Trouble

SATA trouble

32 22452
pieter  
hello,

A blue WD - connected to the sata port, but externally powered (own made cable), works without issues.  speed over Samba and gigabit: little bit higher than over USB2.

Conclusion: was the power (or the cable, but this is not likely).

Kind regards, Pieter

Glad to hear it works now!

MrK1982  
Well actually I'm testing with external powering and this seems to work ... going to run some speed tests after this post ...
For testing purposes I use the original cable (I have built an 'adapter' for this to an other 5V power source), this looks like there is a problem with the BPi's onboard powering ...
@mikronauts: You said you have got some schematics? Could you  send me PM with them attached? I'm interested if there is any power stabilizer between the USB power input and the two pin power output for SATA.

MrK1982  
Post Last Edited by MrK1982 at 2014-8-30 04:46

Because I'm an electronics engineer (and only a software enthusiast) I want to build my own bpi ups and therefore I would need an internal circuitry for powering the sata device. If there is a real problem in powering the sata device from bpi I am going to build up a pi power supply with two 5V outlets powered by a 12V supply. This even could ease up the ups design based on a 12V PB accumulator (7.2Ah) and some charging and switching circuitry.

MrK1982  
Post Last Edited by MrK1982 at 2014-8-30 03:08

@sashijoseph: Thanks a lot for providing the schematics!

Speed Test results: (The following values are mean values which are calculated by 10 consecutive measurements.)
First the most basic testing:

  1. root@dut:~# hdparm -t /dev/sdb1
  2. /dev/sdb1:
  3. Timing buffered disk reads: 279 MB in  3.02 seconds = 92,52 MB/sec
Copy the Code
This is quite a good value but tests only the fastest sectors of the hdd.
Going deeper now ...

First I assure that the hdd cache is activated:

  1. root@dut:~# hdparm -W1 /dev/sdb

  2. /dev/sdb:
  3. setting drive write-caching to 1 (on)
  4. write-caching =  1 (on)
Copy the Code
Then writing test data to partition. WARNING: !!! This destroys data on the chosen partition !!!
To work out this test ten times also means some stress to your hard drive, but this load should be acceptable.

  1. root@dut:/# dd if=/dev/zero of=/dev/sdb1 bs=512M count=2 oflag=direct
  2. 1024+0 Datensätze ein
  3. 1024+0 Datensätze aus
  4. 1073741824 Bytes (1,1 GB) kopiert, 28,4705 s, 37,7 MB/s
Copy the Code
A write speed of 37,7 MB/s also seems to be quite a good value.

Testing read out:

  1. root@dut:/#mkfs.ext4 /dev/sdb1
  2. [...]
  3. root@dut:/# mkdir dutsDUT
  4. root@dut:/# mount /dev/sdb1 /dutsDUT/
  5. root@dut:/# cd dutsDUT/
  6. root@dutdd if=/dev/zero of=tempfile bs=1M count=1024 conv=fdatasync,notrunc
  7. 1024+0 Datensätze ein
  8. 1024+0 Datensätze aus
  9. 1073741824 Bytes (1,1 GB) kopiert, 29,1335 s, 36,9 MB/s
  10. // IT is okay if this is a little bit slower, based on fs operations ...
  11. root@dut:/dutsDUT# dd if=tempfile of=/dev/null bs=1M count=1024
  12. 1024+0 Datensätze ein
  13. 1024+0 Datensätze aus
  14. 1073741824 Bytes (1,1 GB) kopiert, 10,7155 s, 93,1 MB/s
Copy the Code
resulting in a read out speed of 93,1 MB/s ... I am satisfied :-D

All this test have been done with external power source ... because this is nothing for me ... I'm going to analyse the internal powering problems now ... I write a post if i find out something interesting ...

Well actually I'm testing with external powering and this seems to work ... going to run some speed  ...
MrK1982 Posted at 2014-8-30 07:26


On banana pi board, output power to sata disk is filtered by coil. But the coil have too much resistance, and few of disk didn't start, or work bad.
You may change power adapter plugged to the banana pi (with little more voltage: 5.2-5.3V), or short circuit the coil on banana pi board...

Reply 27# sylwek2k

The ferrite bead used is rated at 1.5A wherein typical dcr values are around 0.1ohm max.
It's weird if the value is too high like you say.
Have you measured it's dcr(I don't have a low-ohms meter at hand)?

btw it's a nice find.

It's FB3 element. Near to SATA power adapter.

Maybe i haven't luck, and power adapter or SATA disk is working with resonant frequency of this coil... but i have 0.6 voltage drop on this element (with connected disk).

MrK1982  
FB3 should be a cbf (chip-bead-ferrit) which is technically more than just a coil. The DC resistance of this part should be nearly zero. But I also measured a voltage drop of 0.5 V on this part. So it must be wrong part or defective soldering ...

You have to log in before you can reply Login | Sign Up

Points Rules