Bootable 64-bit RPi3 Gentoo image (OpenRC/Xfce/VC4) UPDATED

Sakaki · Guru Joined: 21 May 2014 Posts: 409

Zucca,

one additional thing while I remember - in addition to changing /etc/fstab, you will also need to modify /boot/cmdline.txt (i.e. the file cmdline.txt, located at the top level of the microSD card's first partition's filesystem), as this currently specifies root=PARTUUID=<...> (the UUID of the second partition of the microSD card).

It's also possible to boot directly from the USB incidentally, no microSD card required (but try this at your own risk!)
_________________
Regards,

sakaki

Zucca · Posted: Sat Sep 02, 2017 1:21 pm Post subject:

Sakaki · Guru Joined: 21 May 2014 Posts: 409

Zucca · Posted: Sat Sep 02, 2017 2:52 pm Post subject:

Sakaki · Guru Joined: 21 May 2014 Posts: 409

Zucca · Posted: Sat Sep 02, 2017 4:48 pm Post subject:

Sakaki · Guru Joined: 21 May 2014 Posts: 409

roylongbottom · n00b Joined: 13 Feb 2017 Posts: 64 Location: Essex, UK

[quote="Zucca"]

Zucca · Posted: Sat Sep 02, 2017 6:35 pm Post subject:

NeddySeagoon · Posted: Sat Sep 02, 2017 9:33 pm Post subject:

Zucca,

The Red led goes out when the Pi is under voltage, so flashing is a bad sign.
The lightning symbol is a 5v over current indicator. I've only seen it once, when I tried to power a SSD from the Pi USB system.
_________________
Regards,

NeddySeagoon

Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail.

Zucca · Posted: Sun Sep 03, 2017 3:18 pm Post subject:

Mates,

The problem I had was an user error. Both of those power supplies were insuficent for the Pi. 2.0A and 2.3A. For some reason I remembered that both were capable of powering Pi.
So I xzcat the image to one spare 16GB µSD. The Pi powered up ok. But the lightning logo kept appearing frequently even when there was no USB devices plugged in. The red led was almost constantly on. This bothers me, since Raspberry Pi should be pulling most of 1A when there's no other devices attached. Can anyone confirm this?

I'll start looking for proper power sources in hopes I have some luck with them. I have some old USB hub power sources, which deliver 5DCV. I think I'll have to do some soldering then. I might eventually hook a female µUSB port directly to the pins (I have some with mounting holes lying around). That way I can free the OTG port.
_________________
..: Zucca :..

dr_wulsen · Posted: Sun Sep 03, 2017 5:20 pm Post subject:

Hi Zucca,

I can confirm that my Pi3 runs fine on a 2.5A supply, Overclocked to 1,3GHz and compiling the whole system without issues.
No red LED or whatsoever. Running idle I'd expect a 2.3 to do the job.

Important thing (no, I'm not esoteric, I'm a plain mechatronics guy) a proper cable between the supply and the Pi. Enough wire gauge/cross-sectional area and the shorter the better in terms of voltage drop on the cable.
Not a Fanboy of specific brands, I use USB cables from Anker, which have a good cross-section and 90cm length.

I know it's possible to power the Pi from the USB outlets (those Y-spliced cables for USB HDD are great for it). But: the micro USB inlet has a fuse that cuts in case of wiring error etc. using the outlets or GPIO to apply power to your Pi bypasses the fuse and you can make a beautiful fried pi in case of issues.
_________________
There's no stupid questions, only stupid answers.

Zucca · Posted: Sun Sep 03, 2017 5:45 pm Post subject:

Zucca · Posted: Sun Sep 03, 2017 7:16 pm Post subject:

I found some parts from the attic:

USB hub power supply: 2.5A@5DCV
a 3A fuse with holder
cables and connectors

When in use the voltages are between 4.99 to 5.05. Should be whiting spec. The red led it lit almost all the time. Sometimes it turns off for a sec or so. The lightning logo however isn't in sync with the red led. It does not appear on the screen as frequently as the red led is lit.
Strange...

Next I'll measure the current draw... I'll expect it to be ok, since nothing is really heating up...

EDIT: Preliminary measurements:

max: ~500mA
idle 280-300mA
after shutdown: 96mA

To me it seems that the idle current draw is bit too high...

A strange side note; while measuring the current the red led was less often lit. Maybe I need to add a filter cap to make sure the voltage stays stable...
_________________
..: Zucca :..

dr_wulsen · Posted: Sun Sep 03, 2017 8:40 pm Post subject:

Terve,

I expect the Pi is clean and dry and shows the same behaviour on an isolating surface?

I'm thinking of a creeping current on the measurement side, that would mess up the data it receives.
A damaged/scratched PCB or bad soldering spot could do the same.

96mA/500mW seems high to me too.
on the other hand i found this: https://www.raspberrypi.org/forums/viewtopic.php?f=63&t=6050&start=75
look for shutdown - h
they say the lan chip draws quite much and their measurements are close to yours.
_________________
There's no stupid questions, only stupid answers.

NeddySeagoon · Posted: Sun Sep 03, 2017 9:02 pm Post subject:

Zucca,

The HDMI output is a load on the 5v. You can alter the drive strengths in config.txt.

Don't bother with a fuse. Its a source of volt drop and too slow acting to protect the Pi. Just don't mess up.
If you have a lab PSU set its current limit to about 5A. That will be plenty.

Not all SD cards are created equally. Power load is proportional to speed.
SD cards that are capable of higher speeds tend to need more power anyway.

Be careful of mobile phone chargers. They are often only 4.7v as they are intended to charge Li-Ion batteries.
They may well be marked 5.0v though.
_________________
Regards,

NeddySeagoon

Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail.

Zucca · Posted: Sun Sep 03, 2017 9:19 pm Post subject:

Ok. I went back to power the Pi using my most powerful USB brick with 2.3A. Now the red led barely flashes, but the lightning logo still appears quite often.
I hooked my multimeter to the 5V GPIO pin and to one of the GNDs to measure the voltage on board. Once I saw the voltage drop to 4.98V. Otherwise it was 5.10V to 5.12V. Then I started four cat /dev zero > /dev/null -processes to roughly simulate full load on CPUs. Voltage was exactly 5.00V.

After reading some of the forums posts on the Raspberry Pi site it seems that the current draw and voltages are ok.
_________________
..: Zucca :..

NeddySeagoon · Posted: Sun Sep 03, 2017 9:26 pm Post subject:

Zucca,

A multimeter will measure the average voltage. The voltage needs to be kept in spec under transient conditions too.
That needs an oscilloscope to make the measurement.

e.g. The voltage could be going between 4v and 6v and sill read 5v on a multimeter.
_________________
Regards,

NeddySeagoon

Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail.

Zucca · Posted: Sun Sep 03, 2017 10:32 pm Post subject:

NeddySeagoon · Posted: Sun Sep 03, 2017 10:57 pm Post subject:

Zucca,

Correct, A CMOS logic gate is like a very small capacitor.
Every time it switches it either charges or discharges and a small amount of charge is transferred across the power terminals.
Switching cycles per second is current flow (charge * time)

The more gates switching, the higher the current.

Power is current * voltage.

Beware multimeters on low AC voltages. How well they read depends on the waveform.
They are usually calibrated for a sinewave. They may also apply the input voltage to a diode, so won't read anything useful under 1v.
You won't need a very large capacitor. Its impedance needs to be low compared to the input impedance of your meter at the frequency of interest, since the capacitor and meter form a high pass filter.
_________________
Regards,

NeddySeagoon

Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail.

Zucca · Posted: Sun Sep 03, 2017 11:57 pm Post subject:

dr_wulsen · Posted: Mon Sep 04, 2017 6:23 am Post subject:

If it is labelled "true RMS" (root mean square) it should do for quite some deviation from sine too, but as neddy said, an oscilloscope will give you the most proper reading, in combination with a precision shunt resistor.

The power draw on the Pi is not so high, so a cap of in the thousands of microfarad will do.
Also the current fluctuations are not so high, so a standard type will. do the job.

I remember the many blown-up cas on early Pentium4 boards which were not low-ESR and died because of too quick changes in current (100A/microsecond). But the Pi is not such a monster, a standard cap should do here.

I'd use one electrolytic cap for buffering and one ceramic one (47-100nF) for filtering.
_________________
There's no stupid questions, only stupid answers.

roylongbottom · n00b Joined: 13 Feb 2017 Posts: 64 Location: Essex, UK

Probably for a completely different problem, there were protracted messages regarding power supplies causing that rainbow coloured square whilst running my stress tests on a Raspberry Pi 2. I bought a DROK Digital Ampere Voltage Multimeter and that did not identify any power issues. Also, there was no difference in the frequency of failures using 1.5 amp and 2.5 amp power supplies. Extensive testing revealed that the problem was caused by overheating, particularly associated with installation of a new OpenGL GLUT driver.

The same problem occurred with my Raspbery Pi 3, the rainbow display appearing after a few minutes of stress testing. This time I did not disconnect power immediately and noted that all programs, including an OpenGL test, logged correct termination. Continued running with a faulty display suggests that there was nothing wrong with power supplies.

I installed a newer version of Raspbian and no display failures occurred. Higher temperatures, up to 85.4°C were recorded, but probably more effective CPU throttling prevented the failures. The high temperature issue was fixed by using a FLIRC case.

In your case, perhaps there are other things to consider besides power supplies.
_________________
Regards

Roy

Zucca · Posted: Mon Sep 04, 2017 1:33 pm Post subject:

I currently use one USB 2.3A power brick and a thick cable. The lightning logo appears less frequently and the red led stays lit except for few flashes during boot and shutdown. I have haven't yet placed any caps for filtering. I'll evetually put caps when I install my Pi into a bigger box.

I wanted to boot my SDcard + USB stick Gentoo now, but it seems that only the bootloader runs and then the operation stalles. Blank screen, but the signal is there.
_________________
..: Zucca :..