„CMP framing“ – what the ** you mean ?

[mige0]

I suspect "delayed energy dissipation" would be too much a mouthfull during conversation unless we called it DED as apposed to CMP.

At the Thermal Distortion thread it became most obvious that "some" are in biiiiiig troubles when it comes to time related effects.

This came at no small surprise to me back then - but on the other hand this insight into such specific limitations of personal comprehension makes me patient with respect to spreading of CMP concept

Frequency domain thinking versus time domain thinking - its simply another way to attempt things - and - you have to get used to it like every beginner.

I suppose this gets at the root of my dislike of the phrase "energy storage".

I suspect "delayed energy dissipation" would be too much a mouthfull during conversation unless we called it DED as apposed to CMP.

as long as you do not accept CMP concept as such - no wonder quite *anything* I relate to CMP is "out of context" for you


Michael

DED is more understandable.

Possibly -
but possibly this is just depending on the level of understanding / context you like to be limited to ?


Michael

 

[DualTriode]

Not always a good to A**-u-me

There is no instability with any reasonable driver. Don't you think driver manufactures would have addressed this by now if it were so? Audio is not exactly new science.

A problem with SET amps is they typically have high output impedance. Like the case of a cap in series with the woofer impedance peak, a resistor in series also has an affect. It is not a resonance but rather a voltage divider which can alter the response across the complete audio band. With any amp with high output impedance it is best to try to add impedance comp to make the entire speaker, at its terminals, look like a resistive load of constant value.

What is a reasonable driver? They tend to ring at resonance all of them to some extent. The higher the Q the more they tend to ring. When I am speaking unstable I am not talking whoppy floppy like the front wheel on the shopping cart at the grocery store, just a tendency to ring.

With SET amplifiers higher Zo is not so much a problem as a characteristic that needs to be accounted for when selecting and applying speakers. In my case it required a lot of fun effort selecting low impedance tubes and impedance matching transformers to build a SET to match with Sennheiser 300 ohm headphones. The HD600 ‘s have an impedance peak of over 600 ohms. The SET and HD600’s seem to work best with a SET Zo of 6 ohms. Higher and the bass tends towards flabby (tends to ring around the resonance frequency).

A few pages of posts back I mentioned the power output difference of speakers driven by current source amplifiers and voltage source amplifiers. An odd thing is that the speaker simulation software assumes voltage source amplifiers, not always a good to a**-u-me.
DT
All just for fun!

 

[dlr]

New HOLM measurements

I decided that it would useful to do my own distortion measurements. After some testing with HOLM, I've got a set. These are for the Scan-Speak 15W/4531 in my dipole. The mic was set 3" (picked randomly) in front of the driver using the UE for crossover duties.

I have an MSB MVC-1 8-channel volume control fed by the output from the Delta 410 sound card. It uses switched resistors described by MSB:

MSB has created the “4096 level digital stepped attenuator”. The volume control is actually a ladder of resistors that are switched in and out using the digital information provided to the DAC.

Each press of the volume button provides precisely 0.5db change. In the UE is added a 10db Q=10 470Hz peak into the crossover for the 15W section. Only the 15W output was enabled. When the 10db peak was added I used the MSB MVC to precise reduce the output by an equivalent 10db so that the signal at the resonant peak would have the same relative level.

The measurements using HOLM look as if there's not quite a match, but at a 3 inch mic placement, the FR is not going to be the same as the far field, the measurements used to feed into the UE. The unaltered response at 3 inches is not flat, but the far field response is. Room reflection and background noise issues prevent me from doing the distortion measurements in the far field.

I did confirm and it's shown in the measurements that adjusting the level in HOLM after measuring a nominal level then -10db is precisely 10db different. The overlay of one measurement by the shifted one that is 10db different according to the MSB setting shows the rolloffs to be nearly perfect.

I found the answer to my earlier question about the HOLM display values. If the "Normalize to zero" check box is set, HOLM adjusts the display, but it's difficult to compare, so I did these unchecked. This provides absolute displays making it far easier to compare.

With that as background, here are a few of the measurements, all showing 2nd HD:

No peak vs. 10db peak, signal adjusted by -10db, no correction to display in HOLM:

Notice the noise floor at the upper end is a good overlay.

No peak vs. 10db peak, signal adjusted by -10db, adjustment in HOLM to add 10db to lower value measurement:

Notice the nearly perfect overlay in the general passband and the 10db difference in the noise floor at the upper end.

What's most telling is that it follows what John pointed out. If you precisely match the level at the peak of an added resonance so that the driver is excursing the same amount at that frequency, there is no increase in distortion. More interestingly, the rest of the bandwidth shows a drop in distortion. What it says is that there is no increase in distortion due to the resonance, any increase is due to the motor displacement change. The resonance itself adds zero.

More to come in another post.

Dave

p.s. I'm doing this a bit late, I have a cold, so if some correction is needed, it's probably due to posting this quickly.

 

[dlr]

Next measurements

Full screen showing settings use in HOLM for each graph (to remove all doubt as to settings):

Dave

 

[dlr]

An interesting comparison

I thought it might be interesting to show the different with different crossover settings. Using the UE, it was a simple task to do. Shown below is the 2nd HD for my normal 250-1200Hz LR8 vs. 100-5000Hz LR8. I leave it to you to analyze:

I will add one note here. I did not leave the start time marker at the HOLM default. That places it at the beginning of data from what I can tell. This, however, includes a long segment of pre-impulse data. It includes a lot of background noise, skewing the result. I set the start marker at what appears to me to be the best point to capture driver data. The stop time marker is done similarly. I can change that if someone desires to see it done differently. I'll hand on the measurements in HOLM for now.

Oh yes, one more thing. All of these measurements were with the UE set to EQ phase flat, though you can't see that. It's not important in this set of tests in any case.

Dave

 

[soongsc]

George

I am rather surprised that you don't know the problem with this statement. Sure EQ can clean up the CSD at any partiicular point in space, that is well known (how audible this is, is another question), but the problems off axis are not always the same as the ones on-axis, so what corrects the problem at one point can make a problem significantly worse at another point. So instead of "killing two birds with one stone", you just winged the first one and scared the other one away.

I did a quick and dirty EQ using SE measurement for UE EQ at 0 deg,


Results measured at 0 deg and 30 deg

An externally hosted image should be here but it was not working when we last tested it.

 

[john k...]

What is a reasonable driver? They tend to ring at resonance all of them to some extent. The higher the Q the more they tend to ring. When I am speaking unstable I am not talking whoppy floppy like the front wheel on the shopping cart at the grocery store, just a tendency to ring.

With SET amplifiers higher Zo is not so much a problem as a characteristic that needs to be accounted for when selecting and applying speakers. In my case it required a lot of fun effort selecting low impedance tubes and impedance matching transformers to build a SET to match with Sennheiser 300 ohm headphones. The HD600 ‘s have an impedance peak of over 600 ohms. The SET and HD600’s seem to work best with a SET Zo of 6 ohms. Higher and the bass tends towards flabby (tends to ring around the resonance frequency).

A few pages of posts back I mentioned the power output difference of speakers driven by current source amplifiers and voltage source amplifiers. An odd thing is that the speaker simulation software assumes voltage source amplifiers, not always a good to a**-u-me.
DT
All just for fun!

You are confusing the behavior of a linear system and the affect of an amplifiers output impedance on the system Q with instability. Ringing in and of itself is not a sign of instability. Nonlinearity is not a sign of instability.

If you want the software to correctly model a current source all you need to do is place a large series resistance between the amp and the load and scale up the output appropriately. If the resistance is much greater than the load impedance the result will be that of a current source, at lease in terms of relative magnitude.

 

[DualTriode]

There is no confusion. It is a matter of degree. Q is adjusted to sculpt the frequency response. More Q means less control.
To assume that the amplifier is a perfect voltage source is a leap of faith. Start adding a few ohms of impedance; capacitors, inductors, the resistance of cables or amplifier output impedance and things change. It is not all or none.
Engineering is an iterative process that is not complete until you run out of time or money.
DT
All just for fun!

 

[gedlee]

More Q means less control.
DT
All just for fun!

What do you mean by "less control"? Sounds synonimous to "less damping" in this context.

 

[mige0]

You are confusing the behavior of a linear system and the affect of an amplifiers output impedance on the system Q with instability. Ringing in and of itself is not a sign of instability. Nonlinearity is not a sign of instability.

There is no confusion. It is a matter of degree. Q is adjusted to sculpt the frequency response. More Q means less control.

Your are really way off.

FR at whatever Q by itself simply is what it is - no *more control* is any needed. There is *plenty of (electric) control* besides the highish impedance as even you should easily see in the plots shown, where those Q's are molded at will.
What John and others try to teach you patiently is that it simply does not matter if the Q is caused by the mechanic properties - or - if you create a "synthetic" Q by means of an electrical behaviour upstreams.

In sum, this is what can be labeled as "response shaping at will" and what all those fancy new software and DSP units perform so well

Best re-read what John has pointed out or go and get some books if you are more happy with "carved in stone" LOL - there is really nothing to add at all...




Completely another story is the story about CMP systems
Michael

 

[fntn]

Your are really way off.

FR at whatever Q by itself simply is what it is - no *more control* is any needed. There is *plenty of (electric) control* besides the highish impedance as even you should easily see in the plots shown, where those Q's are molded at will.
What John and others try to teach you patiently is that it simply does not matter if the Q is caused by the mechanic properties - or - if you create a "synthetic" Q by means of an electrical behaviour upstreams.

In sum, this is what can be labeled as "response shaping at will" and what all those fancy new software and DSP units perform so well

Best re-read what John has pointed out or go and get some books if you are more happy with "carved in stone" LOL - there is really nothing to add at all...




Completely another story is the story about CMP systems
Michael

I didn't bother to mention that the response I provided was EQ'd with a boost at approximately 3K of approximately 3db - similar Q to the boost that was applied at 150hz.

Now class, any idea why the high distortion appearing at 150hz doesn't appear at 3 khz?


The obvious answer, had any of you been paying attention, is that there is no energy storage of significance at 3khz. The Neo 3 has a very slight resonance at around 600hz. On the other hand, the resonance of the PL18 Vifa is approximately centered on 40hz. With the 150hz boost being essentially within the resonance band - it's clear why 2nd and 3rd harmonic distortion increased dramatically for the Vifa with nothing really observable on the Neo 3.

Now, Dave, why would you be using an LR8 cutoff at 200hz in this "experiment" when the data I provided had no cutoff at the lower end - flat response with all of the energy stored at resonance intact? And instead of applying your 10 db peak near the Scan's 35hz Fs, why did you decide to apply it at 470hz?



I have to say, if nothing else, this thread is definitely entertaining.

 

[SY]

Now class, any idea why the high distortion appearing at 150hz doesn't appear at 3 khz?

Excursion.

 

[fntn]

Excursion.

Wrong. The excursion just below the applied 150hz 4 db boost should be comparable or higher - especially at frequencies closer to resonance 40 hz. Clearly, there is no high level of distortion on either the boosted or unboosted response down low where one would expect it. This is further supported by the fact that the "boosted" response was set approximately 10db lower overall than the unboosted response. Thus, the only real boost for the "boosted" response doesn't amount to a boost at all. In fact, the actual level applied at 150hz is actually 6db down from the reference "untouched" sample.


Transducers, like most spring loaded masses, are second order devices. The stored energy interferes with the amplifier's ability to control motion. The obvious result is increased second order distortion. Third order increases as well, but not as much.

 

[dlr]

Now, Dave, why would you be using an LR8 cutoff at 200hz in this "experiment" when the data I provided had no cutoff at the lower end - flat response with all of the energy stored at resonance intact? And instead of applying your 10 db peak near the Scan's 35hz Fs, why did you decide to apply it at 470hz?

Because it easily and clearly demonstrates that adding EQ boost, whether passively or via DSP, does not add distortion. Any distortion increase is due to change in excursion, precisely as John described.

Dave

 

[SY]

The excursion just below the applied 150hz 4 db boost should be comparable or higher - especially at frequencies closer to resonance 40 hz. Clearly, there is no high level of distortion on either the boosted or unboosted response down low where one would expect it. This is further supported by the fact that the "boosted" response was set approximately 10db lower overall than the unboosted response. Thus, the only real boost for the "boosted" response doesn't amount to a boost at all. In fact, the actual level applied at 150hz is actually 6db down from the reference "untouched" sample.

It's hard to follow what you're saying here. Are you saying that a 4dB (or some other arbitrary number) boost at two different frequencies causes the same increase in excursion?

 

[gedlee]

Transducers, like most spring loaded masses, are second order devices. The stored energy interferes with the amplifier's ability to control motion. The obvious result is increased second order distortion. Third order increases as well, but not as much.

Wow, is that wrong. So you are saying that all second order systems are nonlinear? Or just when forced?

 

[fntn]

Wow, is that wrong. So you are saying that all second order systems are nonlinear? Or just when forced?

I'm saying they become non linear with respect to voltage input vs. acoustic output when the amount of energy stored in the moving mass's motion dwarf's the energy of applied force exerted by the amplifier. I think I've been saying the same thing now for many pages in this thread. If we disagree on this point, can't we just leave it at that and move on?

 

[fntn]

It's hard to follow what you're saying here. Are you saying that a 4dB (or some other arbitrary number) boost at two different frequencies causes the same increase in excursion?

The plot that contained the 4db boost at 150hz was in fact 10db down in applied drive level from the superimposed plot that did not contain a boost at 150hz. Thus, there really wasn't an absolute energy boost at 150hz at all. Yes, it was boosted 4db higher than the rest of the swept signal but in comparison to the other "flat" plot, it was actually 6db down because the "flat" plot was 10db higher in drive level. The increase in distortion over essentially the entire bandwidth for the "flat" or unboosted plot is evidence that the drive level was significantly higher - which is also born out by the numbers appearing at the left hand margin.

 

[SY]

Post 899.

 

[fntn]

Post 899.

Yes, what about post 899?

DLR didn't know that the plots were normalized to zero and the db numbers to the left showed the level of normalization. He did some reading and figured it out. So?