The posts on stability a few back are interesting to me, as that's the issue I'm fooling with right now. My pet theory is that oscillations do not necessarily average to the desired point on a waveform, and you can have periodic distortions that don't show up in steady state measurements, simply because that's not when the amp is oscillating. I believe this is audible even to the group that measures and tries things, but can't hear anything (me!)
I'm working on a "sniffer" probe that can be used by clipping it on the amp outputs during listening. It would filter out the audio, and trigger on any small signals above audio frequency. There are a lot of problems like huge s/n ratio between full power audio peaks and millivolt level oscillation bursts, and pickup of ambient RF on the speaker leads, but I think it's surmountable. For the moment, I'm going to try looking at the amp output through a high pass filter, using the scope. (filter- 1k series input resistor, 2.31nF series cap, .957mH coil to ground, .957nF series cap, 2.31mH coil to ground, 1k loading resistor to ground- simple passive 4 pole butterworth LC filter) The only concern I have is if there is some oscillation within the audio band, there's no way to separate it from the signal. Fortunately, I don't think this is common. If anybody wants to spice that filter, I'd be curious if you think it's suitable for what I've described.
I'm working on a "sniffer" probe that can be used by clipping it on the amp outputs during listening. It would filter out the audio, and trigger on any small signals above audio frequency. There are a lot of problems like huge s/n ratio between full power audio peaks and millivolt level oscillation bursts, and pickup of ambient RF on the speaker leads, but I think it's surmountable. For the moment, I'm going to try looking at the amp output through a high pass filter, using the scope. (filter- 1k series input resistor, 2.31nF series cap, .957mH coil to ground, .957nF series cap, 2.31mH coil to ground, 1k loading resistor to ground- simple passive 4 pole butterworth LC filter) The only concern I have is if there is some oscillation within the audio band, there's no way to separate it from the signal. Fortunately, I don't think this is common. If anybody wants to spice that filter, I'd be curious if you think it's suitable for what I've described.
Here we go again
Gentlemen,
From what has been said about the subject matter, I offer a round up. Welcome fundamented bashing, encourage refrain from hand waving.
Take 2 uHy output coil and assume an 8 ohms resistive load, for the sake of focusing on a specific situation. Acknowledging real world situations will be different.
- 3 dB cutoff is about 640 KHz, well above audio band.
- Negligible attenuation at 20 KHz, phase angle -1.8 deg.
- A few meters of cable may well match or exceed 2 uHy.
- Inherent driver and crossover reactances usually orders of magnitude higher.
On the other hand:
- 1 KHz square wave passed through this network gets harmonics distorted in amplitude *within the audio band*. That is, if we consider the output composed of the original signal plus an error introduced by the network, this error is roughly in the order of -60 dB down.
- We worry about any type of distortion exceeding -80 dB (or even less).
- Real music depending on instrument or voice, comprises fundamental plus partials whose amplitude and (sometimes) phase ratios define timbre. We are very sensitive to timbre.
Any comments?
Rodolfo
Gentlemen,
From what has been said about the subject matter, I offer a round up. Welcome fundamented bashing, encourage refrain from hand waving.
Take 2 uHy output coil and assume an 8 ohms resistive load, for the sake of focusing on a specific situation. Acknowledging real world situations will be different.
- 3 dB cutoff is about 640 KHz, well above audio band.
- Negligible attenuation at 20 KHz, phase angle -1.8 deg.
- A few meters of cable may well match or exceed 2 uHy.
- Inherent driver and crossover reactances usually orders of magnitude higher.
On the other hand:
- 1 KHz square wave passed through this network gets harmonics distorted in amplitude *within the audio band*. That is, if we consider the output composed of the original signal plus an error introduced by the network, this error is roughly in the order of -60 dB down.
- We worry about any type of distortion exceeding -80 dB (or even less).
- Real music depending on instrument or voice, comprises fundamental plus partials whose amplitude and (sometimes) phase ratios define timbre. We are very sensitive to timbre.
Any comments?
Rodolfo
Re: Re: Here we go again
Jan,
Unless I messed up badly, the "negligible" linear distortion for harmonics within the audio band for this network measures up to a -60 dB error, that is what I wanted to remark.
It may be argued a sqare wave is not an audio signal, which is true. But then a simple thought experiment. Assume we insert an ideal brickwall filter at 20 KHz. Since this filter will not alter in any way the signal withing the audio band, the aforementioned distortion will still be present but now with a genuine audio signal.
Rodolfo
PS Recall a 1 KHz square wave has harmonics at 3, 5 ... KHz. Other test signals will have even harmonics also, the best of course being very narrow pulses featuring an (almost) uniform comb spectra.
janneman said:
Jan,
Unless I messed up badly, the "negligible" linear distortion for harmonics within the audio band for this network measures up to a -60 dB error, that is what I wanted to remark.
It may be argued a sqare wave is not an audio signal, which is true. But then a simple thought experiment. Assume we insert an ideal brickwall filter at 20 KHz. Since this filter will not alter in any way the signal withing the audio band, the aforementioned distortion will still be present but now with a genuine audio signal.
Rodolfo
PS Recall a 1 KHz square wave has harmonics at 3, 5 ... KHz. Other test signals will have even harmonics also, the best of course being very narrow pulses featuring an (almost) uniform comb spectra.
PMA said:Yes, here we go again
Could anybody explain me destroying sound effect of 1uH//1ohm coil and no problem with at least 2 - 4 uH of a speaker wire?? I am just one ear
Hi Pavel,
No body can. Coil haters just BELIEVE it affects the sound. Don't ask that kind questions, particularly unanswerable questions.
Cheers, Edmond.
PS: How about your little gem?
Just to put some numbers to this, I'm reposting a chart from an article reviewing speaker cables, where they measured the R, L and C of the cables. These numbers are per foot. The article was at audioholics.com
If the amps used were monoblocks and placed right behind the speakers using a very short length of the Goertz wire and no output coil, it seems like one could get some pretty low series inductance values between the amp output and the speaker. Would the improvement be audible? Dunno.
If the amps used were monoblocks and placed right behind the speakers using a very short length of the Goertz wire and no output coil, it seems like one could get some pretty low series inductance values between the amp output and the speaker. Would the improvement be audible? Dunno.
Re: Re: Re: Here we go again
Rodolfo,
I think I'm still missing something.
An audio signal has (ideally) nothing beyond 20+kHz, so the 640kHz roll-off doesn't do diddly to it.
OTOH, a 1kHz square wave, as you say is not an audio signal. It has harmonics in the audio band, which will NOT be linearly distorted. It has harmonics beyond 600kHz which may be linearly distorted, but they are not in the audio band.
Agreed?
Jan Didden
ingrast said:
Rodolfo,
I think I'm still missing something.
An audio signal has (ideally) nothing beyond 20+kHz, so the 640kHz roll-off doesn't do diddly to it.
OTOH, a 1kHz square wave, as you say is not an audio signal. It has harmonics in the audio band, which will NOT be linearly distorted. It has harmonics beyond 600kHz which may be linearly distorted, but they are not in the audio band.
Agreed?
Jan Didden
andy_c said:Just to put some numbers to this, I'm reposting a chart from an article reviewing speaker cables, where they measured the R, L and C of the cables. These numbers are per foot. The article was at audioholics.com
If the amps used were monoblocks and placed right behind the speakers using a very short length of the Goertz wire and no output coil, it seems like one could get some pretty low series inductance values between the amp output and the speaker. Would the improvement be audible? Dunno.
Hi Andy,
One thing that is nicely shown is the tradeoff of inductance vs capacitance. The lowest L cable has highest C and vice versa, as far as I can see.
I wonder what that very low L (with highest C) with very short length would do to amp stability? Depends on the speaker a lot, I guess.
Jan Didden
janneman said:
It doesn't look like it would be a problem. Three feet of the Goertz would be less than 1000pF.
Another thing that was mentioned earlier about electrostatics was that their input transformer would prevent the speaker's input impedance from looking purely capacitive at frequencies where it could affect amp stability.
Hi PMA,
PMA :
Could anybody explain me destroying sound effect of 1uH//1ohm coil and no problem with at least 2 - 4 uH of a speaker wire?? I am just one ear "
I think everybody is expecting that answer from some of the expert in the field here.There's much misunderstanding and miscommunication in this thread.
Yes, there might be audible difference in using output inductor, but that also depends on the amplifier characteristic and the test setup.
Lets assume the case of "ideal" inductor is used as output coil:
Graham Maynard mentioned something like "inability of NFB to fix error in time" thus causing first cycle of single sinewave to have significant error.That would mean in slowly reacting amplifier there is significant error going on.But people with faster reacting amplifier might not have experienced such big error.
other thing that me and some other feel might need to be clarified is, and as PMA pointed out, how about other inductance present in the system ?
I assume the "no-output inductor" point of view would be that:
1.It is better to reduce inductance at the amplifier output. So output inductor is "bad". THD doesn't tell everything.
or
2.Change your amp design to a faster reacting one.
3. or both.
the pro-inductor point of view is :
1.How to make the amp stable into any capacitive load while maintaning very low THD without inductor ? and to maintain low THD we need to maximize global negative feedback up to as high possible frequency, and without compromising stability.
2. Cable and loudspeaker inductance already present in the system. so, how could small value output inductor possibly have much audible difference ? and it doesn't show up in the THD .
now, let's not get upset at each other, guys
Cheers
Hartono
PMA :
Could anybody explain me destroying sound effect of 1uH//1ohm coil and no problem with at least 2 - 4 uH of a speaker wire?? I am just one ear "
I think everybody is expecting that answer from some of the expert in the field here.There's much misunderstanding and miscommunication in this thread.
Yes, there might be audible difference in using output inductor, but that also depends on the amplifier characteristic and the test setup.
Lets assume the case of "ideal" inductor is used as output coil:
Graham Maynard mentioned something like "inability of NFB to fix error in time" thus causing first cycle of single sinewave to have significant error.That would mean in slowly reacting amplifier there is significant error going on.But people with faster reacting amplifier might not have experienced such big error.
other thing that me and some other feel might need to be clarified is, and as PMA pointed out, how about other inductance present in the system ?
I assume the "no-output inductor" point of view would be that:
1.It is better to reduce inductance at the amplifier output. So output inductor is "bad". THD doesn't tell everything.
or
2.Change your amp design to a faster reacting one.
3. or both.
the pro-inductor point of view is :
1.How to make the amp stable into any capacitive load while maintaning very low THD without inductor ? and to maintain low THD we need to maximize global negative feedback up to as high possible frequency, and without compromising stability.
2. Cable and loudspeaker inductance already present in the system. so, how could small value output inductor possibly have much audible difference ? and it doesn't show up in the THD .
now, let's not get upset at each other, guys
Cheers
Hartono
Re: Re: Re: Re: Here we go again
Jan,
The point is we cursorily disregard linear frequency response distortions well within the passband for they are indeed small, but they are there.
In this example we get 1e-3 dB at 1KHz /90 e-3 deg., 2 e-3 dB at 10 KHz / 0.9 deg. and so on.
But if we look at the error itself and in dB, it turns out to be in the mentioned -60 dB range. I am not arguing whether this is of consequence or not given the fact one may also posit it is swamped by other much larger errors from cabling, drivers etc. But .... Are we ready to dismiss what others say about audibility without further consideration?
I am not.
Rodolfo
janneman said:
Jan,
The point is we cursorily disregard linear frequency response distortions well within the passband for they are indeed small, but they are there.
In this example we get 1e-3 dB at 1KHz /90 e-3 deg., 2 e-3 dB at 10 KHz / 0.9 deg. and so on.
But if we look at the error itself and in dB, it turns out to be in the mentioned -60 dB range. I am not arguing whether this is of consequence or not given the fact one may also posit it is swamped by other much larger errors from cabling, drivers etc. But .... Are we ready to dismiss what others say about audibility without further consideration?
I am not.
Rodolfo
We seem to have an education system today that produces people with the attitude - "If I cannot understand it, it cannot be" - very sad.
I thought science was propelled by open minded spirit of inquiry.
I work for a company doing scientific research every day - some well qualified, clever people - but still very often it is experiments that lead to our understanding and not theoretical analysis.
If you are interested in the reality of this situation and not just winning the argument - why not go fit a coil / resistor to your amp o/p and see what you can hear ?
If you cannot hear a difference - fine, but if you can, you may have to accept there some things you do not yet understand.
Is that so hard to accept ?
I thought science was propelled by open minded spirit of inquiry.
I work for a company doing scientific research every day - some well qualified, clever people - but still very often it is experiments that lead to our understanding and not theoretical analysis.
If you are interested in the reality of this situation and not just winning the argument - why not go fit a coil / resistor to your amp o/p and see what you can hear ?
If you cannot hear a difference - fine, but if you can, you may have to accept there some things you do not yet understand.
Is that so hard to accept ?
PMA said:
Well, okay, this kind of calculation is straightforward for something like zip cord. That same site shows the formulas here. I don't have the instrumentation for measuring this, so I have to trust someone else who does (maybe too much so) . BTW, I believe one of the people listed as a peer reviewer for that inductance calculation page is our own jneutron, so maybe he could chime in.
Of course, calculations of L and C for a transmission line can be pretty complex. These calculations are as much about structure as they are about the components (wires) that make them up. It's not difficult at all to envision structures with a large capacitance per unit length, and small inductance per unit length. It's already known that such cables have caused amps to oscillate, so it's more than just an academic thing.
- Status
- Not open for further replies.