My view too. Having an actuator for VTF seems tempting, but I cannot for the life of me work out how you could use that to damp the arm/cart oscillations.
Some people on here are re-investigating the old strain gauge cartridges, as they have response to DC there are some interesting possibilities around that in terms of feedback to control VTF and antiskate, but the pot o goop for damping seems less fuss. But fun to muse on new options to polish the vinyl ****.
She's back at uni for last 2 years. Still doing well. Daresbury has offered to fund her PhD, which given the bloodbath that research funding is going to be for the next 4 or 5 years in UK seems to be the sort of thing you bite an arm off for.
Ah, yes, and... did we really have to go through all that as if somebody was in the dark? I have probably measured a thousand drivers. May I ask, have you?
So now that we agree on the basics of operation, is there anything beyond that? Maybe not, maybe there is. And what is meant by 'if an amp is well designed' as you may well get many responses to what that is, so I can hardly be asked to comment on that and I don't want to play God.
But I think it is still fair to ask questions, what does the speaker track, the voltage of the amplifier or the current of the amplifier?
They are not the same. The amplifier 'fixes' and wants to maintain the voltage, but to do so it has to relinquish all control over the current. So the very force that determines the final dB-SPL is being generated by current, this what is the actual physics say. The final force is proportional to current in a dynamic loudspeaker.
There is an emerging hypothesis that current stability is important for the correct function of a conventional Class A-B amplifier as well as for the speaker itself. We know with certainty that aberrations of almost any kind, from poor driver design and many more factors, can modulate the impedance of a speaker, which in turn can modulate the amplifier current - the very thing that the amplifier has no direct control. The base impedance is the DC resistance of the driver and is stable - can we all agree on that? So any added impedance above the Re is not so stable, because here we observe changes that shoves around the current - the word corrupting the current comes to mind. Finally, a current is presented to Re of the driver and that forms an internal voltage, the current here and the voltage formed there determines the final dB-SPL of the driver. This is not only possible to prove with modeling, but also in physical tests. If the current is reduced (and this current has no direct control by the amplifier) to half, the voltage will also be half and the output of the driver will be reduced by exactly 6dB. But the stability of current that formed that new dB-SPL is at the mercy of the less than stable impedance above the Re value of the voice coil.
So if the total impedance is 20 Ohm, then 6 Ohm is a stable impedance being the DC resistance Re and the remaining 14 Ohm is not so stable and Re is not then the dominant part of the impedance. The final dB-SPL of the driver is only relative to the current, this is not just theory, tests with a microphone totally confirms it, measure half current and it becomes -6dB. It is after all a current device.
If the hypothesis proves true, then what can be done to stabilise the current? The use of current EQ is now being done by designers right now - and some of these speakers are definitely on the market. One designer even uses them in two commercial speakers that are 3-way active, hence three amplifiers - and he EQs the current of the amplifier and there is no passive crossovers. Extensive listening tests shows that is audible and for the better. The guy is a physicist and has now convinced several other speaker designers to incorporate current EQ into their designs.
Now I can hardly understand why something like this should somehow be viewed as heretical or whatever. I mean, this is what is happening and why then should we not talk about it? Exchange views, by all means, but shutting it down, is that wise?
Last edited:
Me too. And I want to choose my own amplifier and not get locked in - but I admit, I can in theory see the advantages.
Hi Joe. What exactly is the acoustic advantage of current drive? A disadvantage is that current drive sacrifices damping.
Yes, I am under that misconception. Mainly because your sig links to a current-driven speaker you designed. Also, you complain about the variable impedance of drive units "corrupting" the vocie-coil current when voltage driven.
What are you promoting?
What are you promoting?
Joe,
It is not easy to figure out if you even know what you are talking about from your verbage. I attribute that to the descriptors you are using, which are somewhat unconventional.
A hybrid control scheme might sound better or worse depending on implementation,I've never tried nor heard it.
In motion control, they like to use a parameter called feed forward velocity to lower the following error(distortion). It is a predictor based on the desired trajectory, giving stimulus to the output to increase accuracy without relying on negative feedback. It can work very well as long as the device is linear, I can get motion errors down in the 200 nanometer range where simple NFB can't do it due to stability issues with high gains.
If a bass driver is assumed linear, I wonder if velocity and acceleration feed forwards could assist in reducing distortion. This in fact kind of sounds like what you speak of.
And yes, I have measured speakers, as well as some other e/m objects to accuracies well beyond what is considered SOTA. Measuring thousands of widgets doesn't necessarily equate to an understanding of the underlying physics, just an understanding of how they react to stimulus. Neither good, nor bad. Just is.
Ps..
The only snag I can think of, is what parameter needs to be optimized.. Acceleration, velocity, or pressure of the air in front of the driver.
Jn
It is not easy to figure out if you even know what you are talking about from your verbage. I attribute that to the descriptors you are using, which are somewhat unconventional.
A hybrid control scheme might sound better or worse depending on implementation,I've never tried nor heard it.
In motion control, they like to use a parameter called feed forward velocity to lower the following error(distortion). It is a predictor based on the desired trajectory, giving stimulus to the output to increase accuracy without relying on negative feedback. It can work very well as long as the device is linear, I can get motion errors down in the 200 nanometer range where simple NFB can't do it due to stability issues with high gains.
If a bass driver is assumed linear, I wonder if velocity and acceleration feed forwards could assist in reducing distortion. This in fact kind of sounds like what you speak of.
And yes, I have measured speakers, as well as some other e/m objects to accuracies well beyond what is considered SOTA. Measuring thousands of widgets doesn't necessarily equate to an understanding of the underlying physics, just an understanding of how they react to stimulus. Neither good, nor bad. Just is.
Ps..
The only snag I can think of, is what parameter needs to be optimized.. Acceleration, velocity, or pressure of the air in front of the driver.
Jn
Last edited:
From your drawing, you just have to maintain sufficient distance between coppers of different polarity where the air gaps will be. Caps by design tend to run voltage as high as the plastics can withstand by going thin, and air breaks down around 50 to 70 volts per mil. Nomex is 1300 volts per mil, kapton about 6 kV, don't know other materials off the top of my head, epoxies run about 400 though.
Jn
Is that equivalent to what is called predistortion in other fields? I have an infinity servostatic woofer which uses accelerometer feedback, but in a fairly basic analog method. Considering more advanced schemes I always wonder if you can accurately enough characterize performance with a complex waveform to get a good result, vs doing a proper job and getting a driver that is loafing along (think 15" as per M2).
Note I was not good at control theory as an undergrad. With the ball and beam lab I could not get it to act as a trebuchet however hard I tried 🙂
Last edited:
It does sound like it.
The advantage in motion control is that the path is pre-calculated, so the amount of goose is also determinable. Random sound, there would have to be a delay so the calculations could be done in advance.
One task I have on the plate is getting a very nonlinear slave device to track a master device with high accuracy, mainly because I do not have permission to fix the master software. My current plan is to code in adaptive tuning, making PID and feed forwards a function of position.
I'm confident that these types of schemes could be used at least into upper midrange given all the display horsepower now available off the shelf.
Edit. Hmm, just remembered, high order cabs (reflex, TL, 6th, would require memory quite a few cycles deep in bass, maybe 10?) the algorithms would need at least 500 milliseconds to be bass accurate. Also, I suspect it would mean bi and triamping, each speaker algorithm would be daunting by themselves, never mind one channel through a passive crossover..
Jn
Last edited:
There are things permitted in reproduction algorithms that would be quite silly in live reinforcement.
I am afraid you spend time chasing a vague target .
Neverteless, regarding the question "what causes it" I would suggest to get a cheap car speaker driver, plot the complex impedance/phase of it on air, housed on an enclosure and finally with the motor locked (glue the coil former on the pole piece). Extract the real and imaginary parts on each case, study and compare them.
George
Regardless of whether you are in position to appreciate what some participants like jneutron and Scott can offer to broaden your knowledge for free, on good will and under some proven professional excellency record, you really are acting negatively when you reject the offered basic answers to your basic questions.
George
- Status
- Not open for further replies.