LayerscapeのトピックRe: 1021a board bring up replace CPLD with Atmega328P

1021a board bring up replace CPLD with Atmega328P

jiye — Sun, 24 Nov 2019 16:44:41 GMT

Hi Guys,

I have a standalone ls1021a, eMMC, a Atemega328P chipset on the board and a JTAG interface. I want to bring the board up to initiates the power-on reset flow. I am trying to implement the CPLD logic into C using Atemega328P to achieve the board initialization stage.

Does any one know how many signals do I need to consider which I need to trigger them in the right sequence?

According to the QorIQ LA1021A Reference Manual.

I also have a question specifically during the POR sequence,

Do I have to give a signal to the cfg_eng_use0 or once the POR starts everything will be handled by itself.

Re: 1021a board bring up replace CPLD with Atmega328P

ufedor — Mon, 25 Nov 2019 09:31:07 GMT

It is reasonable to refer to the TWR-LS1021A design documents (including CPLD source) available at:

QorIQ® LS1021A Tower® System Module | NXP

The cfg_eng_use0 has to be connected as stated in the AN4878 - QorIQ LS1021A Design Checklist, Table 8. LS1021A reset configuration signals. Also in this application note 4.3.1 Configuration signals sampled at reset it is written:

"Reset configuration signals are sampled at the negation of PORESET_B. However, there is a setup and hold time for these signals relative to the rising edge of PORESET_B, as described in the chip's data sheet."

So this is the board designer's responsibility to properly strap (by resistor) the cfg_eng_use0.

> Do I have to give a signal to the cfg_eng_use0

After PORESET_B is deasserted the pin function is IFC_WE0_B which is pure output, so it is not possible to drive the signal externally during normal operation.

Re: 1021a board bring up replace CPLD with Atmega328P

jiye — Mon, 25 Nov 2019 15:27:09 GMT

(1) "Reset configuration signals are sampled at the negation of PORESET_B" --------------------------> does negation mean the falling edge of this signal then to low?

According to the data sheet Section 4.4.1 of LS1021a reference manual:

6. External system logic negates PORESET_B after its required hold time and after
POR configuration inputs have been valid for their required setup times.
7. The device samples the RCW source POR configuration inputs (cfg_rcw_src[0:n])
on deassertion of PORESET_B to determine the RCW source. Note that the POR --------------------------> does deassertion means rising edge then to high
configuration inputs are sampled only on a PORESET_B.

(2)so basically after drive the PORESET_B to high for 1ms minimum, it will start validate the POR inputs to determine the boot source, so all I need to do is to drive this pin high for 1ms high then low for amount of time after the POR configuration inputs have been valid for their required setup times then what should I do after?

Re: 1021a board bring up replace CPLD with Atmega328P

ufedor — Mon, 25 Nov 2019 16:27:08 GMT

1) PORESET_B is an active-low cignal (_B), thus:

negation = deassertion = low->high edge

2) Not correct - see 1) and refer to the POR sequence from the processor's RM.

Re: 1021a board bring up replace CPLD with Atmega328P

jiye — Mon, 25 Nov 2019 16:44:34 GMT

so basically :

4.3 Power-on reset recommendations
Various chip functions are initialized by sampling certain signals during the assertion of PORESET_B. These power-on reset
(POR) inputs are pulled either high or low during this period. While these pins are generally output pins during normal
operation, they are treated as inputs while PORESET_B is asserted. When PORESET_B de-asserts, the configuration pins
are sampled and latched into registers, and the pins then take on their normal output circuit characteristics.

so by saying asserted in this case means the signal is active low?

Just want to double confirm with you last time:

So basically this signal is active low and then if I make it stays low for 1ms then after 4 system clocks drive it to high (the falling edge requires max 1 sys clock cycle) then it will start sampling the boot source cfg_rcw_src[0:n]?

before asserting this PORESET_B signal do I need to perform a toggling event on this pin?

Re: 1021a board bring up replace CPLD with Atmega328P

ufedor — Mon, 25 Nov 2019 16:51:35 GMT

> so by saying asserted in this case means the signal is active low?

Yes.

> So basically this signal is active low and then if I make it stays low for 1ms

> drive it to high (the falling edge requires max 1 sys clock cycle) then it will start
> sampling the boot source cfg_rcw_src[0:n]?

Correct - also refer to the LS1021A Data Sheet, Table 24. RESET initialization timing specifications

> before asserting this PORESET_B signal do I need to perform a toggling event on this pin?

No.

Re: 1021a board bring up replace CPLD with Atmega328P

jiye — Mon, 25 Nov 2019 17:50:59 GMT

how about the power reset control signal from the PMIC?

is this be asserted automatically once PORESET_B is high?

Re: 1021a board bring up replace CPLD with Atmega328P

ufedor — Tue, 26 Nov 2019 04:24:04 GMT

> is this be asserted automatically once PORESET_B is high?

What do you mean?

Re: 1021a board bring up replace CPLD with Atmega328P

jiye — Tue, 26 Nov 2019 13:18:49 GMT

There is POR_B control pin coming out of the PMIC

I am not sure if this one POR_B is the one in the CPLD por_assert

parameter [2:0]por_assert = 3'b101; //POR reset

(1) If so what does the pin do in terms of PORESET_B, how does POR_B relates to PORESET_B ?

Re: 1021a board bring up replace CPLD with Atmega328P

jiye — Fri, 29 Nov 2019 19:51:51 GMT

Hi ufedor,

Any specific singal requirements before the ls1021a standalnoe prototype to detect the codewarrior tap via the usb interface?

Re: 1021a board bring up replace CPLD with Atmega328P

ufedor — Sat, 30 Nov 2019 04:14:09 GMT

Sorry, the request is not clear.

What does it mean "to detect the codewarrior tap via the usb interface"?

Re: 1021a board bring up replace CPLD with Atmega328P

jiye — Sat, 30 Nov 2019 04:35:12 GMT

I meant in order to flash the ls1021a standalone chip using The codewarrior tap are there any signals have to be enabled first

Sent from my iPhone

Re: 1021a board bring up replace CPLD with Atmega328P

ufedor — Sat, 30 Nov 2019 08:19:45 GMT

In the CPLD source:

input por_b; //PORB of 34VR500

Re: 1021a board bring up replace CPLD with Atmega328P

ufedor — Sat, 30 Nov 2019 08:20:38 GMT

The processor has to be connected as described in the AN4878 - LS1021A Design Checklist.

Re: 1021a board bring up replace CPLD with Atmega328P

jiye — Sun, 01 Dec 2019 20:21:35 GMT

can you take a look at his regarding the JTAG/COP port is used case

This is from the LS1021a data sheet below:

Also this is my design below:

(1)I used a resistor R88 to connect TRST and POWER RESET Line does this

match the requirements?

Also,

(2)

this graph here shows that the ALSLEEP signal goes up after the PORESET_B

deasserted. I used my oscilloscope to check both signals I found that the

asleep signals goes high at the point when PORESET_B goes low. Is that

actually normal ?

On Sat, Nov 30, 2019 at 12:34 AM Elias Bitbaker <elias.bitbaker@gmail.com>

Re: 1021a board bring up replace CPLD with Atmega328P

jiye — Sun, 01 Dec 2019 21:31:47 GMT

The one on the top is the POWERET_B and the bottom is the ALSLEEP signal

the behaviour is not following what the datasheet says.

On Sun, Dec 1, 2019 at 4:21 PM Elias Bitbaker <elias.bitbaker@gmail.com>

Re: 1021a board bring up replace CPLD with Atmega328P

ufedor — Mon, 02 Dec 2019 07:03:55 GMT

> asleep signals goes high at the point when PORESET_B goes low

This is OK considering that ASLEEP is tri-stated and has a pull-up.

Re: 1021a board bring up replace CPLD with Atmega328P

jiye — Mon, 02 Dec 2019 13:32:05 GMT

can you take a look at his regarding the JTAG/COP port is used case

This is from the LS1021a data sheet below:

Figure - A

Figure - B

Figure - C

(1)I used a resistor R88 to connect TRST and POWER RESET Line does this

match the requirements as shown in Figure A? These two pins are then pulled up by a 10k resistor to Vcc_O1VDD, my design does not have HRESET_B pin.

(2)So as picture shows below: why there are HRESET_B and PORESET_B on the left side what source triggers them?

Re: 1021a board bring up replace CPLD with Atmega328P

ufedor — Mon, 02 Dec 2019 14:19:33 GMT

The R88 is OK, but you have to consider the QorIQ LS1021A Data Sheet, Table 24. RESET initialization timing specifications:

"Maximum rise/fall time of PORESET_B is 1 SYSCLKs"

Re: 1021a board bring up replace CPLD with Atmega328P

jiye — Mon, 02 Dec 2019 14:33:11 GMT

(1)by saying that because I use c logic to implement the start up logic what I am doing is before asserting the PORESET_B, I drive it to high then low (asserted) for 50ms (tho only needs 1ms min) then I de-assert PORESET_B then I did DDR reset, PCIE reset some I/O interface enabled. How am I able to manage the rise/fall time via the C code on AVR studio IDE which I am using to flash the code into the atmega on my board

(2) What is the relation between PORESET_B and TRST_B

seems like HRESET_B and TRST_B has relation but TRST_B and PORESET_B does not according to the CPLD code on the TWR-LS1021A eva board:

/CPU, DDR and peripheral devices reset
assign poreset_b_18 = pwr_hrst_n && jtag_rst_b_18 && sw_rst_n;
assign cpu_trst_b_18 = pwr_hrst_n && k20_trst_b_18;
assign rst_flsh_n = pwr_hrst_n;
assign elev_rstout = hreset_b_18;
assign ddr_rst_b = hreset_b_18;
assign ethphy_rst_b = hreset_b_18;
assign rst_usbhub = hreset_b_18;
assign hdmi_rst_b = hreset_b_18;
assign rst_pcie_n = hreset_b_18;