if there is such a thing, you may be over-researching a subject.
It is nice to see what others have learned on the subject matter. But unless you actually do it yourself, your learning will be limited.
It is nice to see what others have learned on the subject matter. But unless you actually do it yourself, your learning will be limited.
I think .... no offense here, the basics have completely eluded you. For instance, please, give me at least ONE GOOD REASON why in the hell you'd want to design something in an "unorthodox way"..... does your car have square tires too?
Seriously, work out that pro Vs cons list already mentioned, do a little research on the matter to help you get it done.. you're not going to want to keep that design objective for very long.
It's expensive, it's clunky, it's silly.
Oh, and forget about animals would you? Can't take you seriously at all when you mention that. Animals shouldnt' be listening to amps anyway.
Seriously, work out that pro Vs cons list already mentioned, do a little research on the matter to help you get it done.. you're not going to want to keep that design objective for very long.
It's expensive, it's clunky, it's silly.
Oh, and forget about animals would you? Can't take you seriously at all when you mention that. Animals shouldnt' be listening to amps anyway.
zenmasterbrian!
There is absolutely no exact limit! In some cases 8 kHz is acceptable (eg. vehicle diving), but nobody can guarantee that 40 kHz can't cause problem to your pets. If this bother you, then simply try it!
2 well known examples of audible freq sources: TV 15625 Hz, FM Radio pilot signal 19 kHz. Some of the radios pilot is not filtered out, and I didn't hear anybody complaining about this (unless it interferes with something, eg.: oscillator of tape recorder).
Are you really able to controll a loop contains a multipole LPF over cutoff freq?
And are you able to design and make some ~300uH 30A 20mohm choke wich have low loss at 20 kHz?
And do you know that a base of saturated, big BJT needs several amperes negative current to turn off "fast", and ~10% of the collector current to keep is saturated? Are you aware of all problems, or you just saw an SMPS, and found it simple?
There is absolutely no exact limit! In some cases 8 kHz is acceptable (eg. vehicle diving), but nobody can guarantee that 40 kHz can't cause problem to your pets. If this bother you, then simply try it!
2 well known examples of audible freq sources: TV 15625 Hz, FM Radio pilot signal 19 kHz. Some of the radios pilot is not filtered out, and I didn't hear anybody complaining about this (unless it interferes with something, eg.: oscillator of tape recorder).
Are you really able to controll a loop contains a multipole LPF over cutoff freq?
And are you able to design and make some ~300uH 30A 20mohm choke wich have low loss at 20 kHz?
And do you know that a base of saturated, big BJT needs several amperes negative current to turn off "fast", and ~10% of the collector current to keep is saturated? Are you aware of all problems, or you just saw an SMPS, and found it simple?
mmmhmmm. Without looking into it any further too I'm sure.
Still I'm unable to take this at all seriously. Let me show you what I mean:
"I want a 1000W BJT switchmode amp, but how low can the frequency be before it becomes harmful to animals".
Duuuuuuuuuuuuuuhhhh. Most likely it will be the 100Hz at 1000W that your animals dislike.
troll
Zen,
There are all kinds of things that switch at low kHz... look at AC motor drives for machine tools... 4 kHz ish is very normal.
ClassD audio is probably closest to sine-inverter technology. Those are running 8 to 30 kHZ these days.
But for audio, you're seeking info that was never there in the first place. Class D audio wasn't invented really. Variable speed AC motor drives etc... had put all the theory in place. Audio came into the picture when switching speeds of 100 kHz and up became feasible. In a way, you are looking for milestone in the development process that the audio guys were smart enough to leapfrog over.
🙂
There are all kinds of things that switch at low kHz... look at AC motor drives for machine tools... 4 kHz ish is very normal.
ClassD audio is probably closest to sine-inverter technology. Those are running 8 to 30 kHZ these days.
But for audio, you're seeking info that was never there in the first place. Class D audio wasn't invented really. Variable speed AC motor drives etc... had put all the theory in place. Audio came into the picture when switching speeds of 100 kHz and up became feasible. In a way, you are looking for milestone in the development process that the audio guys were smart enough to leapfrog over.
🙂
Guys, let this thread die now. It is obvious it will continue to head no where. I still haven't seen answers to my two questions like 3-4 pages back. They are simple answers (for those who understand) and answering those two should make it very clear there is no problem with audible 23.5kHz. Of course, this assumes the 100dB filter can be designed. Once again, I am out. Hmm, maybe not let it die, sort of entertaining to read.
-SL
-SL
I just checked, there are 1200 topics in the class D section. So if people want to talk about class D running over 30khz, there are 1199 topics they can post on.
Fortunately, a number of people have raised very relevant issues, and I'm going to try to speak to them. If I miss something, it was just an oversight.
My own interest in Class D goes back to the 1970's, when these higher frequencies were not being used. Class D was interesting, but not well accepted.
My present interest is only on the context of multiamping and powered speakers.
Of course only the LF drivers need that kind of power.
The example was mentioned of the TV horizontal sweep, 15625 is guess in Europe? 15750 in the US.
I do hear this sometimes, and that is without a loudspeaker. It is of course from the magnetics.
The 19khz stereo subcarrier for FM radio is a good point. I've never heard that. I believe it is filtered out. But since FM radio has to go to 15khz audio, good filtering must be real difficult.
TO BE CONTINUED
Fortunately, a number of people have raised very relevant issues, and I'm going to try to speak to them. If I miss something, it was just an oversight.
My own interest in Class D goes back to the 1970's, when these higher frequencies were not being used. Class D was interesting, but not well accepted.
My present interest is only on the context of multiamping and powered speakers.
Of course only the LF drivers need that kind of power.
The example was mentioned of the TV horizontal sweep, 15625 is guess in Europe? 15750 in the US.
I do hear this sometimes, and that is without a loudspeaker. It is of course from the magnetics.
The 19khz stereo subcarrier for FM radio is a good point. I've never heard that. I believe it is filtered out. But since FM radio has to go to 15khz audio, good filtering must be real difficult.
TO BE CONTINUED
<<<
Are you really able to controll a loop contains a multipole LPF over cutoff freq?
>>>
This is an important issue. I have not worked this out entirely. It is a root locus problem, and phase margin problem.
The high order LPF will put poles in a semi circle or an elipse. Loop gain could push them over into the right half plane.
I see as a starting point and audio bandwidth of 300hz, a filter 3db down point of 1khz, and a switching freq of 23.5khz.
Notice that these relations are the same as for the amps that switch at 100khz.
The second order LPF's give conjugate poles in the left half plane. The issues are the same.
These are some of the initial options I see for making it work:
1. Provide additional compensation in the feedback path.
2. Provide feedfoward compensation.
3. Have two closed loops. One outside the filter, the second including it.
4. Raise my filter 3db down point some, inoder to introduce a dominant pole, between 300hz and this filter. This could allow a very very high loop gain for the lower part of that audio range.
These are some of the options. Overall, I am comfortable that this can work because of the narrow 300hz audio bandwidth. That opens up lots of possibilities.
TO BE CONTINUED
Are you really able to controll a loop contains a multipole LPF over cutoff freq?
>>>
This is an important issue. I have not worked this out entirely. It is a root locus problem, and phase margin problem.
The high order LPF will put poles in a semi circle or an elipse. Loop gain could push them over into the right half plane.
I see as a starting point and audio bandwidth of 300hz, a filter 3db down point of 1khz, and a switching freq of 23.5khz.
Notice that these relations are the same as for the amps that switch at 100khz.
The second order LPF's give conjugate poles in the left half plane. The issues are the same.
These are some of the initial options I see for making it work:
1. Provide additional compensation in the feedback path.
2. Provide feedfoward compensation.
3. Have two closed loops. One outside the filter, the second including it.
4. Raise my filter 3db down point some, inoder to introduce a dominant pole, between 300hz and this filter. This could allow a very very high loop gain for the lower part of that audio range.
These are some of the options. Overall, I am comfortable that this can work because of the narrow 300hz audio bandwidth. That opens up lots of possibilities.
TO BE CONTINUED
<<<<
And are you able to design and make some ~300uH 30A 20mohm choke wich have low loss at 20 kHz?
>>>>
The average power put out by this amp will not be very much. The inductors will be air core, so they won't saturate.
The filter components would be the very components used for speaker crossovers. They would have the same value and physical size.
Going to high order does not increase the values of each component, it decreases it.
Doing this compactly is not important to me. This is for a powered speaker. I could even imagine this filter laid out under a piece of plexiglass.
Why shouldn't solid state be as showy as the tube amp stuff?
<<<
And do you know that a base of saturated, big BJT needs several amperes negative current to turn off "fast", and ~10% of the collector current to keep is saturated? Are you aware of all problems, or you just saw an SMPS, and found it simple?
>>>
Yes, I do know about BJT drive requiremens and would have made the same estimates you did.
My interest in Class D goes back to the 1970's. Here my interest is only in the context of multi-amping.
TO BE CONTINUED
And are you able to design and make some ~300uH 30A 20mohm choke wich have low loss at 20 kHz?
>>>>
The average power put out by this amp will not be very much. The inductors will be air core, so they won't saturate.
The filter components would be the very components used for speaker crossovers. They would have the same value and physical size.
Going to high order does not increase the values of each component, it decreases it.
Doing this compactly is not important to me. This is for a powered speaker. I could even imagine this filter laid out under a piece of plexiglass.
Why shouldn't solid state be as showy as the tube amp stuff?
<<<
And do you know that a base of saturated, big BJT needs several amperes negative current to turn off "fast", and ~10% of the collector current to keep is saturated? Are you aware of all problems, or you just saw an SMPS, and found it simple?
>>>
Yes, I do know about BJT drive requiremens and would have made the same estimates you did.
My interest in Class D goes back to the 1970's. Here my interest is only in the context of multi-amping.
TO BE CONTINUED
<<<<
There are all kinds of things that switch at low kHz... look at AC motor drives for machine tools... 4 kHz ish is very normal.
ClassD audio is probably closest to sine-inverter technology. Those are running 8 to 30 kHZ these days.
>>>>
This experience could be highly relevant. Are precautions taken to prevent audible sound? At what frequency do they condsider such precautions unnecessary?
<<<<
But for audio, you're seeking info that was never there in the first place. Class D audio wasn't invented really. Variable speed AC motor drives etc... had put all the theory in place. Audio came into the picture when switching speeds of 100 kHz and up became feasible. In a way, you are looking for milestone in the development process that the audio guys were smart enough to leapfrog over.
>>>>
I appreciate your POV. My recolection is different, being that Class D audio came into existence before such high switching speeds were used. True, it was rare, and problematic. The problems were because they were trying to do broad band audio. This is not broadband audio.
So I believe there is experience with the physiological problems of low ultrasonic.
That is what I want to explore.
So if people have such experience, it would be of use. Otherwise lets just let this thread sit.
There are all kinds of things that switch at low kHz... look at AC motor drives for machine tools... 4 kHz ish is very normal.
ClassD audio is probably closest to sine-inverter technology. Those are running 8 to 30 kHZ these days.
>>>>
This experience could be highly relevant. Are precautions taken to prevent audible sound? At what frequency do they condsider such precautions unnecessary?
<<<<
But for audio, you're seeking info that was never there in the first place. Class D audio wasn't invented really. Variable speed AC motor drives etc... had put all the theory in place. Audio came into the picture when switching speeds of 100 kHz and up became feasible. In a way, you are looking for milestone in the development process that the audio guys were smart enough to leapfrog over.
>>>>
I appreciate your POV. My recolection is different, being that Class D audio came into existence before such high switching speeds were used. True, it was rare, and problematic. The problems were because they were trying to do broad band audio. This is not broadband audio.
So I believe there is experience with the physiological problems of low ultrasonic.
That is what I want to explore.
So if people have such experience, it would be of use. Otherwise lets just let this thread sit.
OK, that's true, but then why do you want extremely low loss on semiconductors? Their losses are proportional to the output power too! Making a 1% loss switch and 20% loss filter has no much sense from my point of view.
Air core? You must be mazochist! Do you want to build a VLF radio station, or are you going to build really gigantic toroids? Do you want to spend as much money on copper, as a whole PWM amp module costs? 😀
Then you will build a 20 % loss filter (at least), wich radiates magnetic VLF signal.
Solve it! How much phase margin do you have at gain=1? Its already negative!
With 2nd order, and a single P controller phase margin is little, but positive at any gain. With PD controller, phase margin can be increased to ~90 degrees. No problem.
What controller would you use with multipole LPF? PDDDD? 🙂 (Theoretically it could work, put practically not realisable.) Or cascade controller, if you place current sensors on the chokes. Very "simple" and "economical"...
If impedance and cutoff freq is fixed, then values are the same (exept for the first and last components). If freq is lower, then values should be obviously higher. If attenuation and freq were fixed, then you would be right, value would be decreased, but your freq is lower, and you want much bigger attenuation! The mentioned L=300uH allowes ~5 kHz freq at Rl=4ohm. Would you really use lower inductance then this?
If you accepted a lower attenuation, lower component values, then idle current (idle losses) would be increased. It's already 3-4A at L=300 uH. (First element should be L/2!)
Calculate these things, and if you get reasonable results, then just build it!
BTW: do you have any pets?
And let's compare this waste of copper to an average fullband 1 kW ClassD amp: ~500g mass, 25W+8%PO loss (mostly on heat sink), ~40$ parts cost.
rather than building one, you may be able to get a large smps power supply and change the feedback scheme to make it work as a power amp.
Pafi,
Power lost in the semiconductors make for the power limit of the amp. Power lost in the filter inductors does not contribute to this limit.
I'm looking at a LPF at 1khz. Its about the same as a speaker crossover at 1khz. For audiophile gear those usually are air core. Sometimes even foil wound. Its not that big.
Remember, this is a big power subwoofer. The electronics will be laid out so it looks nice, and be under plexiglass.
there can be metal screening for shielding.
The closed loop problem can be made to work. It only has to pass a 300hz audio signal max. I could lower this. I could raise the filter higher, so there could be dominant pole between the top of the audio band and the filter. Then I have very high phase margin. There are lots of ways to do this.
i do not have any pets
Power lost in the semiconductors make for the power limit of the amp. Power lost in the filter inductors does not contribute to this limit.
I'm looking at a LPF at 1khz. Its about the same as a speaker crossover at 1khz. For audiophile gear those usually are air core. Sometimes even foil wound. Its not that big.
Remember, this is a big power subwoofer. The electronics will be laid out so it looks nice, and be under plexiglass.
there can be metal screening for shielding.
The closed loop problem can be made to work. It only has to pass a 300hz audio signal max. I could lower this. I could raise the filter higher, so there could be dominant pole between the top of the audio band and the filter. Then I have very high phase margin. There are lots of ways to do this.
i do not have any pets
I know that the standard now is for switching amps and power supplies to run at 40khz, 100khz, or even more.
But this was not always the case.
Looks like Poobah and others have experience with SMPS and inverters in the 4khz to 30khz range. That is very relevant.
Right now, I just want to consider the ultrasonic physiological issues. There must be generally accepted standards for how much is allowable in various frequency bands.
There has been a history of power supplies in the range of interest.
There are also ultrasonic sensors. There must be in that industry some conventions about how low you can go, and at what power levels.
If I had such material I'd of posted it.
So lets just let this thread sit. Some folks will come along that have experience with low ultrasonics, 23.5khz, and what power level is in some way perceptible with prolonged exposure.
This is the first issue. I want to deferre all others until some more progress is made here.
So this thread can sit, unless people have experience with the physiological issues of low ultrasonics.
But this was not always the case.
Looks like Poobah and others have experience with SMPS and inverters in the 4khz to 30khz range. That is very relevant.
Right now, I just want to consider the ultrasonic physiological issues. There must be generally accepted standards for how much is allowable in various frequency bands.
There has been a history of power supplies in the range of interest.
There are also ultrasonic sensors. There must be in that industry some conventions about how low you can go, and at what power levels.
If I had such material I'd of posted it.
So lets just let this thread sit. Some folks will come along that have experience with low ultrasonics, 23.5khz, and what power level is in some way perceptible with prolonged exposure.
This is the first issue. I want to deferre all others until some more progress is made here.
So this thread can sit, unless people have experience with the physiological issues of low ultrasonics.
Hi zenmasterbrian,
Keep in mind that industrial limits are higher than in home limits. I've never seen homeowners walking around with ear plugs constantly.
Right at the front door: "Hearing protection must be worn beyond this point".
-Chris
Keep in mind that industrial limits are higher than in home limits. I've never seen homeowners walking around with ear plugs constantly.
Right at the front door: "Hearing protection must be worn beyond this point".
-Chris
Poobah,
Great idea to use the Yahoo search to find information you want to know. It seems so much more efficient than waiting for someone to come along with the answer...You are a very clever fellow 😉
I suppose that Google.com would work too, right?
Great idea to use the Yahoo search to find information you want to know. It seems so much more efficient than waiting for someone to come along with the answer...You are a very clever fellow 😉
I suppose that Google.com would work too, right?
- Status
- Not open for further replies.