i.mx28 RTC backup current on 24MHz Xtal

[slay]

Hi Everyone,

As the title says: I'd like to know what the typical current would be when the i.mx28 is in RTC backup mode when running on the 24 MHz Xtal. The idea is, that when the (5V) power is removed, that the battery connection is backed up by a goldcap as can be found in schematics as "Optional RTC backup". However: we do only have a 24 MHz Xtal and no 32.768 KHz Xtal. With the 24MHz RTC persistent bit in the hw_rtc_persistent0 register I did manage to keep the 24 MHz Xtal active when 5V power is removed and the RTC indeed keeps running.

However, this only lasts for a couple of minutes, since the goldcap voltage drops quite fast. I know that the typical RTC backup curent is about 13,5 uA but that is for running on the 32.768 KHz Xtal. Is there any known value for the 24 MHz operation in this mode?

Regards,

Ruud

[markwilliams]

Hi Ruud,

How long long do you want to keep the clock running for?

I originally designed my 5V only imx28 board to use a NiMh battery to keep the internal rtc alive (Re: i.MX28 RTC Backup Battery Connection ) but in the end I switched to use an external rtc with super cap.

There are a few issues trying to keep the clock alive:

1) the keep alive current is large compared to RTC ICs

2) the voltage must be above 3.1V according to the data sheet to keep the clocks running so you only have 3.3 to 3.1V drop. In practice I found it to work down to around 2.5V but I don't like designing off-datasheet

3) using the 24MHz crystal clock for the imx28 is not particularly accurate. Especially as it needs to be placed next to the imx28 which gets warm under 5V only operation due to the internal LDO so your run-time accuracy may be tens of minutes out per month

Using an external rtc I could charge my super cap to 5V, the RTC runs down to approx 2V and only takes 1uA. I could also place it away from warm components but I am now switching to a TCXO to ensure accuracy of <10 mins error per year.

I am not sure you will achieve much better than minutes of keep alive directly with the imx28. I would expect the 24MHz keep alive to be much worse than 32.768kHz too.

One more thing to check - are you charging your super cap through a schottky? If so check its reverse leakage as some of the low vf schottkys have terrible reverse current that may be helping to drain your battery back via your charging supply!

Mark

[slay]

Hi Mark,

Thanks for your reply. We know this is not the most efficient RTC backup circuit, but we do aim for "a couple of hours". I came across this 3.1V limit also, but in practice the 24MHz is even still running when the voltage drops to about 1.5V and when powering up from this low voltage, the RTC is still correct. Indeed not specified according to the datasheets, but it seems that 2V could be a safe margin... The starting voltage however is not 3.3V, but 4.0V, isn't it? With a 2V drop and a 47mF goldcap we have about 26 uAh to consume. Not too much, but enough for almost 2 hours if we could reach the 15 uA.

We do charge the supercap via a Schottky (from the VDD4P2) so we will certainly check this for leakage current. Thank you! One question though: The supercap is also connected via the 1K R146 to VDD4P2. If there would be leakage current, wouldn't this 1K be a much bigger leak then the reverse leakage of the diode? Or should R146 be removed as well?

Thanks,

Ruud

[markwilliams]

sorry I forgot the battery pin was ok up to 4v (4.2v?).

With regards to the resistor I couldnt get a clear answer from freescale on this. Their an4199 document on the pmu contradicts a couple of other documents.

if you look at the link in my post above I put extracts from two freescale reference designs. Both remove the resistor to vdd4p2 For supercap or coin cell applications. I removed this resistor in my design too prior to switching to an external rtc.

i needed at least a week of backup. The nimh battery was costly and also had safety certification and soldering issues. The supercap works great (I used a Panasonic part) with a microchip mcp7940n rtc.

maybe you could break the track to the battery pin and measure the imx28 backup current to see if it truly is going into the imx28?

good luck!

mark

[Yuri]

There are no specs just for the case, the i.MX28 Datasheets provide

Offstate Current in Table 8 (Recommended Power Supply Operating

Conditions) for 32 KHz. When using 24 MHz, a little more value is expected.

What may be important : the current of several hundred uA  is detected for

VDD5V only applications mode, shown on the Figure 9 of AN4199. 

R145 should be removed, BATTERY and DCDC_BAT has to be connected without

resistor to get tens uA.

Have a great day,
Yuri

-----------------------------------------------------------------------------------------------------------------------
Note: If this post answers your question, please click the Correct Answer button. Thank you!
-----------------------------------------------------------------------------------------------------------------------

i.MX28 VBat power consumption (Power off) RTC only

[slay]

Hi Yuri,

Thanks for your reply. I know about the correction for AN4199, so that was already in our design by an extra 0E resistor between BATTERY and DCDC_BATT. At first we left in the R145, since this only changes the reference to the 4P2 to 500E in stead of 1K. I don't know how important this is to be 1K and not much higher or lower?

But also with R145 removed, the higher current was still there. I did some experimenting with HW_POWER_CTRL and HW_POWER_5V_CTRL bits, but sofar no luck (except for a few percent improvement). But therefore it is good to know that at least it should be feasible even with the 24MHz used.

So I think the answer is in the power configuration bits then. Would you happen to have any tips for setting these bits for this (5V-only with Goldcap RTC-backup) configuration?

Thanks,

Ruud

[Yuri]

  Let me know Your e-mail so, that I forward You discussion from internal Community,
which can be helpful.

Regards,
Yuri.