I want to tune my output filter so that the higher frequencies sounds right.
The speaker manufacturer hasn't published frequency/ohm specs either. I've tried some 15uH and 1uF caps / 22uH and 0.680uF but nothing sounds just right. Is there anyway to tune it by how it sounds? For example more/less timbre or something like that?
I am modding tpa3116 boards which sounds good, but the upper treble isn't as lifelike as I know it could be.. If you have any experience on this matter please share it!
The speaker manufacturer hasn't published frequency/ohm specs either. I've tried some 15uH and 1uF caps / 22uH and 0.680uF but nothing sounds just right. Is there anyway to tune it by how it sounds? For example more/less timbre or something like that?
I am modding tpa3116 boards which sounds good, but the upper treble isn't as lifelike as I know it could be.. If you have any experience on this matter please share it!
Have you tried ferrite beads!?, If you are not using long length cables, could be the easiest and cheaper solution.
The output filter is a compromise between being fast enough to allow audio through but also being low enough to stop carrier signal through.
The higher the carrier frequency the easier it is to get good top end on the audio without allowing carrier to get through.
The higher the carrier frequency the easier it is to get good top end on the audio without allowing carrier to get through.
Hi
In my experience with TPA3123 board, the nature of the high frequency is greatly influenced by placing low value eg 102/ 101 ceramic caps as close to the power pins of the TPA chip as possible.
My set up is TDA1543 dac, keantokean buffer and markaudio12p.
Most of my friends comment very positively of the natural high freq reproduction..
cheers
kp93300
In my experience with TPA3123 board, the nature of the high frequency is greatly influenced by placing low value eg 102/ 101 ceramic caps as close to the power pins of the TPA chip as possible.
My set up is TDA1543 dac, keantokean buffer and markaudio12p.
Most of my friends comment very positively of the natural high freq reproduction..
cheers
kp93300
I have come to the same conclusions too but I wan't to try to custom tune the output LC filter also.. I use long cables so ferrite beads are out of the question.. I use aluminium electrolytics as power caps which remove most of the noise and they improve the sound very much. But I think the upper treble could be even better!
If you've got a true RMS detecting DMM, oscilloscope or other means of measuring the voltage at the amp output, drive the amp with a frequency generator and measure the amp's frequency response with the load connected.
If it's peaky, rolls off too quick, etc... then that gives you some idea of what you have to change.
If it's peaky, rolls off too quick, etc... then that gives you some idea of what you have to change.
Thats one way to do it.. But the whole purpose is to make the amp sound good.. If the output filter doesn't impact hearable differences, what's the point of tuning it??
My take on this, based on some LTSpice simulations........ Choose 2 order low pass with butterworth response at 80KHz or so at the rated impedance. But recall that speakers are anything but resistive and they do vary vs. frequency. Lower impedance than the design impedance causes the output to roll off sooner. A filter designed for 8 ohms will roll off some 20KHz (or less) if being driven into 4 ohms or less. All of engineering is a compromise.
Another goal of Class D filters is to reduce EMI/RFI from all that switching garbage as well.
By ear??? No way.
Another goal of Class D filters is to reduce EMI/RFI from all that switching garbage as well.
By ear??? No way.
Yes, very easy to sim in SPICE(I use TINA, the free spice sim program from TI) Here is LC filter and typical reactive driver as represented by lumped TS elements. This is for TPa3116d2 with 10uH and 680nF caps and snubber circuit. The gain and phase with reactive load is actually better than pure resistive load.
http://www.diyaudio.com/forums/multi-way/258487-why-crossover-1-4khz-range-76.html#post4027841
Thanks stoc005 for that useful information! So basically the tuning of the filter should not affect the instruments "timbre" of any kind.. Just the upper treble spl level?
One other tip. If you are designing a filter for a half bridge output, use the nominal impedance of the speaker. But if you are designing a filter for a full bridge output (BTL, 4 fets, both speaker wires above ground) use 1/2 the nominal impedance of the speaker. Why? the 2 inductors with 2 or 3 cap filters only see 1/2 the impedance per half bridge output. In essence there are two filters each driving 1/2 the load.
The TI data is somewhat dated since they think you might be able to go filterless. They talk about a 25KHz cutoff. Nope. You will have some early rolloff in the highs if you choose to go that low. I choose 80Khz since I see under 1 db rolloff at 20KHz with resistive loads. The goal is a full range filter with no rolloff of highs.
The TI data is somewhat dated since they think you might be able to go filterless. They talk about a 25KHz cutoff. Nope. You will have some early rolloff in the highs if you choose to go that low. I choose 80Khz since I see under 1 db rolloff at 20KHz with resistive loads. The goal is a full range filter with no rolloff of highs.
- Status
- Not open for further replies.