Odroid-H2 weak HDD power supply

Post Reply
Peedy2495
Posts: 4
Joined: Thu Jan 03, 2019 8:53 pm
languages_spoken: english
ODROIDs: H2
Contact:

Odroid-H2 weak HDD power supply

Unread post by Peedy2495 » Thu Jan 03, 2019 9:24 pm

I'm using two 3,5" HDDs (HGST HUS724030ALA640)
It's impossible to use a softraid level 1 with mdraid!

Reason: by accessing both HDDs synchronous (moving the heads) a lack of power spinns both HDDs down for a second.
The same prob during boot initializing both HDDs via ioctl.

I'm using a 15V PSU with 5A (Toshiba PA3755U-1ACA).
4 Amps are recommended by documentation.

Has the PSU current limitations for both voltages to be able to buffer the peaks via capacities?
Any advice would be welcome.

User avatar
venkatbo
Posts: 494
Joined: Mon Feb 25, 2013 3:10 pm
languages_spoken: english
ODROIDs: C0/C1+/C2; U3+; XU3-Lite/XU4; H2; VU/VU7+; Touchscreen Show TFT Cloudshell SmartPower mAHRS DAC GPS Multiscope Ocam
Location: Bay Area, California
Contact:

Re: Odroid-H2 weak HDD power supply

Unread post by venkatbo » Thu Jan 03, 2019 9:46 pm

If you are suspecting that the onboard circuitry is unable to supply the needed (peak or otherwise) power, is it possible to have the psu supply power directly to the sata drives?

Peedy2495
Posts: 4
Joined: Thu Jan 03, 2019 8:53 pm
languages_spoken: english
ODROIDs: H2
Contact:

Re: Odroid-H2 weak HDD power supply

Unread post by Peedy2495 » Thu Jan 03, 2019 10:01 pm

Well, this not a preferred solution.
& really bad for my planned housing ... :-(

Is it possible to get a excerpt from the psu-schematic?
As a HAM-OM I'm possibly able to modify the supply by adding buffers in the power-cord.
But I have to know the schematic to determine the limits of the PSU.

Cheers & 73

User avatar
odroid
Site Admin
Posts: 29994
Joined: Fri Feb 22, 2013 11:14 pm
languages_spoken: English
ODROIDs: ODROID
Contact:

Re: Odroid-H2 weak HDD power supply

Unread post by odroid » Fri Jan 04, 2019 9:12 am

You can find the schematics in our WiKi pages.
https://wiki.odroid.com/odroid-h2/hardw ... datasheets

What is the peak current of your HDD?
We've tested a few different two HDDs with the Case Type-1 and there was no issue so far.
Or you can try a 19~20V/4A PSU if your current PSU output peak current is not sufficient.

Peedy2495
Posts: 4
Joined: Thu Jan 03, 2019 8:53 pm
languages_spoken: english
ODROIDs: H2
Contact:

Re: Odroid-H2 weak HDD power supply

Unread post by Peedy2495 » Fri Jan 04, 2019 6:41 pm

Now, I tryied a Source with 19V/4.22A.
The same prob, but doesn't sound as massive as with 15V.
According do the overvoltage-detection including the tolerance of BZ84-B22, the max. input voltage is about 21.6V
The step-down-converter will be destroyed by adding a high buffer cappacity and a current limitation by inductivity will decrease/egalize the buffer-effect :-(

Any hints?

colink
Posts: 4
Joined: Sat Jan 05, 2019 1:36 am
languages_spoken: english
ODROIDs: H2
Contact:

Re: Odroid-H2 weak HDD power supply

Unread post by colink » Sat Jan 05, 2019 3:18 am

I have the same issue with a pair of Toshiba MG05ACA800E 8Tb HDDs. The disks fail to initialise the SATA link at 6Gb/s. It renegotiates to 1.5Gb/s but the drives eventually go offline. Plugging the disks into a separate ATX power supply solves the problem. They connect to the H2 at 6Gbps and have run without problem. According to the disks' data sheet, random read power consumption is 11.4W so ~23W in total. From a cursory look at the H2 schematic, the 12V rail supplies only the SATA port and the regulator has a maximum output of 3A so this should easily deal with the disks' requirements. The issue may be with the 5V line as there's only 30W available. I'm using a 15V 4A supply and will try a 19V 6A supply once it arrives. If the H2 PSU is still unable to power the disks, I'll use a separate 'pico' ATX PSU for the SATA disks. It's a shame as a separate power adapter and ATX PSU will cost about $50 which is a big chunk of the H2's purchase price just to make it work as the spec says it should.

Cheers
Colin

User avatar
rooted
Posts: 6435
Joined: Fri Dec 19, 2014 9:12 am
languages_spoken: english
Location: Gulf of Mexico, US
Contact:

Re: Odroid-H2 weak HDD power supply

Unread post by rooted » Sat Jan 05, 2019 3:39 am

Are either of you using the power supply Hardkernel sells?

tkaiser
Posts: 671
Joined: Mon Nov 09, 2015 12:30 am
languages_spoken: english
ODROIDs: C1+, C2, XU4, HC1
Contact:

Re: Odroid-H2 weak HDD power supply

Unread post by tkaiser » Sat Jan 05, 2019 4:29 am

colink wrote:
Sat Jan 05, 2019 3:18 am
According to the disks' data sheet, random read power consumption is 11.4W so ~23W in total.
Is the data sheet also mentioning peak spinup consumption (usually a lot higher than worst case access consumption when platters already rotate since in the latter situation only the head movements cause additional consumption)? And funnily almost always people look just at amperage ratings but forget about voltage drops in peak load scenarios :)

BTW: Usually those newer Toshibas can be configured for staggered spinup via one pin of the SATA power connector.

tkaiser
Posts: 671
Joined: Mon Nov 09, 2015 12:30 am
languages_spoken: english
ODROIDs: C1+, C2, XU4, HC1
Contact:

Re: Odroid-H2 weak HDD power supply

Unread post by tkaiser » Sat Jan 05, 2019 5:34 am

colink wrote:
Sat Jan 05, 2019 3:18 am
The disks fail to initialise the SATA link at 6Gb/s. It renegotiates to 1.5Gb/s but the drives eventually go offline.
Out of curiosity: can you please share the relevant dmesg output from this to get an idea about the timing?

colink
Posts: 4
Joined: Sat Jan 05, 2019 1:36 am
languages_spoken: english
ODROIDs: H2
Contact:

Re: Odroid-H2 weak HDD power supply

Unread post by colink » Sun Jan 06, 2019 1:43 am

@rooted, I'm using the power adapter sold by Hardkernel's UK distributor who trade as Odroid UK. It's 15V 4A.

@tkaiser - Toshiba's datasheet does not mention peak power/current. It's only spec is as regards voltage (12V +/-10% and 5V +/- 5%). I might put a scope on the 5V and 12V rails to measure the voltage drop on spin-up using the on-board supply but not until next weekend.

Full dmesg outputs are at https://gist.github.com/colinkn/899d3d8 ... 772ed0534e.

- With an external PSU powering the disks, the drives are brought up between 2.394s and 2.558s. The delay before 20s, by the way, is for me to key in the decryption password for the HDDs.

- With the on-board power supply, ata negotiation starts 2.394s between 2.61s at 6Gbps. ATA bus errors then start as soon as the rootfs (on nvme) decryption password has been entered at 21.427s and the hdd decryption starts. The disks eventually came online at 1.5Gbps at 77s and 84s respectively and the md-raid 1 device they provide was detected. The automated decryption of the hdds was not successful so I manually decrypted them at 298s. The md device then goes offline with a disk failure. The ata links again go down and finally come up again, one at 1.5Gbps and the other at 6Gbps but with a different device node. I rebooted at this point.

Other points - the disks make intermittent noises as they fail to spin up. BIOS POST is also much longer with the on-board PSU - sometimes minutes longer. The board works like a dream with the HDDs powered externally. Here's hoping a 19V 120W supply will do the job.

Yes, the Toshibas support Pin 11 power-up in standby (PUIS) - enabling staggered spin-up in the BIOS made no difference so perhaps the ODROID hardware doesn't support this? I'm reluctant to enable PUIS manually with hdparm -s1 in case I can't get the disks back on line as I understand BIOS support (to recognise disks in standby) is required. I also wonder whether it's just spin-up that's a problem given the continued bus errors observed with the on-board power supply to the disks once they've come online.

Any other thoughts and ideas welcome!!

mars24568
Posts: 2
Joined: Mon Jan 07, 2019 5:59 am
languages_spoken: english, german
ODROIDs: 3xHC1, 1xH2
Contact:

Re: Odroid-H2 weak HDD power supply

Unread post by mars24568 » Mon Jan 07, 2019 6:28 am

I'm having the same issue - two Seagate 4TB Enterprise Disks (ST4000NM0035) together with the original power supply and power cables sold from Hardkernel.

The disks start running ok, but after some minutes to hours i get entries like below in the log file:

Code: Select all

Dec 14 22:12:56 server kernel: [16212.811154] ata1: hard resetting link
Dec 14 22:13:06 server kernel: [16222.814149] ata1: hard resetting link
Dec 14 22:13:09 server kernel: [16225.306315] ata1: SATA link up 6.0 Gbps (SStatus 133 SControl 300)
Dec 14 22:13:09 server kernel: [16225.310424] ata1.00: configured for UDMA/133
...
Dec 15 04:47:25 server kernel: [39879.697353] ata1: SATA link down (SStatus 0 SControl 300)
Dec 15 04:47:25 server kernel: [39879.697371] ata1: limiting SATA link speed to 1.5 Gbps
Dec 15 04:47:25 server kernel: [39879.707881] ata1: hard resetting link
Dec 15 04:47:26 server kernel: [39880.741259] ata1: SATA link down (SStatus 0 SControl 310)
Dec 15 04:47:26 server kernel: [39880.741276] ata1.00: disabled
Dec 15 04:47:26 server kernel: [39880.741313] sd 0:0:0:0: [sda] tag#1 FAILED Result: hostbyte=DID_OK driverbyte=DRIVER_SENSE
Dec 15 04:47:26 server kernel: [39880.741319] sd 0:0:0:0: [sda] tag#1 Sense Key : Illegal Request [current]
Dec 15 04:47:26 server kernel: [39880.741325] sd 0:0:0:0: [sda] tag#1 Add. Sense: Unaligned write command
Dec 15 04:47:26 server kernel: [39880.741333] sd 0:0:0:0: [sda] tag#1 CDB: Synchronize Cache(10) 35 00 00 00 00 00 00 00 00 00
However, i borrowed an old power supply from a PC which i now use to power the disks directly. They are running ok now since Dec. 21st., so this seems to be ok.

Of course this is not a nice solution - could HARDKERNEL please give some hints if these issues are caused by the power supply, board layout, power cabling, or whatever...

Markus

tkaiser
Posts: 671
Joined: Mon Nov 09, 2015 12:30 am
languages_spoken: english
ODROIDs: C1+, C2, XU4, HC1
Contact:

Re: Odroid-H2 weak HDD power supply

Unread post by tkaiser » Mon Jan 07, 2019 7:54 am

Quick summary: problems occurred with pairs of HGST, Toshiba and Seagate HDDs while https://wiki.odroid.com/odroid-h2/appli ... ata_drives might suggest Hardkernel only tested with WD Red?

Anyway: when we all troubleshooted various XU4 'UAS troubles' caused by undervoltage (just stumbled across this nice example viewtopic.php?f=99&t=28076) @odroid reported that some brands are more prone to undervoltage than others (some HDD fine with only 4.5V on the 5V rail, some already starting to fail at 4.75V -- but IIRC back then only 2.5" HDD were tested)

allanallanf
Posts: 2
Joined: Thu Jan 03, 2019 11:30 pm
languages_spoken: english
Contact:

Re: Odroid-H2 weak HDD power supply

Unread post by allanallanf » Mon Jan 07, 2019 10:08 am

Argh, this thread has put me off buying one now. Luckily they are out of stock - it's the only reason I haven't bought one already.

Hopefully they can address/fix this this issue before the units are back in stock.

tkaiser
Posts: 671
Joined: Mon Nov 09, 2015 12:30 am
languages_spoken: english
ODROIDs: C1+, C2, XU4, HC1
Contact:

Re: Odroid-H2 weak HDD power supply

Unread post by tkaiser » Mon Jan 07, 2019 3:47 pm

mars24568 wrote:
Mon Jan 07, 2019 6:28 am
The disks start running ok, but after some minutes to hours i get entries like below in the log file
Are you able to conduct a test to see whether excessive head movements really trigger the behavior (should be rather simple on Linux using the iozone3 package and then running a quick random IO test).

I'm thinking about the symtoms described here. Some users report spinup problems while you clearly say that isn't an issue. The motors that rotate the platters are fed by the 12V rail (so a potential current limitation could be a problem or maybe a voltage drop too?). But there's 5V used to feed the drive's controller (which might reset with voltage drops on the 5V rail) and there are the motors that move the arms. No idea whether those are fed by the 12V or 5V rails and whether this differentiates by vendor.

mars24568
Posts: 2
Joined: Mon Jan 07, 2019 5:59 am
languages_spoken: english, german
ODROIDs: 3xHC1, 1xH2
Contact:

Re: Odroid-H2 weak HDD power supply

Unread post by mars24568 » Tue Jan 08, 2019 5:48 am

tkaiser wrote:
Mon Jan 07, 2019 3:47 pm
Are you able to conduct a test to see whether excessive head movements really trigger the behavior (should be rather simple on Linux using the iozone3 package and then running a quick random IO test).
Sure. The disks fail almost instantly after starting the test (see logfile below). Maybe I should note that the root-fs is on an emmc-Card. So during bootup the SATA disks are not really under stress - they are just mounted as a -clean- RAID1-Array.

When powered by an external PSU the disks don't fail during iozone -a:

Code: Select all

Jan 07 20:54:23 server root[1210]: IOzone Test, disks powered by separate PSU
...[no relevant information here]
Jan 07 21:01:37 server root[1245]: IOzone Test completed
When powered by the H2 PSU the disks boot up without problems, but fail only 3 seconds after starting iozone -a:

Code: Select all

Jan  7 21:19:16 server kernel: [    1.680254] ata2: SATA link up 6.0 Gbps (SStatus 133 SControl 300)
Jan  7 21:19:16 server kernel: [    1.686734] ata2.00: ATA-10: ST4000NM0035-1V4107, TN04, max UDMA/133
Jan  7 21:19:16 server kernel: [    1.686738] ata2.00: 7814037168 sectors, multi 16: LBA48 NCQ (depth 31/32), AA
Jan  7 21:19:16 server kernel: [    1.688248] ata1: SATA link up 6.0 Gbps (SStatus 133 SControl 300)
Jan  7 21:19:16 server kernel: [    1.688304] ata2.00: configured for UDMA/133
Jan  7 21:19:16 server kernel: [    1.697272] ata1.00: ATA-10: ST4000NM0035-1V4107, TN04, max UDMA/133
Jan  7 21:19:16 server kernel: [    1.697277] ata1.00: 7814037168 sectors, multi 16: LBA48 NCQ (depth 31/32), AA
Jan  7 21:19:16 server kernel: [    1.698746] ata1.00: configured for UDMA/133
Jan  7 21:19:16 server kernel: [    1.699385] scsi 0:0:0:0: Direct-Access     ATA      ST4000NM0035-1V4 TN04 PQ: 0 ANSI: 5
Jan  7 21:19:16 server kernel: [    1.733824] scsi 1:0:0:0: Direct-Access     ATA      ST4000NM0035-1V4 TN04 PQ: 0 ANSI: 5
Jan  7 21:19:16 server kernel: [    1.772279] sd 0:0:0:0: [sda] 7814037168 512-byte logical blocks: (4.00 TB/3.64 TiB)
Jan  7 21:19:16 server kernel: [    1.772316] sd 1:0:0:0: [sdb] 7814037168 512-byte logical blocks: (4.00 TB/3.64 TiB)
Jan  7 21:19:16 server kernel: [    1.772413] sd 0:0:0:0: [sda] Write Protect is off
Jan  7 21:19:16 server kernel: [    1.772493] sd 1:0:0:0: [sdb] Write Protect is off
Jan  7 21:19:16 server kernel: [    1.772502] sd 0:0:0:0: [sda] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA
Jan  7 21:19:16 server kernel: [    1.772606] sd 1:0:0:0: [sdb] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA
Jan  7 21:19:16 server kernel: [    1.780547] sd 0:0:0:0: [sda] Attached SCSI disk
Jan  7 21:19:16 server kernel: [    1.818490] sd 1:0:0:0: [sdb] Attached SCSI disk
Jan  7 21:19:16 server kernel: [    1.860917] md: bind<sdb>
Jan  7 21:19:16 server kernel: [    1.862892] md: bind<sda>
Jan  7 21:19:16 server kernel: [    1.865830] md: raid1 personality registered for level 1
Jan  7 21:19:16 server kernel: [    1.866565] md/raid1:md127: active with 2 out of 2 mirrors
Jan  7 21:19:16 server kernel: [    1.866613] md127: detected capacity change from 0 to 4000785981440
...
Jan  7 21:21:26 server root: IOzone Test, disks powered by H2
Jan  7 21:21:29 server kernel: [  146.180827] ata1: hard resetting link
Jan  7 21:21:39 server kernel: [  156.183581] ata1: hard resetting link
Jan  7 21:21:40 server kernel: [  157.055098] ata1: SATA link up 6.0 Gbps (SStatus 133 SControl 300)
Jan  7 21:21:40 server kernel: [  157.059546] ata1.00: configured for UDMA/133
Jan  7 21:21:40 server kernel: [  157.059549] ata1.00: retrying FLUSH 0xea Emask 0x10
Jan  7 21:21:40 server kernel: [  157.059653] ata1: EH complete
Jan  7 21:21:49 server kernel: [  166.000684] ata2: hard resetting link
Jan  7 21:21:59 server kernel: [  176.002314] ata2: hard resetting link
Jan  7 21:22:00 server kernel: [  176.877055] ata2: SATA link up 6.0 Gbps (SStatus 133 SControl 300)
Jan  7 21:22:00 server kernel: [  176.881515] ata2.00: configured for UDMA/133
Jan  7 21:22:00 server kernel: [  176.881518] ata2.00: retrying FLUSH 0xea Emask 0x50
Jan  7 21:22:00 server kernel: [  176.881617] ata2: EH complete
Jan  7 21:22:09 server kernel: [  186.130059] ata2: hard resetting link
Jan  7 21:22:19 server kernel: [  196.132038] ata2: hard resetting link
Jan  7 21:22:20 server kernel: [  197.003669] ata2: SATA link up 6.0 Gbps (SStatus 133 SControl 300)
Jan  7 21:22:20 server kernel: [  197.007883] ata2.00: configured for UDMA/133
Jan  7 21:22:20 server kernel: [  197.007886] ata2.00: retrying FLUSH 0xea Emask 0x50
Jan  7 21:22:20 server kernel: [  197.007981] ata2: EH complete
Jan  7 21:22:20 server kernel: [  197.403599] ata2: hard resetting link
Jan  7 21:22:30 server kernel: [  207.408329] ata2: hard resetting link
Jan  7 21:22:31 server kernel: [  208.283122] ata2: SATA link up 6.0 Gbps (SStatus 133 SControl 300)
Jan  7 21:22:31 server kernel: [  208.287326] ata2.00: configured for UDMA/133
Jan  7 21:22:31 server kernel: [  208.287330] ata2: EH complete
Jan  7 21:22:36 server kernel: [  213.428588] ata1: hard resetting link
Jan  7 21:22:46 server kernel: [  223.431963] ata1: hard resetting link
Jan  7 21:22:47 server kernel: [  224.306479] ata1: SATA link up 6.0 Gbps (SStatus 133 SControl 300)
Jan  7 21:22:47 server kernel: [  224.310801] ata1.00: configured for UDMA/133
Jan  7 21:22:47 server kernel: [  224.310804] ata1.00: retrying FLUSH 0xea Emask 0x50
Jan  7 21:22:47 server kernel: [  224.310937] ata1: EH complete
Jan  7 21:22:47 server kernel: [  224.526293] ata2: limiting SATA link speed to 3.0 Gbps
Jan  7 21:22:47 server kernel: [  224.546824] ata2: hard resetting link
Jan  7 21:22:57 server kernel: [  234.551479] ata2: hard resetting link
Jan  7 21:22:58 server kernel: [  235.426116] ata2: SATA link up 3.0 Gbps (SStatus 123 SControl 320)
Jan  7 21:22:58 server kernel: [  235.430692] ata2.00: configured for UDMA/133
Jan  7 21:22:58 server kernel: [  235.430720] ata2: EH complete
Jan  7 21:23:01 server kernel: [  237.980615] ata2: hard resetting link
Jan  7 21:23:01 server kernel: [  237.993823] ata1: hard resetting link
Jan  7 21:23:02 server kernel: [  239.402003] ata2: SATA link down (SStatus 0 SControl 320)
Jan  7 21:23:02 server kernel: [  239.431335] ata2: hard resetting link
Jan  7 21:23:04 server kernel: [  241.421955] ata2: SATA link down (SStatus 0 SControl 320)
Jan  7 21:23:04 server kernel: [  241.421973] ata2: limiting SATA link speed to 1.5 Gbps
Jan  7 21:23:04 server kernel: [  241.463196] ata2: hard resetting link
Jan  7 21:23:05 server kernel: [  242.493904] ata2: SATA link down (SStatus 0 SControl 310)
Jan  7 21:23:05 server kernel: [  242.493919] ata2.00: disabled
Jan  7 21:23:05 server kernel: [  242.493956] sd 1:0:0:0: [sdb] tag#9 FAILED Result: hostbyte=DID_OK driverbyte=DRIVER_SENSE
Jan  7 21:23:05 server kernel: [  242.493963] sd 1:0:0:0: [sdb] tag#9 Sense Key : Illegal Request [current]
Jan  7 21:23:05 server kernel: [  242.493969] sd 1:0:0:0: [sdb] tag#9 Add. Sense: Unaligned write command
Jan  7 21:23:05 server kernel: [  242.493977] sd 1:0:0:0: [sdb] tag#9 CDB: Synchronize Cache(10) 35 00 00 00 00 00 00 00 00 00
Jan  7 21:23:05 server kernel: [  242.495766] ata2: EH complete
Jan  7 21:23:05 server kernel: [  242.495775] ata2.00: detaching (SCSI 1:0:0:0)
Jan  7 21:23:05 server kernel: [  242.496616] sd 1:0:0:0: [sdb] Synchronizing SCSI cache
Jan  7 21:23:05 server kernel: [  242.496655] sd 1:0:0:0: [sdb] Synchronize Cache(10) failed: Result: hostbyte=DID_BAD_TARGET driverbyte=DRIVER_OK
Jan  7 21:23:05 server kernel: [  242.496656] sd 1:0:0:0: [sdb] Stopping disk
Jan  7 21:23:05 server kernel: [  242.496665] sd 1:0:0:0: [sdb] Start/Stop Unit failed: Result: hostbyte=DID_BAD_TARGET driverbyte=DRIVER_OK
Jan  7 21:23:05 server kernel: [  242.525780] ata2: hard resetting link
Jan  7 21:23:11 server kernel: [  247.998892] ata1: hard resetting link
Jan  7 21:23:12 server kernel: [  248.873499] ata1: SATA link up 6.0 Gbps (SStatus 133 SControl 300)
Jan  7 21:23:12 server kernel: [  248.877134] ata1.00: configured for UDMA/133
Jan  7 21:23:12 server kernel: [  248.877137] ata1.00: retrying FLUSH 0xea Emask 0x10
Jan  7 21:23:12 server kernel: [  248.877230] ata1: EH complete
Jan  7 21:23:12 server kernel: [  248.931284] md: unbind<sdb>
Jan  7 21:23:12 server kernel: [  248.949423] md: export_rdev(sdb)
Jan  7 21:23:12 server root: IOzone Test manually interrupted
Hope that helps...

Markus

User avatar
odroid
Site Admin
Posts: 29994
Joined: Fri Feb 22, 2013 11:14 pm
languages_spoken: English
ODROIDs: ODROID
Contact:

Re: Odroid-H2 weak HDD power supply

Unread post by odroid » Tue Jan 08, 2019 10:34 am

We will order some HDD samples to reproduce the issue.
Give us a week to share our test result.

tkaiser
Posts: 671
Joined: Mon Nov 09, 2015 12:30 am
languages_spoken: english
ODROIDs: C1+, C2, XU4, HC1
Contact:

Re: Odroid-H2 weak HDD power supply

Unread post by tkaiser » Tue Jan 08, 2019 4:28 pm

BTW: Since a few users report they're using mdraid's RAID-1 mode still in 2019 which is a pretty inefficient and anachronistic way to waste a whole disk for almost nothing just some food for thought: https://forum.openmediavault.org/index. ... post146935 (no discussion in this thread please, let's focus here on the HDD powering situation)

User avatar
odroid
Site Admin
Posts: 29994
Joined: Fri Feb 22, 2013 11:14 pm
languages_spoken: English
ODROIDs: ODROID
Contact:

Re: Odroid-H2 weak HDD power supply

Unread post by odroid » Wed Jan 09, 2019 10:51 am

We've ordered this item(2pcs) since we couldn't find any of three models in Korea.
http://a.co/d/f4F0uxy

It may take two weeks for delivery. :(

Peedy2495
Posts: 4
Joined: Thu Jan 03, 2019 8:53 pm
languages_spoken: english
ODROIDs: H2
Contact:

Re: Odroid-H2 weak HDD power supply

Unread post by Peedy2495 » Thu Jan 10, 2019 2:31 pm

@tkaiser
This thread belongs to an hardware issue.
Posts of methods, technologies, religions, aliens, etc. which doesn't belong to this thread are OFF TOPIC.
Almost everyone has good reasons for his project design.

Cheers ... Peedy

colink
Posts: 4
Joined: Sat Jan 05, 2019 1:36 am
languages_spoken: english
ODROIDs: H2
Contact:

Re: Odroid-H2 weak HDD power supply

Unread post by colink » Wed Jan 16, 2019 10:16 pm

I've now tested the H2 with a 19V/6.3A power supply. It exhibits exactly the same issues I reported above with dual enterprise HDDs (Toshibas in my case) - very long POST (the BIOS codes I saw displayed bottom right of the screen were B4 and A2), SATA link renegotiation down to 1.5Gbps and eventual loss of the SATA link. It seems the issue is with the 12V and/or 5V supplies to the SATA connectors and not with the power adapter plugged into the board. I didn't get anywhere trying to measure the voltage drop on them - the pins are too small to attach scope probes and I didn't want to go soldering the board.

tkaiser
Posts: 671
Joined: Mon Nov 09, 2015 12:30 am
languages_spoken: english
ODROIDs: C1+, C2, XU4, HC1
Contact:

Re: Odroid-H2 weak HDD power supply

Unread post by tkaiser » Thu Jan 17, 2019 1:40 am

colink wrote:
Wed Jan 16, 2019 10:16 pm
It seems the issue is with the 12V and/or 5V supplies to the SATA connectors
What kind of cables do you use? Did you already look into this thread: viewtopic.php?f=171&t=33057 ?

colink
Posts: 4
Joined: Sat Jan 05, 2019 1:36 am
languages_spoken: english
ODROIDs: H2
Contact:

Re: Odroid-H2 weak HDD power supply

Unread post by colink » Fri Jan 18, 2019 8:02 am

tkaiser wrote:
Thu Jan 17, 2019 1:40 am
What kind of cables do you use? Did you already look into this thread: viewtopic.php?f=171&t=33057 ?
I used the Hardkernel SATA power and data cable kit sold with the board. They're marked as 18 AWG so a voltage drop of ca. 0.005V per amp drawn across the 30 cm or so of cable (so well within ATX tolerances) - probably less given there are several conductors in parallel. The cables are unlikely to be the cause of the fault, wouldn't you say?

tkaiser
Posts: 671
Joined: Mon Nov 09, 2015 12:30 am
languages_spoken: english
ODROIDs: C1+, C2, XU4, HC1
Contact:

Re: Odroid-H2 weak HDD power supply

Unread post by tkaiser » Fri Jan 18, 2019 8:19 am

colink wrote:
Fri Jan 18, 2019 8:02 am
The cables are unlikely to be the cause of the fault, wouldn't you say?
Unlikely since you use Hardkernel's. But bad cables at least might have explained what happened with your disks in high load conditions...

User avatar
odroid
Site Admin
Posts: 29994
Joined: Fri Feb 22, 2013 11:14 pm
languages_spoken: English
ODROIDs: ODROID
Contact:

Re: Odroid-H2 weak HDD power supply

Unread post by odroid » Fri Feb 01, 2019 5:26 pm

Finally, we've received the HGST Ultrastar 7K4000 samples this morning.
We could reproduce the issue and we will try to find a solution several days later since long Lunar new year holiday starts today. :(
Sorry for the inconvenience caused.

User avatar
odroid
Site Admin
Posts: 29994
Joined: Fri Feb 22, 2013 11:14 pm
languages_spoken: English
ODROIDs: ODROID
Contact:

Re: Odroid-H2 weak HDD power supply

Unread post by odroid » Fri Feb 08, 2019 1:57 pm

We found a root cause.
There is a dedicated DCDC converter RT8296B for SATA 12Volt power supply on the H2 board.
Our DCDC design could supply up to 2.2Amp output.
But some old enterprise grade HDDs need 1.3Amp in-rush peak current.
If we attach two HDDs, 2.2Amp was not sufficient and it causes the stability issue from time to time.
So we have to change values of an inductor and several capacitors to increase the output power up to 2.7Amp.
We had no such high power consumption hard disks when we designed the H2 hardware unfortunately. :(

Anyway, after some experimental tests with various passive components, we found optimal values and simultaneous access with two HDDs has no issue now.
I think there are 5~6 users who have the HDD power stability issue and we will modify their boards by replacing some passive components.

Please send us your boards via cheap shipping method like USPS.
And we will fix your board quickly and ship back via fast Fedex or DHL with our cost.
Before shipping to us, please send an email to "odroid at hardkernel dot com". She will help your RMA process from early next week.
Really sorry for the inconvenience caused.

P/S. You have to consider using a PSU which can supply higher than 60Watt if you need to run CPU+GPU intensive work loads with heavy access to two enterprise HDDs even after hardware fix.

issamu
Posts: 4
Joined: Sat Jan 07, 2017 11:58 pm
languages_spoken: english
ODROIDs: odroid-c1
Contact:

Re: Odroid-H2 weak HDD power supply

Unread post by issamu » Sun Feb 10, 2019 5:18 am

Will the future boards have this revision?

yudouxie
Posts: 1
Joined: Mon Feb 11, 2019 12:06 am
languages_spoken: english
ODROIDs: H2
Contact:

Re: Odroid-H2 weak HDD power supply

Unread post by yudouxie » Mon Feb 11, 2019 12:10 am

what's the address to ship the board back to? thanks.

User avatar
odroid
Site Admin
Posts: 29994
Joined: Fri Feb 22, 2013 11:14 pm
languages_spoken: English
ODROIDs: ODROID
Contact:

Re: Odroid-H2 weak HDD power supply

Unread post by odroid » Mon Feb 11, 2019 11:39 am

yudouxie wrote:
Mon Feb 11, 2019 12:10 am
what's the address to ship the board back to? thanks.
Please send an email to "odroid at hardkernel dot com" first.
They will let you know the process as well as shipping address.

User avatar
odroid
Site Admin
Posts: 29994
Joined: Fri Feb 22, 2013 11:14 pm
languages_spoken: English
ODROIDs: ODROID
Contact:

Re: Odroid-H2 weak HDD power supply

Unread post by odroid » Mon Feb 11, 2019 11:44 am

issamu wrote:
Sun Feb 10, 2019 5:18 am
Will the future boards have this revision?
I think so. But there is no firm CPU delivery schedule from Intel yet. :(

User avatar
synportack24
Posts: 109
Joined: Sat Jan 10, 2015 10:16 am
languages_spoken: English, German, Korean
ODROIDs: C0, C1, C1+, C2, XU4
Location: 대한민국 - 안양시
Contact:

Re: Odroid-H2 weak HDD power supply

Unread post by synportack24 » Thu Feb 21, 2019 10:53 am

Has anyone run into any problems with Seagate Ironwolf series hard drives? Was looking to get two of those for the H2, but might play it safe and get WD reds.
http://magazine.odroid.com/
submit an article get cool stuff!

Post Reply

Return to “Issues”

Who is online

Users browsing this forum: No registered users and 1 guest