possible SATA brownout?

[Sam Gordon]

Hi there,

Cubieboard rev. 12-09-09 here. Was working fine using BerryBoot and the latest Linaro with sata port and 2.5 inch laptop drive. Hard drive was mounted properly and was providing storage for a small file sharing server for my local network as a test to see if the cubieboard could meet my throughput needs.

A week has passed, and now the board is seemingly unable to consistently power the hard disk.

Audibly, I can hear the head crashing on the hard drive repeatedly. The head crashes every time the hard disk begins to spool, making me believe that the circuit that drives the hard disk is experiencing some form of brownout.

1. position hard disk head
2. spool hard disk
3. energy consumption for head positioning & disk spool too high
4. disk spool fails
5. circuit resets, crashes head to disk
6. rinse & repeat until power down

I am using a USB power consumption meter (which i'll use to post about measured power consumption of the board in various tasks later on..). Under normal operation with the SATA disk spooled I was seeing anywhere from 820-1050mW during normal usage. Now, with the HDD clicking I see it idling at about 650-720. Occasionally the clicking will stop for a second (indicating that the hard disk perhaps successfully spooled). During these times the power consumption skyrockets to the normal value of 820-1050. However, these times are short lived, and fail back into clicking eventually. Once the clicking begins again the current values drop back down to the 650-720 range.

As far as symptoms from a usability standpoint, if I hear the clicking, the CubieBoard won't see the disk. Anywhere. If I attempt to reboot the OS in that state, it will hang at boot time complaining about the SATA link taking too long to respond. It tries repeatedly to reset the link, but to no avail. It does this in PuppyLinux and Linaro. In Android 4.0 it simply does not recognize the drive (although it does make it click), however it allows the board to boot.

A few things

• I've tried 3 hard disks. 1 being an SSD that boasts low power consumption. They all share the same symptoms (well, of course the SSD doesn't click..)
  
• I don't have another power supply handy, I do however have a multimeter, if you can tell me how I may use it to diagnose the power circuitry.
• I would suspect user configuration if this exact setup wasn't working for me previously. Unfortunately, I'm not sure what to try and look at next as far as software goes. The drive is definitely powered, I'm just unsure as to what extent.
  

Any input would be appreciated, Thanks!

[regr...@shutko.ru]

I noticed several powerdowns on SSD write tests (power from usb). But they gone when I powered board from 5v 2A source.

[Martin Wild]

Hi,

ive had similar issues (with the harddisk ,not the brownout) appearing after a few days usage and  heavy compiling .
take a look with dmesg and see what the driver tells you! in my case the sata link got lost and reset over and over again and slowed down over time down to ata 33...
also the filesystem got heavily damaged ,but not the hdd itself.
the cause was a faulty 5V psu breaking down under 24/7 usage after some days. then i bought a new (better quality) psu with 5V 2.5A and no problems since then.
it could be the same problem  in your case when you say your measured power output is lower than before.

Regards, martin

[Bastiaan van den Berg]

Powersupply is very important when drawing this much power. 

You could measure the voltage between the +5V and GND while using sata to see if any voltage drop occurs.

Such a drop would cause the problems you describe.

--

buZz

[Sam Gordon]

Update : It was either the USB power consumption gauge that has somehow become faulty, or the CubieBoard has somehow failed in a way that makes it more power hungry, to the point where it ran outside the power limits of the measurement device. I believe it to be the gauge, as that seems more likely a failure point to me given the 5/10 build quality. The symptoms disappear once I remove the meter from the equation. I will replace the device with a good old multimeter and start from scratch. At least the problem was fixed!

It was indeed a brown out of sorts. A powered USB consumption meter shouldn't add to the overhead of running the device whatsoever, and it is well within specs for measurement (5V/2A), but it was definitely the culprit.

Also, it's interesting to note that the circuit that drives the hard disk is isolated sufficiently from the rest of the hardware to where a hard disk brownout does not necessarily affect the system stability (unless of course you're using the hard disk..) My system happily stayed afloat during such times. Perhaps it's because the rest of the hardware is so power efficient.

Thanks for the input, all.

[ne...@radempa.de]

Another experience for those who have similar problems:

When connecting a 2,5'' HDD (http://www.hgst.com/travelstar-5k1000) to a 2012-09-09 cubieboard the disk (which worked quite silent on another sata host) made clack sounds (in intervals around 1s) and also scratching sounds. (According to S.M.A.R.T. it is however still ok.) The reason was a reset (cf. below), and I guessed due to a lack of power. So I tried with a 5V 3.5A power supply, but got the same problem. When using two power supplies (with comon ground of course), one 5V 2A for the board and the other 5V 3.5A for the HDD the problem disappeared; also when I used the 2A supply for the HDD and the other for the board. So I guess 5V 4A should suffice, and I've ordered such a power supply.

However, I guess that even with the 4A power supply I will have to connect the HDD separately, because after several tries the HDD now does no more spin up (nor clack or scratch) when I connect its power cable to the board. So much about the brownout.

ibu


ata1.00: exception Emask 0x0 SAct 0xc SErr 0x0 action 0x6 frozen
ata1.00: failed command: WRITE FPDMA QUEUED
ata1.00: cmd 61/00:10:00:79:82/04:00:0a:00:00/40 tag 2 ncq 524288 out
         res 40/00:00:00:00:00/00:00:00:00:00/00 Emask 0x4 (timeout)
ata1.00: status { DRDY }
ata1.00: exception Emask 0x0 SAct 0xc SErr 0x0 action 0x6 frozen
ata1.00: failed command: WRITE FPDMA QUEUED
ata1.00: cmd 61/00:10:00:79:82/04:00:0a:00:00/40 tag 2 ncq 524288 out
         res 40/00:00:00:00:00/00:00:00:00:00/00 Emask 0x4 (timeout)
ata1.00: status { DRDY }
ata1: hard resetting link
ata1: link is slow to respond, please be patient (ready=0)
ata1: COMRESET failed (errno=-16)
ata1: hard resetting link
ata1: link is slow to respond, please be patient (ready=0)
ata1: COMRESET failed (errno=-16)
ata1: hard resetting link
ata1: link is slow to respond, please be patient (ready=0)
ata1: SATA link up 3.0 Gbps (SStatus 123 SControl 300)
ata1.00: configured for UDMA/133
ata1.00: device reported invalid CHS sector 0
ata1.00: device reported invalid CHS sector 0
ata1: EH complete

IMHO the 0x6 in the first line means it was reset, cf. source:
ATA_EH_SOFTRESET        = (1 << 1), /* meaningful only in ->prereset */
ATA_EH_HARDRESET        = (1 << 2), /* meaningful only in ->prereset */

[Mateusz Mateusz]

Same problem here. I'm powering Cubie with 5V 3A and I connected 3,5 hdd which is powered with 12V 1,5A and 5V from Cubie. Results:

[   22.720000] ata1.00: exception Emask 0x10 SAct 0x2 SErr 0x280100 action 0x6 frozen
[   22.740000] ata1.00: irq_stat 0x08000000, interface fatal error
[   22.740000] ata1: SError: { UnrecovData 10B8B BadCRC }
[   22.770000] ata1.00: failed command: READ FPDMA QUEUED
[   22.770000] ata1.00: cmd 60/80:08:58:b3:06/00:00:00:00:00/40 tag 1 ncq 65536 in
[   22.770000]          res 40/00:08:58:b3:06/00:00:00:00:00/40 Emask 0x10 (ATA bus error)
[   22.800000] ata1.00: status: { DRDY }
[   22.800000] ata1: hard resetting link
[   23.150000] ata1: SATA link up 3.0 Gbps (SStatus 123 SControl 300)
[   23.240000] ata1.00: configured for UDMA/133
[   23.240000] ata1: EH complete

same situation is with 2,5 hdd powering directly from cubieboard

dmesg: http://pastebin.com/2aEXPhpU , sometimes there are more such errors

Have you testes 4A power supply?

[Clinton Lee Taylor]

Greetings ...

 Just jumping in with some silly ideas.

 You sure the 2.5" HDD only needs 1A, believe that the CubieBoards needs at least .5A to run and believe that the DC plug will only support 2A. Looking at the few 2.5" HDD's I have around, they mostly list 5V 1.0A, but that is just a label and who knows what they draw.  If the HDD needs more than 1.5A, even for just spin up or something, it might be maxing the power need that the Cubie can provide, no matter how big your PSU is.

 I have tested a a few 2.5", with external power for the HDD, but have not yet found a PSU, that will power Cubie and HDD from a single source, but I think that is a problem with lots of over rated PSU devices, but I don't have a good inline meter to know for sure.  When I track down a 5V2A PSU that connections directly to the DC socket without converts and other stuff, I might know better for myself.

 The BaseBoard and HDD bay that other people are making might shed some light on this issue, but this is still a dev board, basically show casing all the funky things the A10 can do.

 Would be nice to be able to put these kinds of issues to rest with better data.

Happy Hunting

Mailed

LeeT

[Lawrence Sheed]

I've done some basic power draw tests in this set.
http://www.flickr.com/photos/sheedl/sets/72157632648336414/

Naked cubie pulls about .12A @ 5A.
With ethernet, and a usb plug,
http://www.flickr.com/photos/sheedl/8453464678/in/set-72157632648336414/

SATA HDD sits at about .6A with ethernet, and usb at cpu load.
Peak SATA is about 1A on spinup which brings load to about 1.5A on
spinup for a 1/2 second or so.

There are a few photos with description of whats plugged in with load.
If you need more, let me know.

Lawrence.

> --
> You received this message because you are subscribed to the Google
> Groups "Cubieboard" group.
> To unsubscribe from this group and stop receiving emails from it,
> send an email to cubieboard+...@googlegroups.com.
> Visit this group at http://groups.google.com/group/cubieboard?hl=en
> [1].
> For more options, visit https://groups.google.com/groups/opt_out
> [2].
>
>
>
> Links:
> ------
> [1] http://groups.google.com/group/cubieboard?hl=en
> [2] https://groups.google.com/groups/opt_out

--
LAWRENCE SHEED

COMPUTER SOLUTIONS
Room 401-402 Han Wen Xuan Building 2
No.14, 955 Yan An Middle Road
Shanghai China 200040

Tel: +86 21 62890765
+86 21 62890056
Fax: +86 21 62890700
Mobile: +86 13901 802 269
Skype: computersolutions.cn
Web: www.computersolutions.cn
E-mail: sup...@computersolutions.cn

[C.Lee Taylor]

Mmm, ok, then I am out of ideas ... There seems to be plenty of power
still available ... Bugger.

Nice rig!

> Visit this group at http://groups.google.com/group/cubieboard?hl=en.
> For more options, visit https://groups.google.com/groups/opt_out.
>
>

[Deksanba]

Hmmm I m wondering this as well, seems the power supply isn't really the source of the problem here. I used a WD5000BPVT. Power specifications says 0.55A so I was hoping that with a 5v / 2a power supply i d be good. But Sadly it's not enough. I m wondering what I could use to power the hard disk separately ? 

This might be of any help as well :

https://groups.google.com/forum/#!topic/cubieboard/LNLMcqKx_-E 

[Deksanba]

Could you try to check the load if you copy lot of files over ethernet (100mb) on the hard disk, this is where I get some resets and hangs sometimes ? Maybe the ethernet traffic + load on cpu + hard disk working is too much for 2A ?

Thanks a lot.

[Mateusz Mateusz]

Good solution is to use external hd case with eSata interface. eSata is 1:1 compatible with sata. There are plenty of cases for 2,5 and 3,5 drives.

I connected 3,5 drive with separate +12V and it seems to be working fine so far.

[theRat]

I don't know where you are getting 0.55A from as the WDC datasheet specifies max power at 0.95A +-10% so more than 1A is a possible draw from that hard disk.  See the datasheet:

http://www.wdc.com/wdproducts/library/SpecSheet/ENG/2879-701278.pdf

The other thing to remember is not all 5VDC/2A power supplies are created equal.  Most of the cheap ones have very poor regulation and will cause issues when a device suddenly asks for an extra 0.5A.  These cheap ones aren't designed for rapidly changing loads and have very small output capacitors to save money and space.

[Deksanba]

Maybe that's the problem, any link with a good and reliable power supply I can order online ? 

Thanks.

[Mateusz Mateusz]

I do not understand.. I connected hard drive to PC PSU ( bequiet 430W ), cubieboard to 3A 5V power supply and what I saw? same errrorrrrrrrrrrrrrrs...........

Mr Cubie what's going on? is my board broken?