Questions the hardware design

[rayc345]

Hello, Eli H. Thank you very much for sharing your design about the IMXRT685 DSP chip! I'm interested in this chip recently, I found your design and want to take it as one of my references. About the circuit, I think you made a small mistake in the pull-up of PMIC I2C, the two resistors should be connected to 3.3V since the VDDIO_1 is 3.3V in this design. Also I have some puzzles about the USB. According to UM11147, two termination resisters of 45 Ohms are inside the chip. Why did the design use two 33 Ohms resistor? In IMXRT685-AUD, these two resistor do not exist. Best regards. Ray Zhang.

[ehughes]

Hello Ray.

Thanks for taking the time to report your findings. Yes, the I.MXRT685 is a nice chip. There is a lot of availability and it is very powerful!

1.) Yes, you are correct. The Rev A design (which is not on Github) used 1.8v IO for VDDIO1. I forgot to change it. It might work as is w/ the 1.8v pull-up.

2.) Sorrry for the confusion here. Those should be labeled as Zero ohm. I typically have pads on USB "just in case" and need to make adjustments.

I'll make a note on the schematic about 1 & 2 and push up the changes.

Side note.... This design did not include any connections for "recovery" boot. There is a mode where you can boot to RAM via a 1-bit NOR SPI if other boot sources fail (such as the SD card). I designed the SuperMonkey XIP on the FlexSPI but like to to exectute everything from internal RAM. The Recovery boot is simple add.

Let me know your your own design goes. I'll be working on another RT685 design again soon.

[ehughes]

Also, something else to consider. This design uses 0201 capacitors underneath the chip for decoupling. I think 0402's could be used of twice the value (.22uF) and use half as many. It would make routing easier.

[rayc345]

Thank you for your explaination. Currently, I'v finished designing the schematic and PCB. Now I'm waiting for the arrival of the chip and checking the designing concurrently. The PCB manufacturing and SMT is quite expensive, about $214 for 5 PCBA pieces( including material costs) so I want to succeed at the first time.

On the top of PCB, there are the RT685 main IC, power managment and connectors. On the bottom, bypass capacitors, Reset control and ESD protection diodes are placed. The top copper layer is used to route high-speed signal( HS SPI and USB), power managment, XTAL. Inner 1 layer is ground plane and is not shown in the picture below. Inner 2 layer is used to route I2S signals and PMIC communicayion signals. Bottom layer is used for power lines. One serial port is connected to the SWD header to allow communication to PC through a bridge. Four I2S( in two of them only DATA signal is routed out) connected to the header, allowing audio/baseband signal input from another board.

For the past several months, I'v been able to use MATLAB/Simulink as the simulation platform for several Cortex-M MCUs on custom-made boards so I can use Simulink for build up my algorithm and get it executed, verified or profiled by a single click. I would like to extend this software platform to the HiFi4 DSP so I can process audio/radio signals more efficiently. That's why I want to use RT685.

In the layout of the components under the 685 chip, I imitated the design or MonkeyRT685. I use mainly 0201 capacitors so that almost every ball of the chip can get a bypass capacitor which is recommand by the chip vendor.