I'm working on a design that uses a 'filter less class d' amplifier (TFA9882UK).
Since this is my first class-d design, as a sanity check, I'm wondering if there will be any audible distortion from not filtering out the class-d PWM waveform? The speaker is a 20uH 4 ohm speaker.
I've attached pictures of the voltage waveform at the speaker terminal. One is the voltage with the filter and the other is non-filtered. I'm convinced that while the switching frequency is an order of magnitude larger than the audible range there will still be negative audible effect from not putting the PWM signal through a low pass filter.
Thank you!
Since this is my first class-d design, as a sanity check, I'm wondering if there will be any audible distortion from not filtering out the class-d PWM waveform? The speaker is a 20uH 4 ohm speaker.
I've attached pictures of the voltage waveform at the speaker terminal. One is the voltage with the filter and the other is non-filtered. I'm convinced that while the switching frequency is an order of magnitude larger than the audible range there will still be negative audible effect from not putting the PWM signal through a low pass filter.
Thank you!
Attachments
The audible difference is very low. But, unfiltered amps irradiates high RF levels, and can overheat the voice coil and or the pieces in the magnetic path.
As a ham radio, I suggest to use filtering, please !!! HF bands are unusable because of higher noise levels from PWM supplies and amplifiers.
As a ham radio, I suggest to use filtering, please !!! HF bands are unusable because of higher noise levels from PWM supplies and amplifiers.
I agree, but how to legalize it, and how to control them.
And, add to the list frequency converters for AC motors, servos, brush DC motors, microwave ovens, AC phase angle TRIAC speed control, and a very extended list.
As ham, I definitively will not use any HF band down to 10 meters. I usually receive a LT newspaper printed as a magazine, and when I see that some switching regulators are up to 5 MHz, I think in our old 80 m band, a single 6DQ6 with a pi coupled and a regen RX. Impossible nowadays.
Most countries have EMC regulations, but most countries don't bother to enforce them. If there was a general climate where interference was deprecated, rather than tolerated, and commercial equipment was blameless then there would be social pressure on DIYers to do the same.
I have heard that the UK military are getting concerned that if they had to use HF communications within the UK (e.g. during some disaster/emergency situation) then it would not work unless the local electricity supply was switched off so that interferers were disabled. Of course in a disaster the power often goes off anyway, but if it can be on it should be on wherever possible.
I have heard that the UK military are getting concerned that if they had to use HF communications within the UK (e.g. during some disaster/emergency situation) then it would not work unless the local electricity supply was switched off so that interferers were disabled. Of course in a disaster the power often goes off anyway, but if it can be on it should be on wherever possible.
As much as unwanted advertisements everywhere.
There are well known and established tests to measure the amount of RF energy radiated and conducted into mains lines by appliances to ensure radio communications are not disrupted. Filter-less class D can comply in certain conditions, though.
Last edited:
"I have heard that the UK military are getting concerned that if they had to use HF communications within the UK (e.g. during some disaster/emergency situation) then it would not work unless the local electricity supply was switched off so that interferers were disabled."
I can see them waiting for the tubes to warm up before they start tapping. If military is ever forced to use analog HF communication the power in the area will already be down. Besides, I'm sure they have the ability to select any ERP that makes Their things work.
I can see them waiting for the tubes to warm up before they start tapping. If military is ever forced to use analog HF communication the power in the area will already be down. Besides, I'm sure they have the ability to select any ERP that makes Their things work.
What ! No filters ?
I've just posted a related thread on this. I have found that class d amps all leak some hf despite filtering and this has a blurring effect on the sound. It's like a background noise is present that you can't hear. It's confusing because the digital amp is still producing lots of detail. At low levels the switching frequency leakage is far stronger than the actual signal. I do worry for my ribbons but they;ve not blown yet. I fixed my old Tripath 300 watter with a low pass 24 octave filter between the amp board and the speaker terminals.
In short I think it may sound bad but far worse is the interference and shocking signals sent to tweeter. My vote is for even more filtration.
Incidentally my original class T amp jammed all FM listening for a range of 10metres.
I've just posted a related thread on this. I have found that class d amps all leak some hf despite filtering and this has a blurring effect on the sound. It's like a background noise is present that you can't hear. It's confusing because the digital amp is still producing lots of detail. At low levels the switching frequency leakage is far stronger than the actual signal. I do worry for my ribbons but they;ve not blown yet. I fixed my old Tripath 300 watter with a low pass 24 octave filter between the amp board and the speaker terminals.
In short I think it may sound bad but far worse is the interference and shocking signals sent to tweeter. My vote is for even more filtration.
Incidentally my original class T amp jammed all FM listening for a range of 10metres.
Bad EMC practice, a problem often overlooked by DIYers, and SMPS Class d are realy require specialised layout and EMC testing. The most important factor for minimal EMC and thus as a result the best audio quality out is the layout. Some information regarding switcher design and EMC.
http://www.diyaudio.com/forums/power-supplies/193705-switcher-emc-design.html
http://www.diyaudio.com/forums/power-supplies/193705-switcher-emc-design.html
Power line communication adds to this problem, this has an exemption from EMC regs. because it works by adding noise to the power lines... Even military radios have to pass complience testing for any country your going to sell them to.
Hi,
I am under the impression that for class D you need the filter to achieve
the high efficiency that presumably is the reason for choosing class D.
The filter cuts switching losses by returning power to the supply,
so no filter really isn't a good option ? Am I missing something ?
rgds, sreten.
I am under the impression that for class D you need the filter to achieve
the high efficiency that presumably is the reason for choosing class D.
The filter cuts switching losses by returning power to the supply,
so no filter really isn't a good option ? Am I missing something ?
rgds, sreten.
As a side effect the LC filter is also helpful to minimize switching losses during idle and low output power, because it forces the sloping of the halfbridge during dead time and the switches can run pretty relaxed almost in ZVS mode. Depending on the filter design and components the filter itself can have HF losses in a similar range, which is saved in the switches...
During full power the filter does not help to reduce losses.
Without filter you will of course pass tons of high frequency signal to the speaker, but this HF is not identical to harmonics of the audio signal.
It is a rectangle of several hundreds of kHz, which can of course be expressed by fourier analysis as a sine of multiple hundreds kHz and lots of odd order harmonics of the multiple hundred kHz fundamental.
The frequency content of all this is by one order of magnitude and more above the audible frequency range.
How much of this will come to the speaker is depending on wiring, x-over and speaker impedances.
Most tweeters won't generate much mechanical move from that.
Also the currents in the coil will be low, because of the inductance of the wiring and the voice coil itself.
Audible irritations of the speaker can theoretically happen by nonlinear behavior of the speaker only. No doubt such nonlinearities are present in tweeters and they are not neglectible. Whether this effects cause impact on the sound or not, will depend very much on the speaker itself and on the amount of HF that you feed to the speaker.
The fact that the sound changes when we add a second filter at the output of the class D amp, can - but does not necessarily implicate that the HF irritated the speaker.
Even if the second filter is designed that it does not directly hit into the transfer function below 20kHz, it will definitely change the driving impedance in the higher frequency range, which may also have influence on the sound.
During full power the filter does not help to reduce losses.
Without filter you will of course pass tons of high frequency signal to the speaker, but this HF is not identical to harmonics of the audio signal.
It is a rectangle of several hundreds of kHz, which can of course be expressed by fourier analysis as a sine of multiple hundreds kHz and lots of odd order harmonics of the multiple hundred kHz fundamental.
The frequency content of all this is by one order of magnitude and more above the audible frequency range.
How much of this will come to the speaker is depending on wiring, x-over and speaker impedances.
Most tweeters won't generate much mechanical move from that.
Also the currents in the coil will be low, because of the inductance of the wiring and the voice coil itself.
Audible irritations of the speaker can theoretically happen by nonlinear behavior of the speaker only. No doubt such nonlinearities are present in tweeters and they are not neglectible. Whether this effects cause impact on the sound or not, will depend very much on the speaker itself and on the amount of HF that you feed to the speaker.
The fact that the sound changes when we add a second filter at the output of the class D amp, can - but does not necessarily implicate that the HF irritated the speaker.
Even if the second filter is designed that it does not directly hit into the transfer function below 20kHz, it will definitely change the driving impedance in the higher frequency range, which may also have influence on the sound.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.