TAS2572EVM: TAS2572 boost fail for classH tracking mode

sorry,update the information:

We are unable to get Class-H mode working on TAS2572 (smartPA, 1S battery, VBAT=3.9V, L=1µH, 8Ω load).

Symptom:

• BST_CFG_02 (0x22) = 0x23 (Class-H) + any VBOOST_MAX_CTRL (tested 0x5B/6V and 0xA7/11V) → PVDD stays at VBAT level during 0dBFS playback
• BST_CFG_02 (0x22) = 0xA3 (Always ON) → PVDD correctly ramps to target voltage
• Firmware (tas2572RCA1.bin) loads successfully; audio playback works in Always ON mode

Our understanding from datasheet (Section 7.3.1.3.4): Class-H requires CLASSH_TUNING_x[23:0] registers to be configured via PPC3 software based on system-level parameters (VBOOST_MAX, C_OUT, gain, BST_ILIM, inductor value).

Questions:

1. Are the default reset values of CLASSH_TUNING registers sufficient for Class-H to function with default VBOOST_MAX=0xA7?
2. If not, can you provide the PPC3-generated CLASSH_TUNING values for our configuration: 1S, VBAT=3.9V, L=1µH, C_OUT=10µF, VBOOST_MAX=6V (0x5B), BST_ILIM=default, Gain=21dBV, 8Ω load?
3. Alternatively, can you share PPC3 software access so we can generate these parameters ourselves?

Hardware Parameters (for PPC3 Calculation)

## Hardware Parameters (for PPC3 Calculation)

| Parameter | Value |
|-----------|-------|
| Boost Inductor (L16) | 1µH, 0805 |
| PVDD Output Capacitor | 20µF total (C210 + C211 = 2× 10µF, 0603, X5R, 25V) |
| Decoupling Capacitor (C209) | 100nF, 0201, X5R, 25V |
| VBOOST_MAX target | 6V (0x3B = 0x5B) |
| BST_MODE | Class-H (0x22 = 0x23) |
| Speaker Load | 8Ω |
| Max Output Power | 1.2W |
| VBAT Range | [Please confirm: 1S Li-ion 3.4V–4.4V] |

## Request

1. Could you please provide the complete **PPC3-generated register values** for our configuration, specifically:
- CLASSH_TUNING_01 ~ CLASSH_TUNING_15
- BOOST_TUNING_01 ~ BOOST_TUNING_11
- BST_ILIM recommendation for 1µH / 20µF / 6V configuration

Hi WangQin,

the TAS2572's boost can work properly under the condition of VBAT=3.9V. 

but i seems your device didn't work as expected. can you share your configuration? do you generated it from PPC3 or only get it somewhere?

we can check it whether there is any register issues frist.

Br,

Wenbin

Hi Wenbin,

Thanks for the quick response.

To answer your question: No, we did NOT use PPC3 to generate the configuration（we haven't this tool）. We only modified the following two registers manually based on the datasheet:

Our Current Register Configuration

Hi Wenbin,

Thanks for the quick response.

To answer your question: **No, we did NOT use PPC3 to generate the configuration.** We only modified the following two registers manually based on the datasheet:

## Our Current Register Configuration

| Register | Address | Value | Purpose |
|----------|---------|-------|---------|
| BST_CFG_02 | 0x22 | 0x23 (Class-H) / 0xA3 (Always ON) | Boost mode selection |
| BST_CFG_05 | 0x3B | 0x5B (6V target) | VBOOST_MAX setting |

All other registers remain at **default (reset) values**, including:
- CLASSH_TUNING_01 ~ 15 → all defaults
- BOOST_TUNING_01 ~ 11 → all defaults
- BST_ILIM → default

## Behavior Observed

| Mode | Register 0x22 | Result |
|------|---------------|--------|
| Always ON | 0xA3 |  Works — PVDD reaches 6V |
| Class-H | 0x23 |  Fails — PVDD stays at VBAT level |
| Class-H (11V default) | 0x23 + 0x3B=0xA7 |  Also fails — PVDD stays at VBAT |

## Hardware Parameters

- **VBAT:** 3.9V (1S Li-ion)
- **Boost Inductor (L16):** 1µH, 0805
- **PVDD Output Capacitor:** 20µF total (C210 + C211 = 2× 10µF, 0603, X5R, 25V)
- **Decoupling Capacitor (C209):** 100nF, 0201, X5R, 25V
- **Speaker:** 8Ω, 1.2W max

## Our Questions

1. We understand from the datasheet that CLASSH_TUNING_xx registers need to be recalculated when VBOOST_MAX is changed. Could you provide the **PPC3-generated register dump** for our hardware configuration (L=1µH, C_PVDD=20µF, VBOOST_MAX=6V, VBAT=3.9V)?

2. However, we also tested with **default 0x3B = 0xA7 (11V)** without changing any other registers, and Class-H still didn't work. In this case, all tuning registers should still be valid at their reset values — so why does Class-H still fail?

3. Is there any **initialization sequence requirement** or additional enable bit we might be missing for Class-H mode?

Please let us know if you need a full register dump from our device for further analysis.

Hi Wangqin,

usually, you should get the driver code from TI, and it will include a default init bin file. 

I don't know why you are using this chip like this ( just set several resisters without the init file).

if you need, I can involve TI SW engineer to provide the driver code.

Br,

Wenbin

Yes, please involve the TI SW engineer to provide us with the driver code and the proper init bin file. This would be very helpful.

For the SW engineer's reference, here is our hardware configuration:

Hi Wenbin,

Thanks for the clarification. That makes sense.

Currently, we are using the **Bin File Generator (v1.4.2)** to create the init bin, and only configured a limited set of registers manually based on the datasheet. We do not have the complete TI driver code or the default init bin file that includes the full Class-H tuning parameters.

**Yes, please involve the TI SW engineer to provide us with the driver code and the proper init bin file.** This would be very helpful.

For the SW engineer's reference, here is our hardware configuration:

| Parameter | Value |
|-----------|-------|
| Platform | MTK MT8169 (Linux) |
| Boost Inductor (L16) | 1µH, 0805 |
| PVDD Output Capacitor | 20µF (2× 10µF, 0603, X5R, 25V) |
| VBOOST_MAX target | 6V |
| VBAT | 3.9V (1S Li-ion) |
| Speaker | 8Ω, 1.2W max |
| Audio Interface | I2S / TDM |
| Sample Rate | 48kHz |
| Control Interface | I2C |

We need to resolve this as soon as possible. Please let me know if there's anything else needed from our side. Thanks!

Best regards,
Wanqin

i upload our bin from your provided config generations,and then we changed the values just of  classh didn't working! base version from your provided.

tas2572-json-bin.ziptas2572-json-bin-base.zip

Hi

I have involved out SW expert to this thread

Br,

Wenbin

yeah，thanks

This issue is quite urgent, so please help us look into it as soon as possible. Much appreciated!

Linux driver link (https://tidrive.ext.ti.com/u/-8JXXq5CVyHpwldS/44a3f155-a8d4-46c0-9cb0-73aa25487dc3?l)

access code: C/44p3jq

why did you send the amp drv code?we had already portting the codec drv code from your provided.we tested base the code.and classh traking is no effective.

Regarding the current status: TI @Wu, Zerunhas already analyzed the register dump we provided and identified that our gain was set to 21dB (0x07=0x00). Since 21dB gain corresponds to a 15.8V peak output voltage, which exceeds any Boost Vmax setting (even the maximum 13V), this causes serious clipping that negatively impacts the Class-H algorithm behavior.

we changed the volume gain,and also fail to playback the 0dBfs signal,the pvdd is not ramp up .equals VBAT.

Let me transfer Peter

Hi Wangqin，

first, the higher gain can result in clipping but will not impact the boost function. but if the clipping results in OC or other issue, the chipset stop work, the boost of course will stop. to verify this, you can just leave the load floating and test the boost voltage only.

regarding the new driver code, that's because you mentioned that you didn't have any init processing for this chipset. this is not the normal processing flow in any project previous.

it seems there is more unclear things, can you get the FAE involved and organize a meeting?

thanks,

Br,

Wenbin

ze run had already checked our waveforms which tested from classd out p/n ,vpeak and out power. it's outloaded speaker,just AP Audio Analyzer.

another,why we changed the always on mode,it works fine?

see these waveforms:0x07=0x0A, 0x3B=0xA7

you said it's related about init processing for this chipset?how do you say that?i didn't get all why did you say that?

our codec drv and chipset bin,all of it provided by TI.TI sent us the config,and generated the bin for flashing.

Hi Wanqin, 

I checked the output waveform and it seems that the output amplitude is so small(only ~350mV peak), hence it is normal that the class H don't need to boost to higher voltage to maintain the required output power. Besides, I checked the test signal and found that it is not actually 0dBFS 1kHz sine wave, but only with amplitude of -24dB RMS(-21dB peak). Did the test engineer verify the test signal amplitude before performing the output power test? Thanks.   

Regards,

Peter

let me check the source of audio signal.

We will update you with the corrected test results.can you help this signal,i generated -3dB,

  the 7v is right?

With proper 0dBFS input, PVDD ramps up as expected and the amplifier delivers full output power.

Could you please confirm:

1. Are these output levels consistent with what you would expect for these configurations?

2. For our final application (speaker max 1.2W @ 8Ω, requiring Vrms = 3.10V), what gain setting (0x07) and VBOOST_MAX (0x3B) combination would you recommend?

Thanks, Wanqin

Thank you for the detailed explanation. We understand the recommendation now — use higher gain + higher PVDD with IV Sense protection and PPC3 limiter/DRC to control output power, rather than limiting PVDD itself.

Two follow-up questions:

1. Class-H mode issue: We found that Class-H (BST_MODE=0x00, 0x22=0x23) does not ramp up PVDD during playback — it stays at VBAT level. Only Always ON mode (BST_MODE=0x10, 0x22=0xA3) works. We tested with both 0x3B=0xA7 and 0x3B=0x5B. Does Class-H require specific CLASSH_TUNING_xx register values from PPC3 to function? Could you provide the PPC3 tuning parameters for our system (1S, VBAT=3.9V, L=1.5µH, 8Ω)?

2. PPC3 access: Could you provide PPC3 software access so we can generate the tuning parameters (CLASSH_TUNING, Limiter, DRC, etc.) for our configuration?

Our system parameters:

• Battery: 1S Li-ion, VBAT = 3.0V ~ 4.2V
• Inductor: 1.5µH (L16)
• Load: 8Ω, speaker rated 0.8W / max 1.2W
• Preferred config: 0x07=0x08, 0x3B=0xC5 (per your recommendation)

Thanks, Wanqin

yes,thanks, can you send the pwm parameters for desense?