The question must be this one:
at the tpa3116 i have "just" 1,5db?....is the TPA3116 not sensitive of impedance-missmatch because of this frequency adaptation that ICG wrote in post 24?
Before the experts arrive, I will give it a shot.
The speaker impedance acts as damping of the output filter resonance. When the damping is reduced with a higher speaker impedance, the frequency plot will show a rise near the resonance frequency. That may affect the high treble a little.
The audible range is below this resonance and the carrier frequency is solidly above the resonance frequency. ICG has divulged statements that the carrier frequency is not fully fixed but varies in order to reduce distortion. Perhaps from the same principles as class T (Tripath). But, for EMI reasons, I doubt that the frequency variations from the 400KHz nominal carrier are large. It is rather a spectrum that is spread-out a bit around the 400KHz. As the 400KHz is already well above the filter resonance frequency, I doubt you will notice any practical influence from this variation of frequency on the response of the filter resonance.
Now we are all awaiting The Doctor, ICG and voltwide among other.
at the tpa3116 i have "just" 1,5db?....is the TPA3116 not sensitive of impedance-missmatch because of this frequency adaptation that ICG wrote in post 24?
Before the experts arrive, I will give it a shot.
The speaker impedance acts as damping of the output filter resonance. When the damping is reduced with a higher speaker impedance, the frequency plot will show a rise near the resonance frequency. That may affect the high treble a little.
The audible range is below this resonance and the carrier frequency is solidly above the resonance frequency. ICG has divulged statements that the carrier frequency is not fully fixed but varies in order to reduce distortion. Perhaps from the same principles as class T (Tripath). But, for EMI reasons, I doubt that the frequency variations from the 400KHz nominal carrier are large. It is rather a spectrum that is spread-out a bit around the 400KHz. As the 400KHz is already well above the filter resonance frequency, I doubt you will notice any practical influence from this variation of frequency on the response of the filter resonance.
Now we are all awaiting The Doctor, ICG and voltwide among other.
Last edited:
I am no Doctor. But there's an important point. The thing you're flatout ignoring is the impedance rise on a lot of speakers. Many dome tweeters aren't a problem and AMTs or most ribbons laugh about that but the impedance of a horn tweeter isn't that flat. But it gets really bad if you look at the impedance plot of fullrange drivers. A lot of these got ~30 Ohm at 15 or 20k and that's by no means trivial anymore, we're talking about ~5dB at some amps here! So, you can in many cases disregard that, but for some folks it is actually very important. As always, it depends on what you're actually doing. I do not try to scare anyone off of class D amps, I try to rise awareness of the problem. No more, no less than that.
Hi
I checked the TPA3116 datasheet and compared it with the others e.g. TPA3251. i cannot find in the block diagramm the automatic by TPA3116 with the "regulation" of this osc. frequency.
the effect that the filter mismatch in higher frequencies was alwasy checked by my board mods - but i never tried to split the amps in low and high.
so if i think about what ICG wrote -and that trough for lot of spekaers- then the logical consequence is to split the amps in a first step in low and high = bi amping with an adaptation of the "tweeter" ouput filter - correct?
i try this at the filter designer (i use the setup which i learnt at my TPA3255 board and feeback by the specialists)
so for an impedance of t 20 ohm we need a coil of 50µH
chris
I checked the TPA3116 datasheet and compared it with the others e.g. TPA3251. i cannot find in the block diagramm the automatic by TPA3116 with the "regulation" of this osc. frequency.
the effect that the filter mismatch in higher frequencies was alwasy checked by my board mods - but i never tried to split the amps in low and high.
so if i think about what ICG wrote -and that trough for lot of spekaers- then the logical consequence is to split the amps in a first step in low and high = bi amping with an adaptation of the "tweeter" ouput filter - correct?
i try this at the filter designer (i use the setup which i learnt at my TPA3255 board and feeback by the specialists)
so for an impedance of t 20 ohm we need a coil of 50µH
chris
Last edited:
Hi Chris,
I understand ICG's reply such that the problem relates to the impedance variation of certain speaker drivers with frequency. Actually, it is not the carrier frequency changes of the TPA3116 that is the problem.
Apparently, some horn speakers (I assume conical treble horns) and in particular full range drivers may have highly variable impedance with frequency.
For a pure treble driver, we could in principle put an electronic high-pass filter before the class D amplifier and use an adapted output filter for the higher treble driver impedance. It is then a matter of how constant this treble driver impedance actually is.
The big problem seems to be with full-range drivers having a varying impedance becoming some 30 Ohm in the top. There the output filter is about right for most of the frequencies while at the top the filter is poorly damped due to the increase in impedance. For such full-range drivers, we cannot filter out a part of the frequency range and we cannot make a dynamically adaptable output filter with varying properties for different frequencies. At least it will be very complex.
The conclusion I draw from ICG's qualified analysis is that class D, due to the output filter, is less suited for full-range drivers because of their varying impedance. Impedance variations with frequency is much better handled by class AB amplifiers having a low output impedance in the full audio range.
50uH for 20 Ohm sounds about right.
Last edited:
Hi FF
Thanks for your feedback..
Yes the frequency response of speakers is normal - real - i know.
as a look and read a lot in this forum and i recognizes at stronger amps (apex, hypex, ICE moduls) big fat coils and if i remember right such values about 30µH or so. now it looks more logical for me because of the higher Z at high frequencies
so my conclusion is like the post before...big value for the "tweeter" but with higher current avalibitlity (low DC -bigger coil) to push some current at the bass chassis.
so do not understand me wrong - at all my modification (FX502spro, TPA3255 YJ) i keep still the recommendations and i am happy with this low cost Class D amps sound wise.
chris
Thanks for your feedback..
Yes the frequency response of speakers is normal - real - i know.
as a look and read a lot in this forum and i recognizes at stronger amps (apex, hypex, ICE moduls) big fat coils and if i remember right such values about 30µH or so. now it looks more logical for me because of the higher Z at high frequencies
so my conclusion is like the post before...big value for the "tweeter" but with higher current avalibitlity (low DC -bigger coil) to push some current at the bass chassis.
so do not understand me wrong - at all my modification (FX502spro, TPA3255 YJ) i keep still the recommendations and i am happy with this low cost Class D amps sound wise.
chris
Last edited:
Hi Chris,
As I understand you, it may be better to design the output filter for the speaker driver impedance at higher frequencies (above nominal driver impedance) and accept that the output filter for the mid-range and lower frequencies is over-damped. We cannot avoid that with a varying driver impedance, the damping of the filter is not ideal for all frequencies. But, over-damping at lower frequencies should be better than under-damping at higher frequencies because the lower frequency range is further away from the resonance frequency and thus much less affected by the resonance.
The "price" when using a higher induction value for the choke is the choke will need to be physically larger to handle the maximum currents for in particular the bass driver without saturation, just as you conclude.
Last edited:
The Vienna Theorem, draft version.
Dear Chris,
None of the big brains have objected to your findings until now. I propose that we baptize your principle "The Vienna Theorem" (or something involving your name directly?) for further reference.
It can be drafted as: In case you use a class D amplifier with a speaker having a significant impedance rise at higher audio frequencies (such as for many full-range drivers, ref.: Loudspeaker impedance measurement using a multimeter and 2 resistors , figure in section "Loudspeaker impedance measurement complete"), then design the amplifier output filter according to the impedance at high audio frequencies following the advises of TI in http://www.ti.com/tool/LCFILTER-CALC-TOOL .
This way the higher audio frequencies will be near critically damped while the lower audio frequencies will be over-damped but with little influence from the resonance above the audio range.
Make sure your output filter chokes will not saturate at any currents drawn at the output of the amplifier.
Dear Chris,
None of the big brains have objected to your findings until now. I propose that we baptize your principle "The Vienna Theorem" (or something involving your name directly?) for further reference.
It can be drafted as: In case you use a class D amplifier with a speaker having a significant impedance rise at higher audio frequencies (such as for many full-range drivers, ref.: Loudspeaker impedance measurement using a multimeter and 2 resistors , figure in section "Loudspeaker impedance measurement complete"), then design the amplifier output filter according to the impedance at high audio frequencies following the advises of TI in http://www.ti.com/tool/LCFILTER-CALC-TOOL .
This way the higher audio frequencies will be near critically damped while the lower audio frequencies will be over-damped but with little influence from the resonance above the audio range.
Make sure your output filter chokes will not saturate at any currents drawn at the output of the amplifier.
At the end of the day you will start to think about compensating the loudspeaker impedance on the side where it lives: the speaker.
Now look at what has been done to make tube amps live easy for centurys! The tube amp has an output transformer (inductor)... so it´s output is impedance sensitive.
It is easy to calculate and add some parts to the x-over and have a constant speakert impedance.
But to tell the truce, I do not belive that output coil matching with D-amps will have a real sonic impact that can be translated into "better sound".
Now look at what has been done to make tube amps live easy for centurys! The tube amp has an output transformer (inductor)... so it´s output is impedance sensitive.
It is easy to calculate and add some parts to the x-over and have a constant speakert impedance.
But to tell the truce, I do not belive that output coil matching with D-amps will have a real sonic impact that can be translated into "better sound".
Last edited:
Hi Turbowatch
...hmm...maybe i am not an expert....i do not have an equipment with thd..but i tried it at my YJTPA3255, FX502spro, tpa3116dual chip and my easy amp tda7498. if the coil is "set" to lower impedance and you play a high impedance speaker then the time when you get tired of listening music -this is at the level over the normal listening level -its a level i enjoy the room is full of music, but not loud.
this phenomena is for me the miss match of the coils in upper frequencies.
chris
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.