0
\$\begingroup\$

I have a 2 x 50 W TPA3116D2 (VHM-302) amplifier and a Pioneer TS-G1345R 35 W, 4 Ω speaker. When connecting them, there is no problem and the sound is good.

However, I wanted to build a low-pass filter to hear only the bass sound. For that, I made a Sallen-Key filter using a TDA2030. I selected R1 = R2 = 100 Ω and C1 = C2 = 4.7 μF so that the cut-off frequency is around 300 Hz.

After adding this filter to the circuit, both the volume and the bass level decreased a lot. I can barely hear the bass sound now. I'm sharing my circuit below. What could be the problem?

enter image description here

\$\endgroup\$
1
  • \$\begingroup\$ You need a split power supply - that means: Both, the pos. and the neg. power rail need a voltage against the common ground. In your drawing there is only one single DC supply BETWEEN both supply pins without any ground reference. \$\endgroup\$
    – LvW
    Commented Feb 9, 2023 at 16:08

5 Answers 5

1
\$\begingroup\$

The TPA3116 is a class-D amplifier, which is using high-frequency PWM for driving speakers. The PWM frequency is hundreds of kHz (I can't remember the exact number, please check its datasheet).

Thus, the output from TPA3116 is, essentially, a pulse-width-modulated audio signal at high frequency. You're using a low-pass filter, which results in the PWM signal not passing to the speaker.

\$\endgroup\$
1
  • \$\begingroup\$ Class-D switching amplifiers drive speakers fine with low audio frequencies modulated with PWM. it is your filtering power amplifier that is not working properly because its gain is much too low to be a power amp. Make a filter circuit with an opamp and it can have its output attenuated so that it can drive the TDA2030 that has a gain of 20 times or more. With a 12V supply, the output of the TDA2030 will be about only 3W into 4 ohms. \$\endgroup\$
    – Audioguru
    Commented Feb 8, 2023 at 1:45
0
\$\begingroup\$

Like most linear power amplifiers, the obsolete TDA2030 power amplifier MUST have a gain of 20 or more to avoid oscillating. Its gain is only 1 in your filter circuit.

With a 12V supply, a TPA3116 amplifier produces 15W into 4 ohms when both speaker wires are driven. One speaker wire produces only about 4W into 4 ohms. I think a TPA3116 class-D amplifier must have a load of minimum 8 ohms between its two speaker wires so that its output filters are not a dead short at its switching frequency.

\$\endgroup\$
3
  • \$\begingroup\$ Wouldn't increasing the gain higher than 3 cause the filter circuit to become unstable and oscillate since the Q value will also increase? Also, I don't use the right speaker output of tpa3116, I use only the left speaker output. \$\endgroup\$
    – Ozkan lV
    Commented Feb 8, 2023 at 11:52
  • \$\begingroup\$ Yes - you are right. The closed-loop gain must not be larger than "3". This gain exclusively determines the pole-Q of the filter. \$\endgroup\$
    – LvW
    Commented Feb 8, 2023 at 19:30
  • \$\begingroup\$ Of course a properly designed "equal components" filter has a certain low gain of 1 (Bessel) or a gain of 1.6 (Butterworth) and the minimum gain of the TDA2030 power amplifier must be 20 times. They are completely separate parts. \$\endgroup\$
    – Audioguru
    Commented Feb 8, 2023 at 22:00
0
\$\begingroup\$

Some other recommendations in datasheet applications have not been followed, in addition to gain.

Ground is a floating voltage in your diagram, a single supply should used a half supply voltage reference for the amplifier (RA + RB + CB), and input decoupling capacitor (C1) and output capacitor (2000 uF).

Decoupling capacitors for supply are missing (C5 + C3).

Althoug a low pass filter is used, a zobel network is recommended (R4 + C7), also protection diodes D1 and D2. enter image description here

\$\endgroup\$
0
\$\begingroup\$

Your Sallen-Key lowpass has two equal capacitors and two equal resistors. Hence, it is a so called "equal component design".

In this case, the pole frquency is at wp=1/RC and the quality factor (pole-Q) is Qp=1/(3-Acl).

When the gain of the fixed-gain stage is unity (closed loop gain Acl=1) the filter response is very bad (Qp=1/2). Remember that even a simple passive 2nd-order RC filter will have a quality factor which can be very close to Qp=0.5.

Therefore: Either use the unity-gain design (not the same passive parts) or use higher closed-loop gain Acl (maximum Acl=3) and equal componenets. But you should not realize a mixture of both design alternatives.

\$\endgroup\$
0
\$\begingroup\$

Here is the peaked response that is almost oscillating of a Sallen-Key lowpass filter with equal-component values and a gain of almost 3 times: SK

\$\endgroup\$
2
  • \$\begingroup\$ Your design will have 0 dB gain at very low frequencies (and 7.8 dB gain above app. 1 Hz, as shown in your Bode diagram). This is due to the capacitor C3. Why do you recommend this cap ? \$\endgroup\$
    – LvW
    Commented Feb 9, 2023 at 15:58
  • \$\begingroup\$ I like deep bass. With the gain resistors as 620 and 1k for a gain of 1.62 and a Butterworth response, the 10uF feeding the 1k capacitor produced -3dB at 8Hz for a flat output down to 40Hz. Then I increased the resistor values for an opamp to drive them but then there was too much capacitance. \$\endgroup\$
    – Audioguru
    Commented Feb 10, 2023 at 18:15

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.