I assume that no one uses two different set of speakers to test 2 power amplifiers. The case is that power amplifier with high output impedance, like tube power amps sometimes have, is difficult to set to similar level as high damping factor solid state power amplifier. In our experience, sound level meter is then more effective than electrical measurements at power amplifier terminals.
... on electrical side of crossover, I guess. You hardly get such pronounced dip with acoustical measurement.
Why? becuase of the really high level of distortion seen. It only seems to be the very low Zo of an amplifier which minimizes the distortion produced/generated. And, if the amplifier Zo is not very low, then that generated distortion gets fed back into the amp if it has gnfb loop. What does that produce.... better or worse sound? Is this a reason why solid state amps with high nfb at low freqs sound cleaner, tighter and more accurate, generally, in the bass?
If the distortion at the cable end/speaker load end, is such distorted, will it produce harmonics that get thru the crossover to other drivers?
Maybe another condition of "As IF" the cable produced sounds that resemble distortion compared to other cables with lower Ls/Z or shorter (towards zero) cable length?
Resistor tolerance: There is possibility and there is probability of a resistor being out by as much as 1% (if so spec'ed as 1% tolerance). The probability for modern mfr'ed film resistors is very, very low.
Thx-RNMarsh
Last edited:
For a simple deposited film resistor, the manufacturer will use test by sort methods to bin the resistors based on value and tolerance. As such, the values as purchased will have truncated distributions, and this could lead to values spanning the entire allowed value band.
If the manu does a spiral trim to final value, they most likely will measure the raw value, calculate the laser spiral path length, trim it, most likely either measure it immediately after trim (assuming the heat capacity and tempco allow), or do a final measure after a bit of time. The final distribution using this method can easily be a normal distribution around the final value, or, if the manu has a tighter tolerance product line, cull out the really good ones for higher pricing.
Once a manufacturer has the ability to sell all it's resistors to varying tolerance levels for varying costs, they will try to optimize their profit. So they have no incentive towards providing us bulk resistors with a normal distribution.
jn
Hi Richard,
In fact, also when the Zout is very low, still the 'error' voltage gets into the feedback loop. It has to, because the Zout is made so low because of the feedback action in the first place.
Jan
yes of course... large or small amount depending on Zo. best though is no cable -amp built into spkr cabinet or short length cable as possible. And/or more linear spkr load - like planar or ribbon perhaps. Also, reason why electronic crossover measurs and sounds better - no spurious harmonics get over into mid/hi freq spkr as esily as with passive crossover.... greater isolation. Ditto bi amping and bi-cable... isolation from nonlinear load generated harmonics.
That only leave the line level from source/preamp to power amp... there the Cp of the cable dominates as part of the loading affect and the input Z of the amp... The C can be reduced by appropriate materials and diameter etc and the amp load is pretty much linear so less interfacing trouble if using this cabling route/method. Just more care in grounding/shielding.
-RM
That only leave the line level from source/preamp to power amp... there the Cp of the cable dominates as part of the loading affect and the input Z of the amp... The C can be reduced by appropriate materials and diameter etc and the amp load is pretty much linear so less interfacing trouble if using this cabling route/method. Just more care in grounding/shielding.
-RM
Last edited:
As Patrick remarked, you don't get this kind of measured performance easily, but there is a fundamental reason why you can't at all.
That reason is that the soundfields created by two non coincidental drivers have two different vectors. Even if perfectly aligned in level and phase, this implies that there can never be complete annihilation.
Doing some very acid math, I get with a distance between two drivers of 15 cm and measured at 1 meter a remainder that cannot be interfered away well over 1 %.
Therefore, your claim of a .3% residue is extraordinary and cannot be believed without further proof.
It turns out that there IS a good way to do that in situations where it's significant. Rather than do this like they did in the '50s, in the analog domain, where instability is just around the corner, it can be done digitally. I just saw a demo of some software used for amplifier control in pro installations which implements wire corrections. Very cool.
Wrong guess. This is result from acoustical measurement. Two drivers, two amplifiers.
I'm not making claims, just demonstrating real results.
Best proof is for others to duplicate results using similar setups.
Technique uses measured IR from chosen point on intended listening axis. Inverse transfer functions are generated that linearize each driver response. These are convolved with chosen crossover filters and bandwidth limiting filters. These become filter kernels for playback convolution engine.
The distortion does not go away when the source impedance is low. Its still there in the current waveform. The lower source Z increases the amplitude in the current waveform. More interesting would be the measured acoustic distortion change when the source impedance changes. Since the actual force in the driver is a function of the current not the voltage it may be lower with a higher impedance drive.
How does it sound as you move off the intended listening axis? What sort of change do you get?
Yes, in the fashion world, having a measurement (or an experiment) be correct and replicable so that valid conclusions can be drawn is secondary to the story that's being sold to their non-technical audience. In the real engineering world, that doesn't fly. Maybe the concept is right, maybe it isn't, maybe it's right for certain circumstances and not others. That's what curious people want to know. They care. Their learning didn't stop 40 years ago.
Personally, I'd love to see details of how the measurement was made, how sensitive it is to mike position, and how it was achieved in the speaker. Things like that are actually consequential to the sound. And some of us who aren't "pros" in the fashion market actually care about sound.
- Status
- Not open for further replies.