VBACKUP Current Draw

Added by Anonymous about 11 years ago

I am attempting to replicate the VBACKUP current draw information you supplied in the Hardware Design section of the wiki. I have a 3359-GX-226-RC module on the MityARM development board. I removed the battery (which was dead by the way) from the development board and replaced it with a 3V supply.
I am unable to get the advertised ~10uA current draw on VBACKUP and have noticed some unusual current numbers.

With the SOM inserted in the board and no Main Power (+12V to board), VBACKUP draw is much less than 1uA.
With Main Power Supplied (board boots off SD card supplied with dev kit), VBACKUP draws ~0.3uA.
When I remove Main Power from the board, VBACKUP draws ~175uA (WOW!! this is a lot and maybe explains why the battery was dead)
If I remove the 3359 SOM and re-install it in the development board, VBACKUP now draws much less than 1uA again.

I've replicated the above on two different boards and two different 3359 modules.

In the MityARM-335x RTC Information wiki article, the measured 10.2uA current consumption on VBACKUP was when 3.3VIN was powered down and the PMIC RTC was enabled and active. I wondered if maybe the excessive current draw was because I hadn't initialized the RTC in the PMIC. In order to do that, I powered up the board in u-boot and ran the following u-boot commands:

  • i2c dev 2
  • i2c reset
  • i2c mw 2d 3f 10 (Enable RTC module & 32KHz crystal oscillator)
  • i2c mw 2d 5 13 (Set year to 2013)
  • i2c mw 2d 4 3 (Set month to March)
  • i2c mw 2d 3 26 (Set day to 26)
  • i2c mw 2d 2 15 (Set hour to 3pm)
  • i2c mw 2d 1 30 (Set minute to 30)
  • i2c mw 2d 10 1 (Runi2c clock)

Then verified that the RTC was running with u-boot command "i2c md 2d 0 10".

When I removed the Main Power, the current draw on the VBACKUP went to ~173uA.


Are there other PMIC registers that need to be initialized/set in order to lower the VBACKUP current when Main power is removed?

On the SOM, I see the VRTC power from the TPS65910 goes to a Cap and a pullup resistor tied to the BOOT1 pin, does it go any where else like to the AM3359? Is there a jumper that needs to be removed to isolate this power rail?

What code and hardware configuration did you use to obain the 10.2uA current draw on the VBACKUP input?

Replies (6)

RE: VBACKUP Current Draw - Added by Michael Williamson about 11 years ago

Hi Mark,

We need to confirm what revision of module you have (there was Errata on earlier modules where the VBACKUP draw is quite high). Can you please get the 80- number off the module?

Check the Module Revision History Page and see if you have a module with the correction.


RE: VBACKUP Current Draw - Added by Anonymous about 11 years ago

Hi Mike,

Thanks for the quick response. I believe I have modules covered in the errata. Of the three modules I've tested with, one is an 80-000444RC-1-D and the other two are 80-000509RL-1B.

I also have one of the AM3354 DDR3 modules p/n 80-000517RC-1B. What VBACKUP current draw should I expect from this?


RE: VBACKUP Current Draw - Added by Michael Williamson about 11 years ago

Hi Mark,

For the DDR2 modules, I'm pretty sure you need the -1C revision for the fix. However, the AM3354 DDR3 module should have the RTC fix already applied. I tested the first prototypes here and was able to see 10 uA on them as opposed to the higher 100+ uA draw.

You need to power up the module and run the most recent version of the 3.1 kernel that includes the PMIC RTC driver. The driver takes the RTC out of reset (once it is out of reset, it remains "on" through main VIN voltage off cycles). Once that happens, then you should see 10 uA draw on the VBACKUP with VIN powered down. The default kernel on the MMC probably does not have this patch applied. If you need help getting a kernel image with the patches, let me know.


RE: VBACKUP Current Draw - Added by Anonymous about 11 years ago


I tested the AM3354 DDR3 module here as well and also saw the ~10uA current draw on VBACKUP.

One other thing I tried worked for me in lowering this current. On your module, it appears that you have the PMIC BOOT0 input tied to GND and BOOT1 pulled up to VRTC via resistor R86. There is a register inside the TPS65910 called PUADEN_REG (offset 0x1C) that defaults to a value of 0x9F. The lower two bits of this register control programmable pull downs on the BOOTx inputs. The internal pulldown on BOOT1 is "fighting" the pullup resistor on the SOM. I changed the value of this register to 0x9C which disables the pull-downs. The current draw on VBACKUP dropped to ~5.5uA. This number more closely matches the 6uA typical VBACKUP current specified on page 10 of the TPS65910 data sheet.

If you can confirm this is acceptable, this would almost double the life of any battery connected to the VBACKUP pin.


RE: VBACKUP Current Draw - Added by Michael Williamson about 11 years ago


Great catch! This is certainly acceptable, and we'll likely modify either our u-boot code or kernel code to disable these pull-downs. Both pins are tied externally (BOOT1 to VRTC via 10K, Boot0 to ground), so there is no need for these to be enabled. This would indeed significantly improve battery life.

Thanks for the post.


    Go to top
    Add picture from clipboard (Maximum size: 1 GB)