Keep the Realtime Clock running on the iMX53QSB withouth the +5V Wall supply

[SrikanthKotagir]

Hi,

   I am trying to preserve the date/time values in android when I remove the 5V Wall power supply. The imx53 Hardware reference( IMX53QSBRM) states in Section 3.20

" On the Quick Start board, there are two through-holes (JP1 and JP2) next to the power connector. These through-holes are positioned to hold a Lithium coin cell battery (Sanyo ML1220-VM1 or equivalent). For proper operation, the coin cell posts should be soldered direction to the Quick Start board, with the positive terminal connected to JP1 and the negative terminal connected to JP2. The DA9053 PMIC will charge the coin cell when 5V Wall Power is available. When 5V Wall Power is removed, the coin cell will provide power only to the RTC power rail (VLDO1) supplying power to the i.MX53 processor. The length of time a coin cell can power the RTC subsystem may vary" 

  I have tried it but the date/time settings are lost when the power is removed for more than 5 seconds. Dis anyone else try this?

Thanks,

Srikanth

[martinernst]

I was looking for J1 & J2 I do not see them on my board. They are suppose to be next to the DC power jack. I don't have the board at work today but I will lookup what Rev board I have.

Marty

[JohnGoodman_ng]

The problem is a design problem with the PMIC and which VLDO was chosen for the SRTC of the i.MX53 QSB.  The RTC in the DA9053 is not operational, so you have to use the SRTC in the i.MX535.

The way I solved the problem was to use an AAA battery and cut the SH9 jumper and insert a MMBT2222A SOT23 as a linear regulator.  The linear regulator supplies current to the battery and the SRTC when external power is available and the battery supplies power to the SRTC when external power is not present.                 

The 100 ohm thru-hole resistor is attached to the cathode of the battery to limit surge currents.   I have tested the circuit with Linux and WEC 7 and the SRTC works correctly with both OS'es when external power is present and not present.

The problem is corrected on i.MX53 Quick Start –R board with the MC34708 PMIC.  On the –R board, you will have to select the RTC in the PMIC rather than the SRTC in the i.MX535.  On the -R board, RTC battery connections were not provided.

[GiulioBenetti]

Hi,

the only way is to cut SH9 and use an external LDO with 1,3V output.

This is the only possible workaround.

The problem is that they connected together VDDRED and VDDOUT, as you can see on data sheet.

This is a silicon bug. Nothing to do more.

[c10ud]

Thanks for your prompt response Ture.

Quoting from the HW User Guide:

The first voltage regulator to power on is always VLDO1. This regulator supplies a maximum of 40 mAcurrent at 1.3V and powers on only the Secure RTC module of the i.MX53 Processor. This turns on theRTC Clock (32.768KHz) and Watch Dog features. In the event a System Reset is triggered, or the QuickStart board is placed into Standby, VLDO1 will remain powered ON. The only time that VLDO1 will turnoff is if all power is removed from the Quick Start board, or if a software command is sent to the PMIC toturn off VLDO1. In the case that a developer attaches an optional coin cell to JP1/JP2, the coin cell willprovide power to keep VLDO1 operating.

then i see

1. NVCC_SRTC_POW (VLDO1)

If i got what you meant then:

• The DA9053 will mantain power only to its oscillator when a backup battery is soldered onboard. But that is currently useless since the external pins are grounded (so you cannot provide an oscillator)
• The DA9053 cannot mantain power on VLDO1 in any way when wall power is removed (i.e. no i2c programming or things like that)

I have two questions:

• Is there some "reccomended"/widely tested workaround?
• Will (as far as you know) new boards suffer the same issue?

Thanks in advance

Ture Nielsen said:

Hello Richardo,

Yes, XIN and XOUT are the PMIC oscillator.Maybe the manual say so but i am affraid that the board hasn't been designed according to that. the supply for the RTC on i.MX5 is the NVCC_SRTC_POW but that has got to be 1.3V so you cant connect the back up battery directly to this.

With the correct board design this is working!

[TureNielsen]

Hello Richardo,

Yes, XIN and XOUT are the PMIC oscillator.Maybe the manual say so but i am affraid that the board hasn't been designed according to that. the supply for the RTC on i.MX5 is the NVCC_SRTC_POW but that has got to be 1.3V so you cant connect the back up battery directly to this.

With the correct board design this is working!

[c10ud]

Hello Ture,

I assume XIN and XOUT are for the PMIC's oscillator? The manual says the backup battery is for the processor's RTC. Am i wrong? (see OP)

greets

[TureNielsen]

Hi Ricardo,

I have been looking through the schematic for the QS board and there are a few issues with the board design. You are correct  when you say that you can connect the backup battery to JP1 and JP2 and this is connected to the PMIC DA9053. But as you can see from the attached schematic cut out the oscillator is not connected to the XIN and XOUT pin's.

As far as I know the XIN and XOUT pins are not brought out as test points so you cant get the RTC circuit to work on from the back up battery using the QS board.

[c10ud]

Srikanth, i have reproduced your test (and i think you meant SH9) and the result is the same.

Also, i noticed that the consumption is somehow high, i noticed high voltage drops from the battery ~0.3V in a very short time when wall power is not applied.

I've also seen that the VBAT can be regulated somehow in the DA9053 (at least the user guide says so even if it doesn't specify how) but i'm unsure if that could be the problem..

I'm on some more test though, even if i hope to get a word from Freescale. 

[c10ud]

Hello, i tried both with a Sanyo and a generic CR1220. However i tried to ask to some Freescale employee about this weird issue:https://answers.launchpad.net/linaro-landing-team-freescale/+question/173235 greets

[SrikanthKotagir]

Hi Richardo,

   The JP1 and JP2 are internally connected to the DA9053 to the backup battery pins.  When I was looking at the output of the DA9053 to the RTC clock pin to imx53 ( you can monitor it on the SH29 on bottom of the board), I notice that when the battery is not connected, when I turn off the wall supply, the voltage goes from 1.3V to 0 immediately. However if the battery is connected, then it maintains the 1.3V up to 2-3 seconds after power down.

   What battery have you been using? I was using the VL1220-VC battery from panasonic.

Thanks,

Srikanth

[c10ud]

I'm getting the same problem as yours. Are JP1 and JP2 supposed to be both GND when nothing is attached to them?

Plus, i am also trying with the linaro 11.09 dev images: only rtc0 is available (the mx internal rtc) and it seems to work ok (the whole hwclock thing)...it just loses the time when power goes away, and the battery in JP1/2 doesn't help.

greets

Srikanth Kotagiri said:

Well, the DA9053 clock is not running because the 32kHz external oscillator is not available for da9053. I just looked it up on the imx53qsb schematic.

I will keep posting on any updates.
Srikanth Kotagiri said:

Hi,

   I followed up on Remi's post and to see which RTC is powering up the QSB. The default kernel configuration uses the mxc_rtc, the RTC available on the processor and only the /dev/rtc0 device is available.

the rtc0 is ticking while the rtc1 is not. Any one has any idea why?

Thanks

[SrikanthKotagir]

Well, the DA9053 clock is not running because the 32kHz external oscillator is not available for da9053. I just looked it up on the imx53qsb schematic.

I will keep posting on any updates.
Srikanth Kotagiri said:

Hi,

   I followed up on Remi's post and to see which RTC is powering up the QSB. The default kernel configuration uses the mxc_rtc, the RTC available on the processor and only the /dev/rtc0 device is available.

the rtc0 is ticking while the rtc1 is not. Any one has any idea why?

Thanks

[SrikanthKotagir]

Hi,

   I followed up on Remi's post and to see which RTC is powering up the QSB. The default kernel configuration uses the mxc_rtc, the RTC available on the processor and only the /dev/rtc0 device is available.

   I thought maybe the DA9053 is being supplied with power and enabled the CONFIG_RTC_DRV_DA9052=y and recompiled the kernel. Now I had two RTC devices, the /dev/rtc0 corresponding to the mxc_rtc and the /dev/rtc1 for da9052-rtc.

  However looking at the information from the sysfs the RTC clock on the DA9053 is not running. It is always set to Jan 1 2000.

Here is the commands and output i used for reference

# cat /sys/class/rtc/rtc1/name
da9052-rtc
# cat /sys/class/rtc/rtc1/date
2000-01-01
# cat /sys/class/rtc/rtc1/time
00:00:00
# cat /sys/class/rtc/rtc0/name
mxc_rtc
# cat /sys/class/rtc/rtc0/date
1970-01-02
# cat /sys/class/rtc/rtc0/time
00:14:52
# cat /sys/class/rtc/rtc1/name
da9052-rtc
# cat /sys/class/rtc/rtc1/date
2000-01-01
# cat /sys/class/rtc/rtc1/time
00:00:00
# cat /sys/class/rtc/rtc0/name
mxc_rtc
# cat /sys/class/rtc/rtc0/date
1970-01-02
# cat /sys/class/rtc/rtc0/time
00:15:21

As one can see, the rtc0 is ticking while the rtc1 is not. Any one has any idea why?

Thanks

[rlorriaux]

Make sure you are using the right RTC device. Several i.MX reference boards have two RTCs (one on the i.MX and one on the PMIC - which the DA9053 has) corresponding to /dev/rtc0 and /dev/rtc1. You want to check that you are using the RTC that is actually powered by the coin-cell battery.