which
terminates the cable in something resembling it's characteristic
impedance.

Except that cables don't have any characteristic impednace at audio frequencies and no audio frequency signal has an electrical wavelength short enough to make reflection issues any sort of problem. So the amplifier's stability problems are obviously coming from some other source, such as RF pickup on the inductive loop antenna formed by the cable and the loudspeaker.

Also, reflection and resonance are two completely different animals. While the damping network changes the high frequency load impedance and subsequently changes the nature of the reflections, what you're ultimately doing is damping the electrical resonance of the loop and thereby making it less sensitive to RF at its resonant frequency.

se
 
The one and only
Joined 2001
Paid Member
On the contrary. Resonance and reflection are the
same in this context. If the cable is terminated in other
than it's characteristic impedance, there is a reflection
of the wave, and at the frequency where it results in
a standing wave, there is a resonance. This is why
video cable is terminated in a characteristic cable
impedance, such as 75 ohms.

RF pickup is unrelated to this phenomenon, and the
inductance of the cable is only incidental to forming the
characteristic impedance of the cable.
 
Member
Joined 2002
Paid Member
This thread essentially started with the idea that high voltage amplifiers were easier to make than high current amplifiers:

Or, let’s go a little more extreme. Multiply the amp voltage by 10, have an output series resistance of 72 Ohm. With an 80 Ohm load, I can make a super low distortion amp is the 50 watt range.

So if you really want to try this, build 80 Ohm speakers instead of blowing 9/10 of the energy in a resistor.

I'm imagining some kind of ridiculously large tower, like those old Infinity IRSs, with the speakers all in series, rather than series/parallel...

-- mirlo
 
Intersting topic so far. But let's compare something which is more interesting to me, and which I think gets overlooked to often: the <i>open loop</i> output impedance of the amplifier. Here, I think we will see very large differences between different designs.

Take, for instance, an Aleph or Zen type design. With it's common source output device, the open loop output impedance will be fairly large, and reduced to a more typical value only by the application of feedback. Tube amps are similar. The high impedance outputs are reduced through the transformer, and the application of feedback. Contrast that with a conventional solid-state design which uses emitter follower outputs, and consequently has a much lower open loop impedance.

Nelson touches on an interesting point with the cable resonance. At 1MHz+, it's well out of the audio band, but still affects the amplifier's stability and behaviour at these higher frequencies. Ultimately, this <i>may</i> translate to some audible symptoms.

In theory, with a low open-loop output impedance, phase margin will be less affected by reactances in the load. This, in turn should result in an amplifier which is less sensetive to the load impedance (phase margin will not vary as much with non-resistive loads). So why, then, do tube amplifiers and Alephs consistently sound so good? Is it that the phase margin doesn't matter to sound quality? Does the simplicity of these designs make up for their high load reactance sensetivity? If so, by what mechanism? Or, could it be that a higher open loop output impedance provides some measure of damping for spurious oscillations in the load? There seems to be a missing piece of the puzzle here...
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.