LM4675SDBD: Audio Power Amplifier Output validation using CRO

Hi Jay,

You're correct in terms of the expected output power for the given conditions.

I have a few comments and questions listed below:

• What is the factor that makes the output not satisfactory? Is the audio output distorted? Is it not loud enough?
• If you want to check if the limitation comes from the battery, you can try using a bench supply instead of the battery, this way you can control the amount of current going to the amplifier.
• Regarding the filter for the scope measurements, you may use any RC or LC filter with cutoff frequency around 24k~40k. Lab measurements use AUX-0025, but you can use other filters like this RC-box.
• Once you have the filtered signals on the scope, you can first measure the peak voltage, then calculate the RMS and from there apply the power formula to obtain the RMS power, which should match the data sheet specifications.

Best regards,
-Ivan Salazar
Applications Engineer

Hi Ivan,

Thanks for your detailed explanation.

1) What is the factor that makes the output not satisfactory? Is the audio output distorted? Is it not loud enough?

- Both. Loudness and sound quality. We are also working to improve mechanical enclosure design for better sound

2) If you want to check if the limitation comes from the battery, you can try using a bench supply instead of the battery, this way you can control the amount of current going to the amplifier

- Okay. I will test it and get back to you. meanwhile can you please let me know is LM4675SD suitable for LiPo battery powered products ? or is it made for external DC powered products ? If it is not made for battery powered products can you please suggest any alternative part which can given upto 1 to1.5W output power at 3.7V battery voltage and made for low battery powered products ?

3) Regarding the filter for the scope measurements, you may use any RC or LC filter with cutoff frequency around 24k~40k. Lab measurements use AUX-0025, but you can use other filters like this RC-box.
- Thanks you. RC-Box application note will help me a lot.

4) Once you have the filtered signals on the scope, you can first measure the peak voltage, then calculate the RMS and from there apply the power formula to obtain the RMS power, which should match the data sheet specifications.

- Okay, thank you. 

Hi Jay,

The amplifier is able to operate either or line or battery powered applications. It's a good idea to look into the speaker design as well, my suggestion to check with bench supply is just to double check if there are current limitations in this case, if that could be a problem then using a regulator, adding more decoupling capacitance or using other devices with features such as AGC (TPA2015D1) could be a next step in that case.

Let us know if any further support is required after your tests and analysis.

Best regards,
-Ivan Salazar
Applications Engineer

Hi Ivan,

I had tested with external power supply, result is almost same. Regarding speaker we are using AS01508AO-SC-WP-R from PUI. It is made for low power product. I am working to get external RC filter to measure output power of the amplifier. I will update you tomorrow with results.

As well as at 3.7V, the LM4675SD can give 0.5–0.6W as output, Is it enough to drive speaker (8R, 0.7W input power (1W Max), and 94 dBA SPL specifications) ?

As we have one similar product with the speaker which has same specification and sound loudness is really good. It is almost double. I don't have idea about which amplifier they had used.

Here is block diagram of audio output path.

15uH inductor in series and 1uF cap in parallel of audio out path.

HI Jay,

LM4675 is a Class-D amplifier, so the common-mode capacitor filter is not commonly used. Could you first try without the filter and connect the speaker directly at the OUTP and OUTM from the amplifier? If that's better, you can try replacing the 1uF common-mode capacitor and use 2 capacitors to GND instead.

Best regards,
-Ivan Salazar
Applications Engineer

Hi Ivan,

1) Yes there is bit improvement (not that much only 5%) in the loudness after removing 1uF.

2) I had played continuous 1KHz sine wave tone instead of audio clip and performance is too good. Loudness increased by 80%. so it looks like audio path is performing well in lower frequencies ( in this test case 1uF also removed)

3) I had done voltage measurement at amplifier output with 8.2 ohm resistive load( removed speaker) as well as removed battery and given constant external 3.7V supply. I haven't change 15uH series inductor and added 0.22uF w.r.t GND shown in circuit. As well as I had added 301R between amplifier output as shown in page 15 of the datasheet.

4) Scope results are attached. Probe 1(Yellow) is connected to OUTP and Probe 2(Green) is connected to OUTN. GND of probe is connected to board GND. Then by using math function (Probe 1-Probe2) results are shown in orange trace. Vrms of (Probe1-Probe2) is 1.8V and I had measure Vrms using Multimeter and I got same result 1.8V.

5) By using P=E^2/R, P= (1.8*1.8)/8 = 0.4 W only or do we need to multiplies 1.8V with 1.414 as it is RMS value than, 1.8V*1.414= 2.54V, P=E^2/R, P= (2.54*2.54)/8 = 0.8W. Can you please let me know which one is correct method and Can you please verify output signals and schematic ? As per datasheet it should give 0.5W min.

Hi Jay,

Can you try without the 301R? This shouldn't be required if you have a speaker or 8-ohm load connected in parallel. It seems from the scope capture that the output is clipping at ~2V, just for debugging purposes you may also try without any load (neither 301R nor 8-ohm load) and check if the output signal is able to reach higher voltage.

Best regards,
-Ivan Salazar
Applications Engineer

Hi Ivan,

Okay I will take results without 301R and without load. Meanwhile I am sharing another test result which I had captured by making filter same as RC Box. I had taken combination of 68 ohm and 68nF to make low pass filter and cut off frequency is around 34KHz. I had removed all other components from audio path and directly connected 8 ohm load and after 8 ohm load I had connected RC filter as suggested in RC box application note. 

 Scope results are attached. Probe 1(Yellow) is connected to OUTP and Probe 2(Green) is connected to OUTN. GND of probe is connected to board GND. Then by using math function (Probe 1-Probe2) results are shown in orange trace. Even in this case I am getting clipped output but Vrms value is 2.8V.

Can you please let me know which one is correct method in my previous post ? "P=E^2/R, P= (1.8*1.8)/8 = 0.4 W only or do we need to multiplies 1.8V with 1.414 as it is RMS value than, 1.8V*1.414= 2.54V, P=E^2/R, P= (2.54*2.54)/8 = 0.8W. Can you please let me know which one is correct method?"

Hi Ivan,

In above test condition I had removed 8R load and directly connected RC filter to measure audio data. there is no load and no component in audio path. Only RC filter between amplifier output and scope.

Scope results are attached. Probe 1(Yellow) is connected to OUTP and Probe 2(Green) is connected to OUTN. GND of probe is connected to board GND. Then by using math function (Probe 1-Probe2) results are shown in orange trace. 

In this case I am getting clipped output but Vrms value is 3.1V (improved compare to previous 2.8V with load)

Hi Jay,

Thanks for the updates.
I think the latest captures are as expected. You may use the Vrms value of the math result of (OUTP)-(OUTP), calculating power using this value should give Wrms which is the specified value in data sheet. If you multiply by sqrt(2), then you would be considering Vpk and Wpk instead.

Regarding the clipping, the gain of the amplifier is set to 2V/V (given by 300k/150k), so it will output double of the signal you set at the input, you may reduce the input signal so that the output is just before clipping.
THD+N = 1% is output slightly clipping, while THD+N = 10% is already clipping considerably.

Best regards,
-Ivan Salazar
Applications Engineer

Hi Ivan,

Is it possible to do further communication over email ? I want to share schematic of audio section for further discussion.

Hi Jay,

I just accepted your friend request, and you can now send private messages. You may share your files over there or we can share contact information there as well to follow up offline.

Best regards,
-Ivan Salazar
Applications Engineer