Hey all,
I am new to audio and am following a reference guide from IR for class D audio.
Attached is an image detailing my inquiry.
My two questions are:
1. Why bother with the inductive LPF at the output, if the speaker itself represents essentially an inductance and resistance in series?
2. Assuming 1. is debunked, what type of inductor should I choose? I usually notice people use large circular airgap inductors in such power applications.
Thanks in advance
I am new to audio and am following a reference guide from IR for class D audio.
Attached is an image detailing my inquiry.
My two questions are:
1. Why bother with the inductive LPF at the output, if the speaker itself represents essentially an inductance and resistance in series?
2. Assuming 1. is debunked, what type of inductor should I choose? I usually notice people use large circular airgap inductors in such power applications.
Thanks in advance
Attachments
There are several class D chip amplifiers out there which are intended to operate directly into speakers, e.g. MAX9703/4.
The principal critical factors in the feasability of such designs are power, frequency, spread spectrum clock (if any), and lead length from amp to speaker. As already pointed out, you can create a load of EMI if you don't know what you're doing. In the worst case this is a criminal offence in most countries.
w
The principal critical factors in the feasability of such designs are power, frequency, spread spectrum clock (if any), and lead length from amp to speaker. As already pointed out, you can create a load of EMI if you don't know what you're doing. In the worst case this is a criminal offence in most countries.
w
Fair enough, I realize the EMI could be an issue but I am more just thinking theoretically.
Furthermore, sizing your LR low pass requires some knowledge of the speaker, as it provides not only your R, but adds to your L. So the question becomes, how does one adequately set the parameters of the LR filter on the output stage? Do I require intricate knowledge of the speaker?
Furthermore, sizing your LR low pass requires some knowledge of the speaker, as it provides not only your R, but adds to your L. So the question becomes, how does one adequately set the parameters of the LR filter on the output stage? Do I require intricate knowledge of the speaker?
Also, once we reach the conclusion the output inductor is necessary as I'm certain is the case, what type is appropriate?
I know of current carrying capacity, but aren't there lots of different types of inductors just like there are capacitors. So air core etc.. Normally I see those big hoopy ones on high power output stages, is there anything special about them, or is it just the only option when looking into large currents flowing?
I know of current carrying capacity, but aren't there lots of different types of inductors just like there are capacitors. So air core etc.. Normally I see those big hoopy ones on high power output stages, is there anything special about them, or is it just the only option when looking into large currents flowing?
No, in fact an (additional) inductor is required neither in theory nor in practice.
Yes, there are many types of inductors.
The subject of your enquiry is beyond the scope of a correspondence on a forum. If you are interested you could do worse than read the datasheets for the chips I have mentioned, or do a search for similar amplifier chips or modules, or simply google 'inductor', 'capacitor', 'filter', etc.
w
Yes, there are many types of inductors.
The subject of your enquiry is beyond the scope of a correspondence on a forum. If you are interested you could do worse than read the datasheets for the chips I have mentioned, or do a search for similar amplifier chips or modules, or simply google 'inductor', 'capacitor', 'filter', etc.
w
I have another question if you don´t mind, I once saw a schematic where there were two output chokes in series between output transistors, and the output and feddback were taken from the midle of them.
What are the andvantages and disadvantages of this compared to standard toplogy?
I think there was mentioned no shoot-through in the output transistors which seems like an advantage, so why isn´t it used more often?
What are the andvantages and disadvantages of this compared to standard toplogy?
I think there was mentioned no shoot-through in the output transistors which seems like an advantage, so why isn´t it used more often?
A simple and generic speaker impedance modelling may be done by connecting in series: voice coil resistance, voice coil inductance and the RLC from fundamental resonance. Iimpedance plots of real speakers for reference are easy to find on the web. Multi-way systems are sometimes quite hard to model because they tend to exhibit several resonances.
But if you have to ask such a question, then you are probably miles away from understanding enough electronics for class D.
PD: Some class D control schemes produce flat frequency response regardless of speaker impedance and output filter resonance.
But if you have to ask such a question, then you are probably miles away from understanding enough electronics for class D.
PD: Some class D control schemes produce flat frequency response regardless of speaker impedance and output filter resonance.
Eva said:
I appreciate your response, but find your presumptions of my electronics knowledge inappropriate. I am not being defensive, however feel it added no benefit to the thread and you may wish to consider my remarks before responding as such in the future.
Regardless, I am simply new to Audio and will cease to further this thread and will just do the literature research on speaker modelling and appropriate filtering thereof.
Skeebopstop!
Impedances of speakers are very different, but an amplifier have to work well with (almost) all of them, so you mustn't calculate the filter based on the modell of a speaker. It should be independent from speaker impedance (for a certain degree, within certain limits).
This is a very simple (and not really accurate) modell of a single way speaker:
http://users.hszk.bme.hu/~sp215/elektro/Szub impedancia.gif
Impedances of speakers are very different, but an amplifier have to work well with (almost) all of them, so you mustn't calculate the filter based on the modell of a speaker. It should be independent from speaker impedance (for a certain degree, within certain limits).
This is a very simple (and not really accurate) modell of a single way speaker:
http://users.hszk.bme.hu/~sp215/elektro/Szub impedancia.gif
Pafi said:
Thanks mate.
That explains it quite well. I referenced an IR AN on Class D and their output filter seemed calculated based purely on the 'resistive' component of the speaker, which was the source of my confusion.
What you just described would explain why, as purchasing a speaker based on its resistance is certainly an industry standard thing to do, whereas the rest of the model is very different!
For modulators with post-filter feedback, whose frequency response does not depend much on output filter resonance, there are other criteria for designing the output filter:
- The amount of "resonant" current flowing in the inductor when the amplifier is idle (it determines idle losses together with switching crossover time imposed by gate charge/discharge currents, and parasitic capacitance at the switching node).
- The load imposed by output capacitor at high frequencies (it may end up being a lower impedance than the speaker).
- The maximum output swing at high frequencies (due to the attenuation that arises when the filter is loaded).
You have plenty of research to do before you can understand how to design a proper output filter.
- The amount of "resonant" current flowing in the inductor when the amplifier is idle (it determines idle losses together with switching crossover time imposed by gate charge/discharge currents, and parasitic capacitance at the switching node).
- The load imposed by output capacitor at high frequencies (it may end up being a lower impedance than the speaker).
- The maximum output swing at high frequencies (due to the attenuation that arises when the filter is loaded).
You have plenty of research to do before you can understand how to design a proper output filter.
I would definitely propose to use a damper RC in parallel to the cap of the output filter - even in amps without post filter feedback.
In case of post filter feedback it is helpful to reduce the influence of the connected speaker - which might be helpful for stability without load.
The required values are depending on the values of L & C and intendend speaker impedance and may be also depending on the feedback structure.
A good starting point for ordinary filters is 10 Ohms & 1.5uF.
But take care - the 10 Ohms might generate quite some heat during tests with large 10kHz rectangular signals or similar (simply have a look to the HF-content of the applied signals).
The basic behavior of output filters can be simulated very well, including step response, phase shift, load/no load etc... It is really worth to play with SPICE for this topic.
In case of post filter feedback it is helpful to reduce the influence of the connected speaker - which might be helpful for stability without load.
The required values are depending on the values of L & C and intendend speaker impedance and may be also depending on the feedback structure.
A good starting point for ordinary filters is 10 Ohms & 1.5uF.
But take care - the 10 Ohms might generate quite some heat during tests with large 10kHz rectangular signals or similar (simply have a look to the HF-content of the applied signals).
The basic behavior of output filters can be simulated very well, including step response, phase shift, load/no load etc... It is really worth to play with SPICE for this topic.
In principle, in UcD style self-oscillating modulators there is no point in using RC dampers. All the extra "gain" produced by filter resonance is used to attenuate the own resonance, so the net effect is flat frequency response regardless of the resulting Q of the filter with whatever speaker load that is connected.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.