4
\$\begingroup\$

I'm building a USB-powered demo board with the Atmel SAM E70 controller, and a single hurdle I'm facing is the battery-powered operation. I intend to use a simple 3V coin cell battery as a backup power source, and want to preserve MCU's current date/time in a low-power mode.

I have a minor STM microcontroller background and kinda expected Atmel MCU to have similar capabilities, but E70 (and the whole line of Cortex-M7-based Atmel MCUs, it seems) lacks the VBAT input. E70 datasheet has a separate section (22.4.4 - "Using Backup Batteries/Backup Supply") on the topic, but it doesn't provide any ready answers - only a sketchy schematic and some power sequencing advice.

E70 has a sister chip - the V71, and I've found that V71's demo board from Atmel ("V71 Xplained Ultra") actually does provide a detailed schematic, which is far from being easy & compact:

Backup power schematic of V71 Xplained Ultra

It seems that Atmel expects system engineers to build a backup power system of their own; it's a bit of a letdown, as STM F4/F7 line has an embedded backup powering scheme.

Thus, 2 questions:

  1. Is there any easier (than building a scheme with an opamp and 2 additional ICs) way or a tried-and-tested schematic to provide a backup battery power for Atmel S70/E70/V70?

    Note: I was advised to use a separate RTC IC (like STM M41T62LC6F), but don't feel comfortable with adding an additional IC to the board.

  2. (in case any Atmel engineer happens to visit this question) What was the reasoning behind leaving out the VBAT/low-power controller out of the MCU?

\$\endgroup\$
10
+50
\$\begingroup\$

The idea behind the Vbat input on chips such as the STM32 is that the RTC and other low-power peripherals will normally be powered from the main Vdd, automatically switching to Vbat when Vdd is absent. On-chip, this can be implemented with something as simple as a diode-OR:

schematic

simulate this circuit – Schematic created using CircuitLab

To save pins (I assume), the SAME70 doesn't provide a separate connection for the RTC, instead having it take power from the main Vdd. However, as they show in the datasheet section you referenced, you can add the power-switching logic externally:

schematic

simulate this circuit

The schematic you posted is very complicated, because it's trying to do something slightly different: C103 on that diagram is a supercapacitor (100 mF!), so much of the circuitry is responsible for charging it at a reasonable rate. You're using a coin cell, so none of that is relevant.

There are several practical considerations with the schematic I posted. First, I recommend Schottky diodes, because you lose a lot less voltage that way. Second, I assume that you're generating a 3.3 V rail from the USB Vbus with an LDO linear regulator. If so, you'll want to take the feedback for the regulator from after the diode, so that the regulator can compensate for the diode drop. It would look something like this (ignoring bypass capacitors):

schematic

simulate this circuit

Third, this will try to power the entire chip from the coin cell when USB power isn't present. You have to be very careful to enter Backup mode as soon as you notice Vbus missing, or you'll drain the coin cell quickly. You'll want to use a GPIO pin to detect VBus (see schematic 2 in section 58.2.8).

\$\endgroup\$
  • \$\begingroup\$ Hi Abe! Thank you for your response. A little comment, though: so much of the circuitry is responsible for charging it at a reasonable rate - this isn't quite true; the problem is that Atmel chip has pretty complex power timing requirements (specific bits/pins need to be powered several ticks before Vdd power disappears, etc), and scheme actually tries to account for that. \$\endgroup\$ – Dmitriy Khudorozhkov Feb 22 '17 at 13:58
  • \$\begingroup\$ You're partially right about the charging scheatic, though :/ \$\endgroup\$ – Dmitriy Khudorozhkov Feb 22 '17 at 14:03
  • \$\begingroup\$ @DmitriyKhudorozhkov I'm not sure what you complex power timing requirements you refer to: as long as you're using the internal regulator for supplying the core, the SAME70 just requires that Vddin and Vddio be connected. On the circuit you posted, everything left of U101 is for charging the capacitor. \$\endgroup\$ – Abe Karplus Feb 22 '17 at 17:35
4
\$\begingroup\$

According to your datasheet, the LF oscillator will be powered by VDDIO. This is very annoying, as it means the pin buffers could stay powered also, and draw current if they are not put to the proper state before going to sleep. This should be checked...

You say you do not want to use a separate RTC IC in order not to add another chip to your board...

However, the schematic you post adds 17 parts including 2 ICs to your board, not including the switching regulator...

A RTC clock chip like PCF85063 or MCP7940 will have a full BOM cost of much less than $1 including crystal, and it solves your problem.

It also draws less current (1.2 µA versus 2-5 µA specified for your micro)...

It is also a proven solution that works and you can implement it without banging your head against your desk, in less time that you'll need to solve your current issue... And probably costs less too!

You remind me of myself when I was trying to use a feature in a micro that I shall not name, and the docs were, shall we say not very clear. It didn't work. I wasted lots of time on this. And then... Happiness and joy, here came the errata: "we're sorry, the feature you selected this microcontroller for doesn't actually work, thank you very much and have a nice day!"

So here's the deal: if the docs aren't clear... If the manufacturer doesn't demo the feature on their demo board... and if the chip is recent... and if no-one in the forums made the feature you want actually work...

No errata published yet means... No errata published yet.

Maybe it's there, maybe not... Will you spend a week to know if it's Schrödinger's feature?

Unless of course you're the guy Atmel hires to make their demo boards! In that case, send an email to the chip designers ;)

Maybe not the answer you want. Sorry!

\$\endgroup\$
  • \$\begingroup\$ Hi @peufeu! Thank you for your response. Yes, separate RTC IC is the solution I'm drifting for. Unless of course you're the guy Atmel hires to make their demo boards! - nope, I'm not affiliated with Atmel, unfortunately :) \$\endgroup\$ – Dmitriy Khudorozhkov Feb 22 '17 at 14:06
  • \$\begingroup\$ @peufeu I am often drifting to do just that "get it working because it should..." and hence wasting lots of time... I also want to use Vbat in SAM S70 ... I need to read your reply often as a reminder :) \$\endgroup\$ – VladBlanshey Jul 4 at 1:32

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.