Automatically run after power recovery

[thomas88]

Hi, 

Is it possible to run Rock64Pro without power button? Some time ago the power was cut in my home and when it recovered unfortunately my board didn't comes up by itself... It is a little bit annoy, especially when you're not at home and you would like to use you board remotely but you can't...

Cheers!

[Bullet64]

Normally ROCKPro64 will boot again when power comes back.

[thomas88]

Normally ROCKPro64 will boot again when power comes back.

Unfortunately not... I also noticed weird symptom. When i disconnected the power supply and back it later the board didn't come up. I had to push the power button to run the board. Is there something wrong with my unit? Thanks for any advice!

[Nikolay_Po]

As I can see at the page 18 of rockpro64_v21-SCH.pdf, there may persist auto-boot feature, take a look at a diagram, marked red.

If these components are installed, the board will turn on when the power is applied. If it is not turning on, check the network at  pin #34 (VDC) of U5 RK808-D chip. When the voltage on this pin is raising above 0.6V the power supply IC will turn on the power.

IMHO, the network on this pin is strange. If the voltage will rise slowly (it shouldn't because the voltage is producing by a 12 to 5V buck converter), the capacitor C204 will charge slowly and may not to produce enough VDC pulse. Also the VDC input of supply system IC has a debouncing time of 100ms. And in some circumstances the pulse may be not long and high enough to be detected with a margin.
I have none board yet and can't check the speed of voltage rise at RK808 #34 VDC pin. And I don't know what reasons have board developers to make such capacitor-coupled startup circuit. Probably it is perfectly OK and just was not installed.
Just for a reference, here is the network from RK808 data sheet:

This circuit should turn the power system on at about 8V input. But sure, it is not so simple as at a glance. The LDO above the buck converter is not present in the design of ROCKPro64 v2.1. The voltage at the VDC will stat rising earlier than 5V buck regulator VCC5V0_SYS output and VCC_RTC voltage. That is the cause the RK808 may not turn on. It will not see the voltage rise and power-on interrupt will not trigger, no power-on will happen. Probably, this was a reason to implement the startup circuit as shown at top diagram.

In case of auto-boot circuit is present but not working, I'd recommend to try this circuit (in red):

[mani]

organized schematic

[TheWiley]

For what it's worth, I see this behavior too - my RockPro64 has never turned on when I plugged it in or after a power cut/restoration - I have always had to press the power button. Mine is in a case where it's a bit hard to see the LEDs, so I'd definitely consider it an improvement if the board could power on automatically.

I'll try to have it out of the case in the next few days and look for the circuit you've described.

[Nikolay_Po]

Mine second RockPro64 was arrived with 5A power supply. I tested autostart circuit at VDC pin, see my previous post https://forum.pine64.org/showthread.php?...4#pid46594

In short, with both 3A and 5A Pine64 power supply the circuit provides long enough pulse to turn on CPU power converter. I will add a diagrams from oscilloacope when will be at place.

[Nikolay_Po]

Here are the diagrams of power on at pin 34 of RK808 power management system.
This is a startup from Pine64 12V 3A power supply:
Blue is voltage on 34pin of RK808 "VDC", yellow is system power voltage VCC_5V0_SYS

This is a startup with Pine64 12V 5A power supply:

A close-up:

It is seen that in both cases the VDC pulse is higher than 0.6V (higher than 0.6 screen cells) and longer than 100ms (longer than 1 screen cell).
RK808 power management system power plug-in detection time, TdbPLUG_IN declared as "typical", 100ms, with no maximum value. Pulse length on scope diagram is about 170..175ms. 70% margin from 100ms. Should be OK. But only in case of X5R dielectric for C204 1uF capacitor.

For initial question of @thomas88 and @TheWiley I'd recommend to look at the LEDs on the board and Ethernet connector at the moment of power plugging in. Will they blink somehow? If it is not a hardware problem, may be it is a software one?
Also I'd check the turn on with different board temperature. In case of different dielectric of C204 capacitor, the capacitance drift from voltage and temperature may significantly decrease the pulse length. In conjunction with unknown TdbPLUG_IN higher limit, turn on pulse may be shorter than actual TdbPLUG_IN time.

In other words, the design seems to me OK but two things need to be ensured: 1) C204 should be X5R or X7R dielectric, not Y5V/Z5U. 2) TdbPLUG_IN top limit should be defined by Rockchip as an IC manufacturer (may be I just don't have last RK808 datasheet).

[TheWiley]

I played a bit with this today and found that the auto-power-on seems to work less than 1/3rd of the time for me. When it does work, I do see the ethernet NIC lights come on immediately and the board starts responding to ping 30-60 seconds later. When it doesn't work, I get no NIC lights. I think that means I have a hardware problem?

If it makes a difference, I'm running with the 12v/5A PSU.

It also seems to matter how long the power is off - I haven't checked systematically, but auto-start seems to be more reliable if the board is unplugged for more than a minute or so. Does that make sense?

[neilman]

I played a bit with this today and found that the auto-power-on seems to work less than 1/3rd of the time for me. When it does work, I do see the ethernet NIC lights come on immediately and the board starts responding to ping 30-60 seconds later. When it doesn't work, I get no NIC lights. I think that means I have a hardware problem?

If it makes a difference, I'm running with the 12v/5A PSU.

It also seems to matter how long the power is off - I haven't checked systematically, but auto-start seems to be more reliable if the board is unplugged for more than a minute or so. Does that make sense?

It might be because short power break doesn't lower PSU output to zero - there is some residual voltage so power resumption causes, say, a 9v pulse on power restore.
A longer power break makers the output fall to zero so, power resumption creates the full 12v rise.
Possibly?