I suppose where we disagree is that it doesn’t change anything.
I think most of this comes up in defensive of ultra high priced audio electronics aimed at an over 40 male population.
I don’t see how settling these questions is aimed at benefitting anyone else or how your proposed study will overcome this glaring bias.
Last edited:
So, you have measured or seen any measurements to verify that when an amp is running in third quadrant, that it still faithfully drives a second load device at the 2 to 5 microsecond accuracy, despite the fact that the opposite pass devices must reach across roughly a full rail voltage to pull the load? Remember, a reactive load requires each pass element to pull a load that is way on the other side of zero.
Remember, harmonic distortion does not spot temporal shifts of any specific frequency.
Jn
You have to ask what kind of malfunction would cause an amp to change phase without also being accompanied by severe distortion. Pretty much the only one is thermal drift.
Perhaps the distortion is higher in 3rd quadrant. So how much phase shift difference do you expect to have when distortion rises from 0.003% to 0.005%? Keep in mind that phase is changing relative to input signal phase, so 1 degree phase shift would be 1.75% error voltage between the halves of the LTP. You have to ask why the amp's error would somehow jump from 0.003% to 1.75% without any measurable overload distortion. It doesn't really work like that. That amount of phase shift cannot occur without at least causing the VAS to overload due to excess input from the LTP.
While it is true that phase shift is not a harmonic distortion, the LTP does not treat phase error one way and harmonic error another way. Both are distortions which the LTP applies a correction without bothering to consider the difference in their mathematical derivation.
The main difference in operation in 3rd quadrant is that the active device's Vce is higher, but the transistor does not experience any pin voltage reversals. It is just operating at higher Vce. The transistor can't tell whether it's base or emitter is above or below 0V, it only knows the differences between it's pin voltages.
Perhaps the distortion is higher in 3rd quadrant. So how much phase shift difference do you expect to have when distortion rises from 0.003% to 0.005%? Keep in mind that phase is changing relative to input signal phase, so 1 degree phase shift would be 1.75% error voltage between the halves of the LTP. You have to ask why the amp's error would somehow jump from 0.003% to 1.75% without any measurable overload distortion. It doesn't really work like that. That amount of phase shift cannot occur without at least causing the VAS to overload due to excess input from the LTP.
While it is true that phase shift is not a harmonic distortion, the LTP does not treat phase error one way and harmonic error another way. Both are distortions which the LTP applies a correction without bothering to consider the difference in their mathematical derivation.
The main difference in operation in 3rd quadrant is that the active device's Vce is higher, but the transistor does not experience any pin voltage reversals. It is just operating at higher Vce. The transistor can't tell whether it's base or emitter is above or below 0V, it only knows the differences between it's pin voltages.
Last edited:
What is the 'it' to which you refer?
What is the 'this' that comes up? Who is 'defensive' of said electronics?
Do you have an enumerated list 'these' questions?
Please clarify which particular 'glaring bias' are you referring to? Do you mean a bias against the likely effectiveness of wooden cable lifters, or something else?
Please bear in mind I and perhaps others have not followed your conversations in other threads, so please do not presume you are being clear in the way you have been describing what you believe to be pertinent issues.
Well stated description, nicely put.
The distinction would make more sense if I had actually said what I meant, sorry about that.
I meant second quadrant, where the voltage is near one rail, but the other rail devices are the ones conducting.
Sorry no diagrams here... Consider a reactive woofer, the voltage-current trace is elliptical. Now, add a second frequency and a second reactive load. When the main load is q2, and the second requires q1 vc, how do the active passes do that? Where are the circuit currents at all times.
I believe simple sines may not be adequate. Take a real three way load, use real music, and run a comparison of in/out. Is the output an exact scaled representation through all VI space?
Jn
I believe most of part our body can be train, like an athlete.
Every time an athlete bring a new world record, and what the threshold here (that a man can achieve)?
Yes. and How many can detect reliably the >20KHz (40-50KHz?) of a building alarm system? One in _____ people? Clearly there are people on the far right side of the bell curve. Learning/training can move you further to the right also.
So, who do you design for? The 50%, 75% on that bell curve? Or, as best Hi-End designers do... for the 99.9%.
THx-RNMarsh
So, who do you design for? The 50%, 75% on that bell curve? Or, as best Hi-End designers do... for the 99.9%.
THx-RNMarsh
Last edited:
WOW!
-RNM
I believe that the RTX unit that a few people have here will, with the Virtins pro integrations allow this to be done. Might allow a few issues to be put to bed, even though you poo poo'd it a week ago.
I wasn't asking for research to be done, just for someone to come up with a credible hypothesis as to why falling loop gain with frequency is bad once you are below a certain distortion threshold. The Griesinger paper (thanks Scott) suggests that 0.1% distortion is the threshold, but doesn't go into details as to whether that is THD or spot frequency. But it's a stake in the ground. And even the humble LM3886 is 10x better than that number. The 14k border patrol amplifer might struggle though
As usually it depends and when using music the real requirements might be more relaxing, but form masking theory (and graphs) you already get an impression that a in case of THD much lower levels of higher harmonics are detectable. (Which is used to detect defects in speakers)
If using single tones at higher levels but at 1kHz or below might lead to audible higher harmonics (5th - 7th and above, even when lower than -80 dBr).
Last edited:
Jakob: You might want to look at the ppt Scott linked. I was specifically commenting on the figure that Dave used as the point at which IMD from ultrasonics became audible to his (small) group of listeners.
I don't know if his music samples are still available but there is one that claims to be barely audible with 15% THD.
I don't know if his music samples are still available but there is one that claims to be barely audible with 15% THD.
In that case I meant q2, I just assumed you got it right because I'm not up on my quadrants... Sorry MIT guys.
An amplifier has no trouble driving all quadrants as long as voltage and current margins are adequate. In the attached simulations green is output, blue is current, pink is the differential error voltage. Yellow is the gate drive voltage of both output transistors, orange and red are their source currents. These transistors have no idea they are in Q2. Compare it with the second image which is Q1. If the simulator is wrong about this then I can only suggest to stock up on canned food and stay away from everything electronic, which is already or will soon be on fire.
The question still remains, how you would get significant phase shift change without harmonic distortion, as such phase shift would immediately overload the VAS. A Cordell Distortion Magnifier would be just the thing to do this experiment, although you could do it with a good ADC and your computer's onboard sound. I've done it before while playing music and didn't notice anything.
Attachments
Last edited:
Ah, simulation..well, why didn't you say so.
Edit: btw, nice and thank you for that.
My concern is with the physical wiring, which simulation is not setup to model. When an amplifier is pushing a hard load and runs through all four quads, the hf lower power current is being driven with alternate rails. So rail wiring currents are not simple, and the rail current magfield is not a good representation of output current. As you clearly show, the hf content is alternately being driven by two physical rails. If those rails were delivering current using either a triaxial or three layer stripline, rail magfield could be ignored. If the output stages are physically located apart, is the supply return current for each output stage field neutralized by return path proximity?
Perhaps the problem statement would be clearer if one attacks it from human hearing capability..
Consider a stereo setup, you are listening to the soundstage enjoying imaging. A triangle is 25 degrees off axis to the right of the lead vocal. What interchannel relationship is required to maintain the triangle image in space. IOW, what level of IID or ITD can be allowed and not smear the triangle image, that is, sideways image localization? If you go through the numbers, the IID is ridiculous, like .01dB or such. ITD on the other hand, can be in the 2 to 5 usec range and be discerned.
Does anybody think we can measure that level of delay of a small hf component buried within a complex two channel musical stream?
Is a typical spectrum analyzer capable of seeing such a temporal modulation?
Jn
Edit: btw, nice and thank you for that.
My concern is with the physical wiring, which simulation is not setup to model. When an amplifier is pushing a hard load and runs through all four quads, the hf lower power current is being driven with alternate rails. So rail wiring currents are not simple, and the rail current magfield is not a good representation of output current. As you clearly show, the hf content is alternately being driven by two physical rails. If those rails were delivering current using either a triaxial or three layer stripline, rail magfield could be ignored. If the output stages are physically located apart, is the supply return current for each output stage field neutralized by return path proximity?
Perhaps the problem statement would be clearer if one attacks it from human hearing capability..
Consider a stereo setup, you are listening to the soundstage enjoying imaging. A triangle is 25 degrees off axis to the right of the lead vocal. What interchannel relationship is required to maintain the triangle image in space. IOW, what level of IID or ITD can be allowed and not smear the triangle image, that is, sideways image localization? If you go through the numbers, the IID is ridiculous, like .01dB or such. ITD on the other hand, can be in the 2 to 5 usec range and be discerned.
Does anybody think we can measure that level of delay of a small hf component buried within a complex two channel musical stream?
Is a typical spectrum analyzer capable of seeing such a temporal modulation?
Jn
Last edited:
0.01db is 0.12%. That is getting much closer to plausible but still not there without overload (or an amp with >0.12% THD). Plausible would be where the IID is around the same level as THD.
0.12% is easily measurable with an affordable ADC, so the experiment seems on track. The trick is to get the attenuation matched between both channels so you can get a true residual measurement. This requires at least a trimmer and a trimmer capacitor.
A spectrum analyzer that can see down to 0.12% would probably be adequate as well.
0.12% is easily measurable with an affordable ADC, so the experiment seems on track. The trick is to get the attenuation matched between both channels so you can get a true residual measurement. This requires at least a trimmer and a trimmer capacitor.
A spectrum analyzer that can see down to 0.12% would probably be adequate as well.
I met Dave he was a neighbor and friend of one of the founders of the planned community I was once in. He came by gratis to make suggestions about the disastrous acoustics in our dining room. We ended up sticking with massive damping in the coffered ceiling ruining the architects design intent.
Since most mics are mechanically resonant well before 20k I'm skeptical that a jangling key or cymbal recording could be accurate enough to really nail the ITD effects.
EDIT - By that I mean the reproduction chain to real life.
Last edited:
I'm not so much concerned with amplitude, and don't have the actual dB numbers at hand.. I'm far more concerned with the ITD. I'll try to find the numbers today.
Will a SA see a phase shift modulation at the 2 to 5 usec interchannel? If the bass is introducing such temporal things, is an fft capable mathematically of discerning it at the level we want?
Jn
- Status
- Not open for further replies.