SAMA5D27 - some design questions

Go To Last Post
4 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi,

I'm preparing a design based on SAMA5D27 and there are some things which the documentation doesn't tell or they're buried so deep that I can't find them.
If somebody knows, please let me know what are the answers to these questions.or where they are and what to read then.

1) Why would I need to power VDDFUSE, or why and what for is the fuse box ever necessary? Do I need to do it with any defined timing with respect to other rails?

2) Can SAMA5D27 run at full speed without a heatsink in a closed box?

3) The SAMA5D2B-XULT board contains unidirectional level shifters on lines
SHDN (SAM) -> PWRHLD (ACT8xxx)    (transistors Q2 and Q1)
nPBSTAT (ACT8xxx) -> WKUP (SAM)    (transistors Q8 and Q9)
(both on sheet 4 of schematics)
What's the purpose of using these circuits? According to datasheet, nPBSTAT is open-drain, so it might be simply pulled up to VDDBU or 3V3. SHDN is a logic signal, so high level is ca. 3.3V, while ACT8xxx have logic high input threshold at 1.4V, thus no shift to 5V is necessary. Is there anything that I am missing but should be aware of?

4) At the same board, what is the purpose of using the high-pass filter (R9, C10, R12, R13, C8) for driving Q3? Is this used to extend the time before nPBIN becomes released? However, at this time the processor already has control over ACT8xxx by asserting PWRHLD, so no such extension seems necessary and the true reason must be different.

5) Concerning the ACT8xxx PMIC and SAMA5 as on the same board, is it allowed to make a modification and connect nPBIN permanently to ground through 50kOhm resistor for "always on" applications? This will be exposed to SAM via nPBSTAT line going to WKUP signal, how will the processor handle this? In my case the device will start as soon as it receives power, with no power button available to the user.

6) I will be using eMMC memory via the SDMMC0 interface. This interface comprises SDMMC0_CD (card detect) signal. Is it necessary to provide a defined logic level on this signal's pin, or forcing the eMMC detection via a relevant register is sufficient and the physical pin can be used as GPIO or for other peripheral?

Thanks a lot,
Adam

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I'd suggest you contact your local Atmel FAE and put the questions to them.

As for question 2 - this is a tricky question. Let's say the device consumes 0.2W working at full speed and the junction to ambient thermal resistance is 100C/W. (this is a guess - check the datasheet). So the chip will be around 20C higher than ambient. So you can run a 105C spec chip at a 85C ambient. Thus it depends on the ambient temperature of your closed box.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Well, yeah, I've sent these and more questions to Atmel like, let me check... 2.5 week ago.
Concerning question 2, I was also hoping that somebody has run the thing, maybe in similar conditions, and can tell based on experience.
The junction to ambient thermal resistance from datasheet (34.8 K/W) in this case can be unreliable, as I'm going to have the chip close to housing wall and this may further affect airflow, which I assume to be an important factor.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

In a closed box you're heating the internal air. Where does this heat go? Radiated by the enclosure?
You can't escape the thermodynamics!
An empirical technique to get an idea of how your box works heat wise is to get a suitably sized resistor and calculate the required resistance/voltage to generate then same amount of heat the chips would. In this case around 200mW. Put it in your box with a thermometer and measure how hot it gets in there over 24hours. Note the ambient temperature. This gives you the temperature rise. Based on 34.8K/W, the chip temp is around 7C higher than ambient.