massimo.vighetti
Posts: 6
Joined: Sat Sep 09, 2017 9:07 am

Boot problem with V3 on card designed past year

Sun Sep 10, 2017 8:23 am

In 2016 we designed and produced a card for the compute module.
We have about 300 of these cards that work properly.
The work cycle is for 4 to 10 daily switch-on and shutdown.
Now we have decided to mount the V3 module compute to replace the V1 and find the following problems

5% of the time does not start.
5% of the time starts with a delay of 10 to 30 seconds.
The problems are not related to specific compute module and card couple: one that has turned on regulary, 10 minutes after power off, does not turn on.
On the contrary, one that normally turns on sometimes does not work

Our card has these features:

3.3 volts provides 3 amps
VBAT is on the 3.3V but the problem does not change if it is powered to 5V
1.8V rises 3ms after 3.3V
Reset rises 3ms after 1.8V
EMMC_DISABLE_N not connected
EMMC_EN_N not connected
RUN not connected
HDMI_HDP_N not connected
SDX_VREF not connected
GPIO0-27_VREF 3.3V
GPIO28-45_VREF 3.3V

We added a reset supervisor to keep down RUN for 500 ms after card power on. Problem not fixed.

Any help please!!!!

PhilE
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 1361
Joined: Mon Sep 29, 2014 1:07 pm
Location: Cambridge

Re: Boot problem with V3 on card designed past year

Sun Sep 10, 2017 3:08 pm

Do you see any signs of life when power is applied?

Do you see the same kind of unreliability with a CMIO board?

massimo.vighetti
Posts: 6
Joined: Sat Sep 09, 2017 9:07 am

Re: Boot problem with V3 on card designed past year

Sun Sep 10, 2017 7:24 pm

Hi PhilE

We put a led between EMMC_EN_N and ground.
At power on it goes to 1.7V and after a few seconds it goes to 1.8V, both when it goes wrong and when it starts correctly.

On our carand LAN9512.
When it doesn't start uses 380/450 mA.
When starts 400/650 mA.

CM on IO Board V3 start right everytime.
On IO Board V1 (VBATT 3.3V) show problem with same frequence as on our card.

James Adams
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 94
Joined: Wed Mar 19, 2014 2:58 pm
Location: Cambridge

Re: Boot problem with V3 on card designed past year

Mon Sep 11, 2017 8:23 am

With your LED on EMMC_EN_N you are just measuring the forward LED voltage drop (through 10K to 3V3). Seems to be working to me.

This feels like a power supply issue, the CM3 takes a lot more current from VBAT. What is the capability of your upstream power supply? (I.e. the 5V input)

You need to look at the supply rails on an oscilloscope and check they start up as you suggest and are 'clean' and at the correct voltage.

Finally 1V8 can also take quite a bit of current - can you confirm the specification of this supply?

Note that the only significant difference between CMIO1 and CMIO3 board is that the VBAT is on 3V3 (CMIO1) and 5V (CMIO3).

James
James Adams
Raspberry Pi - COO & Hardware Lead

massimo.vighetti
Posts: 6
Joined: Sat Sep 09, 2017 9:07 am

Re: Boot problem with V3 on card designed past year

Mon Sep 11, 2017 11:21 am

Hi

1.8V 600mA
3.3V 3A

Attached V graph
Attachments
DS1Z_QuickPrint5.jpg
DS1Z_QuickPrint5.jpg (78.1 KiB) Viewed 1290 times
DS1Z_QuickPrint4.jpg
DS1Z_QuickPrint4.jpg (81.29 KiB) Viewed 1290 times

massimo.vighetti
Posts: 6
Joined: Sat Sep 09, 2017 9:07 am

Re: Boot problem with V3 on card designed past year

Mon Sep 11, 2017 11:23 am

Even for us it seems a supply issue.
What we don't undestand is how it is.

CM has 1A on 1.8V and 10A on 3.3V

James Adams
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 94
Joined: Wed Mar 19, 2014 2:58 pm
Location: Cambridge

Re: Boot problem with V3 on card designed past year

Mon Sep 11, 2017 11:42 am

Can you explain why there are 2 3V3 traces (blue and yellow)?

Also can you explain why the traces start at ~500mV (the yellow one then drops down before rising again)? Residual voltages on power rails is a not-uncommon cause of these kind of power-up issues.

Some idea of your powerchain architecture would also help.

James
James Adams
Raspberry Pi - COO & Hardware Lead

massimo.vighetti
Posts: 6
Joined: Sat Sep 09, 2017 9:07 am

Re: Boot problem with V3 on card designed past year

Mon Sep 11, 2017 3:49 pm

We powered VBAT with 5V on IO Board V1.2 and problem still arises.

About previous attachments

light bue: 3.3V
violet: 1.8V
yellow: RUN

RUN Low at power on depends on CM internal supervisior

James Adams
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 94
Joined: Wed Mar 19, 2014 2:58 pm
Location: Cambridge

Re: Boot problem with V3 on card designed past year

Tue Sep 12, 2017 10:30 am

2 things:

Can you try delaying RUN a bit more before bringing it high?

I'm still concerned about the residual voltage on the supplies - why are they starting at ~500mV?

James
James Adams
Raspberry Pi - COO & Hardware Lead

gioreva
Posts: 8
Joined: Tue Sep 12, 2017 4:49 pm

Re: Boot problem with V3 on card designed past year

Tue Sep 12, 2017 5:29 pm

I am the hardware guy of this project.

We added a reset supervisor, DS1233A-10+, to keep down RUN for 500 ms after card power on. Problem not fixed.

On this test, I turned off the card only 1 second before.
Normally wait at least 10 seconds before turning it on again, and tensions start from zero.
I have the doubt that it is a bad lot of CM3, but I do not know how to find out.

With the board IO V3 the problem does not show up, but with board V1.2 and VBATT at 5V the problem exists.

PhilE
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 1361
Joined: Mon Sep 29, 2014 1:07 pm
Location: Cambridge

Re: Boot problem with V3 on card designed past year

Tue Sep 12, 2017 8:29 pm

You haven't address James's point about residual voltage - can't you drop them to ground?

gioreva
Posts: 8
Joined: Tue Sep 12, 2017 4:49 pm

Re: Boot problem with V3 on card designed past year

Wed Sep 13, 2017 7:49 am

Only on that measurement, tension was not zero because I shut the card down 1 second before. Anyway, I add some resistance on power supply devices and measured them on each ignition.
This is exactly the measurement taken when CM didn't start
DS1Z_QuickPrint6.jpg
Zero Start
DS1Z_QuickPrint6.jpg (79.7 KiB) Viewed 1145 times

On this change I increased the delay of 3.3V, this is 5mS after 1.8V, and Reset 500mS after.
This is the measurement taken when CM didn't start
DS1Z_QuickPrint7.jpg
Change, added reset supervisor.
DS1Z_QuickPrint7.jpg (80.89 KiB) Viewed 1145 times

PhilE
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 1361
Joined: Mon Sep 29, 2014 1:07 pm
Location: Cambridge

Re: Boot problem with V3 on card designed past year

Wed Sep 13, 2017 8:18 am

Are you able to take the same measurements on CMIO 3 and compare?

James Adams
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 94
Joined: Wed Mar 19, 2014 2:58 pm
Location: Cambridge

Re: Boot problem with V3 on card designed past year

Wed Sep 13, 2017 8:27 am

When the CM3 does not start can you measure the voltage across C42 on the CM3 (bottom right corner of the module next to L1).

I'm finding it hard to explain why it works on a CMIO3 but on a CMIO1 (with VBAT at 5V) it does not - as these boards really are identical except the VBAT power and the CMIO3 has extra wires to the SD socket, which is not connected at all on the CM3 module (only CM3 Lite).

When you boot the module on CMIO1 and CMIO3 what other things are plugged in? (USB/HDMI etc..?)

Could you provide high resolution pictures of your failing CM3 (and CMIO boards) front and back?
James Adams
Raspberry Pi - COO & Hardware Lead

gioreva
Posts: 8
Joined: Tue Sep 12, 2017 4:49 pm

Re: Boot problem with V3 on card designed past year

Wed Sep 13, 2017 9:59 am

This photo on high resolution.
https://1drv.ms/u/s!AoR_HuKwnabrripi7mmkw6GQMjh2

On C42 there are 1.2716V when CM start correctly, 1.2020V when CM don't start.

I tried with 1.5 seconds of delay, on the rise of the RUN, but not start the same.

On IO board, i use USB-To ethernet adapter.
On My board i have LAN 9512-JXZ

On my board, if no CM present, led FDX1, are always off.
If the CM does not start, the LED will turn on after 10 seconds.
If the CM start, the LED will turn on after 10 seconds, after 2/3 seconds it shuts off again, after 2/3 seconds it turn on and it take IP from DHCP server.

gioreva
Posts: 8
Joined: Tue Sep 12, 2017 4:49 pm

Re: Boot problem with V3 on card designed past year

Thu Sep 14, 2017 6:06 am

I have 200 CM I can not use, and a thousand cards still do not go with the CM3
Or are the CM3s that are defective.
Can I ship my CM3 and my board to designers for check ?

gioreva
Posts: 8
Joined: Tue Sep 12, 2017 4:49 pm

Re: Boot problem with V3 on card designed past year

Thu Sep 14, 2017 6:24 am

I increased the run time of Run, at 3 seconds.
At the fifth turn on the CM started with 2 minutes of delay.

gioreva
Posts: 8
Joined: Tue Sep 12, 2017 4:49 pm

Re: Boot problem with V3 on card designed past year

Thu Sep 14, 2017 9:28 am

CM-IO v1.2 (Not work)
DS1Z_QuickPrint8.jpg
CM-IO v1.2
DS1Z_QuickPrint8.jpg (83.33 KiB) Viewed 1044 times
CM-IO v3 (Work)
DS1Z_QuickPrint9.jpg
CM-IO v3
DS1Z_QuickPrint9.jpg (82.02 KiB) Viewed 1044 times

PhilE
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 1361
Joined: Mon Sep 29, 2014 1:07 pm
Location: Cambridge

Re: Boot problem with V3 on card designed past year

Thu Sep 14, 2017 10:27 am

The only differences I can see are the 500mV residual voltage on the 5V rail on the V1.2 board and the extra half millisecond or so before the 1.8V rail comes up on the V3.

PhilE
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 1361
Joined: Mon Sep 29, 2014 1:07 pm
Location: Cambridge

Re: Boot problem with V3 on card designed past year

Fri Sep 15, 2017 10:07 am

We're a bit stumped here as to what might be causing the boot problems. I'd like to establish how far the boot process is getting, starting with whether the bootcode.bin is being executed. Assuming you have access to a serial port (or an LED):
  1. Put the CM3 into the CMIO V3 board and boot it.
  2. Make sure the firmware is up-to-date - "sudo rpi-update" (possibly after "sudo apt-get install rpi-update")
  3. Add the following to /boot/config.txt:

    Code: Select all

    enable_uart=1
    uart_2ndstage=1
    
  4. Attach the serial port to GPIOs 14 (TXD) and 15 (RXD), or an LED between 14 and GND.
  5. Reboot the CM and look for serial traffic - mine looks like this:

    Code: Select all

    Raspberry Pi Bootcode
    Read File: config.txt, 1792
    Read File: start.elf, 2870564 (bytes)
    Read File: fixup.dat, 6694 (bytes)
    [    0.000000] Booting Linux on physical CPU 0x0
    
    A LED should be on most of the time, flickering when characters are output.
Try this and report your findings with the different carriers.

gioreva
Posts: 8
Joined: Tue Sep 12, 2017 4:49 pm

Re: Boot problem with V3 on card designed past year

Mon Sep 18, 2017 3:54 pm

I can not get the problem out.
I removed all the filters, delays and capacitors.
I removed the console.
The only difference is updating raspbian.
It can be possible ?

PhilE
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 1361
Joined: Mon Sep 29, 2014 1:07 pm
Location: Cambridge

Re: Boot problem with V3 on card designed past year

Mon Sep 18, 2017 3:59 pm

The only difference is updating raspbian.
It can be possible ?
Can you explain what you mean by that? Does it work with an updated Raspbian? Did updating Raspbian break it?

gioreva
Posts: 8
Joined: Tue Sep 12, 2017 4:49 pm

Re: Boot problem with V3 on card designed past year

Tue Sep 19, 2017 7:31 am

Maybe, What fixed the problem was the update of raspbian.
After reverting back everything else the problem has gone away.
Now the only changed thing is the updated raspbian.

Return to “Compute Module”

Who is online

Users browsing this forum: No registered users and 3 guests