[FAQ] AM64x Design Recommendations / Commonly Observed Errors during Custom board hardware design – EVM or SK Schematics updates for Design Update Note

Hi Board designers, 

Here are some recommended updates based on the learnings for performance improvements.

Schematics - Design Value Updates 

1. VCC_3V3_SYS LOAD SWITCH

SoC IO supply rails have slew rate requirements specified.

Refer Power Supply Slew Rate Requirement section of the data sheet.

Add a cap 220 pF or higher on the Load switch CT (Switch slew rate control) pin.

2. SD card power reset Load switch 

Add a cap 220 pF or higher on the Load switch CT (Switch slew rate control) pin.

3. eMMC supply decap

Add decaps as required to the eMMC memory supply rails 

4. USB VBUS divider 

5. Provision to Bypass CMC

6. USB power switch 

Use power switch with OC indication. Connect the OC output of the power switch to SoC input

7. SoC clock MCU_OSC0_XI. MCU_OSC0_XO - connecting external clock 

DNI the R and C placed at the output of the clock buffer connected to the MCU OSC0 clock input. Placement of these components reduces the amplitude and could affect the performance

8. SoC  MCU_OSC0_XO termination - when using external clock 

Data sheet section reference : MCU_OSC0 LVCMOS Digital Clock Source

Do not leave XO un connected 

The oscillator was designed to measure the differential voltage between XI and XO, for creating a internal reference clock.  So there is no guarantee they will get the correct differential input voltage if the XO pin is allowed to float.  The datasheet clearly shows this pin needs to be connected to VSS when using an LVCMOS clock source.

9. Ethernet PHY updates

10. Caution with EPHY 1V and VPP 1.8V LDO

Miniature DQN package used.

Package Outline

PCB pads 

There is a likely chance of assembly error due to the LOD outline, pins orientation and  the land pattern (pads and pitch)

Take note of the following

EPHY 1.8V supply pins when not used should not be shorted

Note the recommended caps for the EPHY supplies

Note the Rbias R and C values 

Note the pulldown for the EPHY Reset

11 EXTINTn RC and pullup 

EXTINTn is a open drain output type IO buffer. The IO has slew rate limit requirements when pulled to 3.3V/

Add an RC to limit the slew rate.

Regards,

Sreenivasa

Hi Board designers, 

Refer below workarounds when load switch is not used for resetting the SD card power supply and the supply is fixed to 3.3V

https://software-dl.ti.com/processor-sdk-linux/esd/AM62X/09_02_01_09/exports/docs/linux/Foundational_Components/Kernel/Kernel_Drivers/Storage/MMC-SD.html?highlight=mmc#steps-for-working-around-sd-card-issues

https://software-dl.ti.com/processor-sdk-linux/esd/AM64X/latest/exports/docs/linux/Foundational_Components/Kernel/Kernel_Drivers/Storage/MMC-SD.html

Given below are the list of various workarounds that can be done in the device tree node to get SD card working. The workarounds are ordered in increasing order of reducing performance.

All the changes mentioned below, are to be done in the MMCSD device tree node corresponding to the SD instance. This is usually the first (index starting from zero) instance.

1. Restricting to a given speed mode
   • By default the kernel driver tries to enumerate a SD card in the highest supported speed mode. Given below is the order in which the driver tries to enumerate a SD card
     • SDR104
     • DDR50
     • SDR50
     • SD HS
     • SD legacy
   • These speed capabilites can be masked using device tree property sdhci-caps-mask.
     • Limit to DDR50: sdhci-caps-mask= <0x00000002 0x00000000>
     • Limit to SDR50: sdhci-caps-mask= <0x00000006 0x00000000>
     • Limit to SD HS: sdhci-caps-mask= <0x00000007 0x00000000>
     • Limit to SD legacy: sdhci-caps-mask= <0x00000007 0x00200000>

·   &sdhci1 {·       /* SD/MMC */·      vmmc-supply = <&vdd_mmc1>;·      vqmmc-supply = <&vdd_sd_dv>;·      pinctrl-names = "default";·      pinctrl-0 = <&main_mmc1_pins_default>;·      ti,driver-strength-ohm = <50>;·      disable-wp;·      sdhci-caps-mask = <0x00000006 0x00000000>; /* Limiting to SDR50 speed mode */·   };

• Limiting to SD HS speed mode can also be done by using the property no-1-8-v. This disable switching to 1.8V which is required for UHS speed modes(SDR104, DDR50, SDR50)

·   &sdhci1 {·       /* SD/MMC */·       vmmc-supply = <&vdd_mmc1>;·       vqmmc-supply = <&vdd_sd_dv>;·       pinctrl-names = "default";·       pinctrl-0 = <&main_mmc1_pins_default>;·       ti,driver-strength-ohm = <50>;·       disable-wp;·       no-1-8-v; /* disabling all the UHS modes */};

Regards,

Sreenivasa

Hi Board designers, 

MMC0 PCB Connectivity Requirements

we have issues with an eMMC connected to MMC0 of a AM6254 processor on a custom board.

Hi Board designers, 

To implement UHS-I refer below steps:

The SD card interface pullups have to be connected to the 3.3V_1.8V VDDSHV_SDIO switched supply output from the PMIC 

The SD card needs to be powered using a fixed 3.3V supply.

The 3.3V supply to the SD card needs to be switched through a power switch

Provision to reset the power switch using SOC IOs is recommended 

An ANDing logic is recommend to reset the SD cars power switch. 

The SOC VDDSHV5 supply needs to be connected to the .3V_1.8V VDDSHV_SDIO switched supply output from the PMIC

Regards,

Sreenivasa

Hi Board designers, 

Refer below for description about SPI signal series resistor

Hi Board designers, 

Refer below for description about SPI signal series resistor use case (for understanding and can be applied for AM6x)

https://e2e.ti.com/support/processors-group/processors/f/processors-forum/1127775/am3358-a-ringing-problem-of-emmc-interface/4191109?tisearch=e2e-sitesearch&keymatch=userdisplayname%2525253A%25252522Peaves%25252522%25252520%25252526%25252526%25252520series%25252520resistor#4191109

AM3358: A ringing problem of eMMC interface

I do not think the voltage change observed on the clock signal while it is high is causing your problem. It appears you are probing the clock signal near the eMMC device based on the shape of the clock signal rise/fall edges. You need to probe the clock signal near the processor. I suspect you will see a non-monotonic transition near mid-supply on the source end of the PCB trace. This non-monotonic transition is likely causing the problem.

The processor sources the clock out to the eMMC device and loops in back at the processor pin to improve timing margin. So there is a good chance the non-monotonic transition occurring at the source is creating a glitch on the clock that is looped back into the processor.

The AM335x schematic checklist tells you to install a 33 Ω series resistor on MMCx_CLK (as close to the processor as possible). I do not see this resistor in your schematic. This resistor is necessary to prevent this problem.

The capacitor is likely changing the signal distortion enough to make it work better, but it is not a good solution.

The resistor doesn't eliminate the non-monotonic transition. It shifts the voltage away from the input buffer switching threshold voltage, where it is less likely to create a glitch on the looped back clock going back into the processor. The only way to resolve the clock glitch issue is install a 33 ohm series termination resistor located very close to the AM335x pin. The trace length between the processor pin and the series resistor should be as short as possible. 

AM3354: MMC clock timing

https://e2e.ti.com/support/processors-group/processors/f/processors-forum/1037185/am3354-mmc-clock-timing/3846519?tisearch=e2e-sitesearch&keymatch=userdisplayname%253A%2522Peaves%2522%2520%2526%2526%2520series%2520resistor#3846519

I did not see any de-coupling capacitors between on the AM335x VDDSHV4 power pin after low pass filter L704. Were these accidently omitted in the snapshots provided? If not, each power pin needs local de-coupling capacitors after their respective low pass filters.

I’m concern with the SD Card ground connections. Hopefully the SD Card VSS1 and VSS2 pins are connected directly to the same ground as the AM335x device. I assume the other grounds shown for the SD Card are simply the metal chassis ground and they are not part of the SD Card signal and power ground connections.

They should not be using a zero ohm resistor for R514. This should be a 22 - 50 ohm resistor. The MMC/SD host controller in AM335x loops the clock back at the pin to improve timing margin for the internal circuits. However, there is a negative side effect of implementing clock loopback on a pin. The clock signal will be distorted on the source end of a signal trace and a series resistor of 22 - 50 ohms is required to modify the signal distortion such that is does not create a glitch on the looped back clock signal.

I will explain what happens. The clock output buffer inside AM335x has a source impedance of about 30- 50 ohms and a typical PCB signal trace has a characteristic impedance in the range of 40 - 60 ohms. When the output buffer toggles the signal from low to high or high to low, the voltage of the pin which sits between the output buffer and the PCB signal trace will not transition to the expected voltage immediately. This is because the output buffer impedance and trace impedance create voltage divider that causes the pin to step to a mid-supply voltage for a short time. This step will last as long as it takes for the clock signal to travel down the PCB signal trace to the SD Card, where it encounters a high impedance load that creates a reflection that allow the voltage  to increase to VDD or VSS. The amplitude of the mid-supply voltage step may be near the AM335x clock buffer input switching threshold, which allows noise to create a glitch on the AM335x internal clock. The series resistor raises the impedance of the PCB portion of the voltage divider such that the voltage at the pin steps above or below the switching threshold.

I suspect this is what is causing the instability and the reduced clock amplitude is contributing to the issue.

What you observe on this signal depends on where you connected the probe, the bandwidth of your probe/scope, and quality of your scope probe ground. 

You will need a very high bandwidth probe and scope to observe short over-shoot, under-shoot, and non-monotonic events on the signal. I recommend using a low capacitance FET probe with a very-very short low loop inductance ground.

You will see a mid-supply step on the signal when probing near the source. This occurs because the output impedance of the MMC0_CLK output buffer, series termination resistor, and characteristic impedance of the PCB signal trace creates a voltage divider. The voltage divider output is applied to the source end of the PCB signal trace. This voltage propagates down the trace to the far end where it encounters a high impedance load which causes a in-phase reflection that returns to the source. Therefore, the voltage transition observed on the far end will be a continuous transition between VSS and VDD. The is not the case for the source end, as it steps to a mid-supply determined by the voltage divider values and the transition only continues to VDD or VSS once the reflection returns from the far end. You need to select a series resistor value that allows the MMC0_CLK pin to step through the voltage of (VDDSHV4 / 2) without pause. I suggest the step observed on the MMC0_CLK pin should be at least 200mv above (VDDSHV4 / 2) on the rising edge and at least 200mv below (VDDSHV4 / 2) on the falling edge.

Regards,

Sreenivasa

Hi Board designers, 

Refer below for description about SPI

SPI_MISO_ voltage was dropped from 3.3V to 2V
Query
Customer connected ADC data output (MISO) to AM2432 through ISO7741 using SPI interface. The expected voltage level of SPI_MISO_ADC was 3.3V, however it dropped down to 2V.
Answer
I can only assume their software has configured pin A14 on the AM243x device as an output and it is in contention with the output of the ISO7741 device.
Have customer check the voltage of this signal while the AM243x is held in reset?
Have customer disable the AM243x output buffer by writing the appropriate value to the TXDIS bit in the respective PADCONFIG register during this condition to see if the signal voltage changes.
Result
Customer disabled TX_DIS then voltage level became normal.

Regards,

Sreenivasa