If you measure the amp as a black box, then I'd argue it improves PSRR.
Unfortunately, it makes sense John
What excitation signal would you suggest for measuring it ?
And for this? (Ed’s wind is one option)
George
Ed, that is in line with my reasoning. A considerable outside force on the cone is required to, dynamically, lower the impedance of a driver to below it's Rdc.
John, this is exactly what I argued, you need movement in the opposite direction of where the amp wants to steer it for the impedance of a driver to dive below Rdc. You call it 'when the amplifier is trying to brake the system', which is the same as what I argued: the movement of the driver needs to be in anti-phase with the signal for this situation to happen. Where we differ is in that in my view this situation cannot occur in real life without exerting an external force on the cone.
The reason some loudspeakers have freakish impedance curves approaching the x-axis is inept crossover design, not back EMF.
John, this is exactly what I argued, you need movement in the opposite direction of where the amp wants to steer it for the impedance of a driver to dive below Rdc. You call it 'when the amplifier is trying to brake the system', which is the same as what I argued: the movement of the driver needs to be in anti-phase with the signal for this situation to happen. Where we differ is in that in my view this situation cannot occur in real life without exerting an external force on the cone.
The reason some loudspeakers have freakish impedance curves approaching the x-axis is inept crossover design, not back EMF.
Unfortunately, it makes sense John
What excitation signal would you suggest for measuring it ?
I'd go with three frequencies simultaneously which show as worst case independently.
And for this? (Ed’s wind is one option)
George
I'd frequency sweep a sinx/x pulse to determine worst case. Using that waveform, I'd plot the V/I response to the system.
If it's resonant or underdamped, the amp will be called upon to quash the movement.
Quash being of course, a technical term...
As to quashing Ed's "wind", perhaps diet??
John
Ya gotta love that low hanging fruit...
We are discussing a mechanical system which converts electrical energy to acoustic. While the conversion efficiency to the acoustic impedance of the external world may be in the 1 to 2% realm that is not the true conversion efficiency of the voice coil mechanism. The voice coil and amplifier do indeed see the acoustic mass of the enclosure, as well as any storage mechanism/resonant system of the enclosure design, reflex or bandpass.
As such, the air mass coupled to the cone can provide tremendous forces on the cone.
John
Yes Vaccy,
You did get the mechanism correct. But your experience is geared to what you do. That is why I mentioned this is one of the areas where I bet with other loudspeaker designers. So I am an evil cuss...
A fair number of times folks have gotten into trouble ignoring the funny things that happen. I do have to explain to amplifier manufacturers that two 8 ohm loudspeakers can present a 2 ohm load to an audio power amplifier. (JN we don't use passive crossovers so multitone is not an issue.
I have had to explain the the Smith and Larsen woofer test will not give accurate readings over long loudspeaker lines. (Cable capacitance.)
Some of the DSP based audio power amplifiers try to show load impedance. Requires quite a bit to get it to actually work in the outside world.
Now the easy way to measure minimum loudspeaker impedance is to use a high value resistor fed from three properly summed oscillators and a true RMS meter. (More or Less)
You did get the mechanism correct. But your experience is geared to what you do. That is why I mentioned this is one of the areas where I bet with other loudspeaker designers. So I am an evil cuss...
A fair number of times folks have gotten into trouble ignoring the funny things that happen. I do have to explain to amplifier manufacturers that two 8 ohm loudspeakers can present a 2 ohm load to an audio power amplifier. (JN we don't use passive crossovers so multitone is not an issue.
I have had to explain the the Smith and Larsen woofer test will not give accurate readings over long loudspeaker lines. (Cable capacitance.)
Some of the DSP based audio power amplifiers try to show load impedance. Requires quite a bit to get it to actually work in the outside world.
Now the easy way to measure minimum loudspeaker impedance is to use a high value resistor fed from three properly summed oscillators and a true RMS meter. (More or Less)
A fine result, but surely one would expect that with any design with high levels of feedback? Sleep deprived today so may have missed something obvious.
CiteSeerX — A General Relationship between Amplifier Parameters and its Application to PSRR Improvement
shows the limit when the dominant pole compensation is referenced to a ps rail: the psrr for that rail follows loop gain
2-pole compensation and high loop gain gives more audio frequency loop gain
in discrete we can modify that with several other techniques such as ground referencing the compensation C
or diffing out the ps-gnd ac with a cancelling C to the other side of the mirror
the new OPA1622 brings out the compensation reference on a pin so it can be grounded - very rare for a op amp
Last edited:
The temp mentioned in the section in Motchenbacher and Fitchen is "approximately 300 C".
Last edited:
I just went and looked at Cordell's book as his chapter on class D was referenced.
The amusing thing about the "correct" TI article is they don't say much about how they close the loop. Stanley has warned about doing this in the symbol domain, as he told me Nielsen was claiming to do.
The funniest experience some people had at Harman was when one Japanese cap manufacturer had developed a special series of electrolytic caps and wanted Harman to hear how much better they made the audio sound. They actually showed up with a boom box that had been fitted with those new parts, and expected Harman's golden-eared listeners to hear the profound difference. I missed the demo which was probably as well. It was not even a sighted A-B comparison between two boom boxes!
Thanks John. Transferred to the “For to do” list.
That fruit, escapes me
Ed, you are turning cryptic again. I promise I’ll pass the ouzo
George
An old integrator reset switch idea. The emitter is now the collector though and the reverse beta starts out so bad not sure this is a fair vehicle to look for the problem.
Yes, inverted mode worked well for audio muting too. Toshiba developed a part with good reverse beta and high emitter-base breakdown voltage, the 2SC2878.
The circuits were chopper stabilized ones. I believe they were indeed integrator resets..
John
The division between amplifier and CM and cascode is sort of a line in the sand. Making things black and white isn't really necessary and can mislead people into believing in divisions that don't really exist in practice - "A CM won't improve PSRR so I won't use it". This is the #1 thing that keeps anyone from being able to explain electronics to their grandmother.
That's one way of looking at it.
We can't just leave the PSRR of the CM flapping in the wind can we? It adds to the amp's PSRR. And many amps have CMs on the PCB - and take it's PSRR with them when they are disconnected from the PSU. Or can a CM not be part of an amplifier design? If you were making a datasheet for such an amp would you specify a PSRR that was not equal to the PSU ripple divided by the output ripple?
Every cascode is essentially a mini-regulator anyways. If we reduce everything to black and white nothing will make sense anymore. A CM is practically just a cascode covering multiple stages at once. Often people will understand things better if rather than simplifying, you just present a picture that is factually consistent. Then instead of learning terminology and trying to derive function from it, they learn function first and then terminology just fits into place. And they don't learn the bad habit of doing it backwards.
- Status
- Not open for further replies.