HP nc8000 - HDD caddy UDMA2 / Multiword-DMa2

Hello everybody!

I found your forum by google and was impressed by the great support and all the little tricks from your users for improving standard manufactured notebooks. So I hope, I can get help here for my problems, too.

I recently got a HP nc8000 as a gift from my father and wanted to change the main hdd drive along with some other upgrading steps. I bought a SATA-2-PATA Multibay HDD Caddy and a Seagate 250GB 7200rpm hdd at ebay.
At first, I put this caddy in the multibay slot and only got this drive running at UDMA2(as assumed). I tried BAR-edit (Intel 855PM datasheet for guidance -> no effects) and hdparm (improved reading speed, but same slow writing performance). No chance...

Second try... I removed the multibay connector from the caddy and replaced the optical drive with the modified caddy. The nc8000 has two IDE channels, optical drive & internal hdd on primary...multibay drive on secondary.
When it booted up I was happy to see the new hdd at UDMA5, but at the same time the internal hdd was declassified to Multiword-DMA2. You might imagine the 'insane' transfer rates.

Now I'm stuck with this caddy and don't know where to put it in. Is there any chance to get the caddy hdd and the internal hdd running at UDMA5 or am I forced to remove the internal hdd? Is it possible to get UDMA5 at multibay slot? Should I try to hardwire pin 34 (80/40-pin cable mode) to GND ?

As You see, I am at my wit's end and hoping for any suggestions which could improve the actual situation. Thanks...

Regards,
sticcio

 

Ok...as Read & Write Utility reported the multibay hdd caddy as 40pin-connection, I hardwired JAE50 pin34 to ground as suggested in another thread. Again, no luck! This prevented the notebook from booting up.

I also tried BAR-Edit, again -> offset 54: from 0x1011, 0x3033, ... to 0xf0ff. Same here, it simply doesn't work.

Another thing I noticed (& doesn't work) is the caddy led. The original was dead/shorted and I replaced it with a new red one. You know it already...nothing seems to work.

I'm more than frustated, now...!

-sticcio

 

Next try...Ubuntu 10.10 (Live-CD)!

New hdd on secondary IDE channel (aka Multibay): old hdd (PATA) on primary channel is detected / new hdd is missing (not detected)

New hdd on primary IDE channel (aka fixed ODD): new hdd (SATA, Master) is detected / old hdd (PATA, Slave) is missing (not detected)

Something strange is going on...

-sticcio

 

it sadly looks like your secondary ide channel is limited to multiword DMA 2.

I don't right now have access to my old nc8000, but i do have a caddy that can go into it, i'll have a butchers when i get home tonight to see if i have the same issues as you.

tbh i never bothered putting another drive into the nc8000, as i remember it was a pain messing with it.

The 855pm chipset does need the drivers installed on windows installing with the -overall -overide switches to be done properly.

what OS you running?

and do you have an external USB caddy for the optical drive? if you do, put the multibay caddy into it and see whats detected.

this will for sure tell us if the caddy itself is duff or not.

 

This GRUB2 config works for me on 8710w.

Not all lines are required, I simply looked how linux configures IDE controller and added it to GRUB

Code:

  setpci -d 8086:2850 54.l=3033
  setpci -d 8086:2850 40.b=5  
  setpci -d 8086:2850 41.b=A3
  setpci -d 8086:2850 43.b=0
  setpci -d 8086:2850 48.b=1
  setpci -d 8086:2850 4A.b=2  

If your IDE controller id is different from 2850 you sholud replace it whith correct one.

 

Hello Siorah & zerg32,

and thanks for coming to my aid.

No, you probably misunderstood me. The secondary channel (Multibay) is limited to Ultra DMA2 at the moment. Anyway, this is too slow for an OS partition.

If I do it the other way and replace the fixed optical drive on primary channel with the new hdd, I get the following result:

Primary Master - Seagate 250GB SATA - UDMA5
Primary Slave - Fujitsu 40GB - Multiword-DMA2

As hdd cloning tools reported, I got too many read errors. 2 hd drives on primary IDE channel won't work (thanks to HP for that wonderful piece of EMC design).

You're right, I forgot to mention my OS. I'm running Windows XP Pro and of course I installed Intels chipset driver. I tested both, the old v5.00.1012 provided by HP and newest v6.3.0.1007 by Intel.

See above, please. The caddy itself is running fine on UDMA5 as master on primary channel, but influences slave drive (internal fujitsu hdd).

IIRC the 8710w has no multibay slot. Is it necessary to edit IDE config with/for GRUB bootloader? Are you running a multiboot system?

I tried BAR-Edit and changed offset 54 of IDE controller to 0x3033 & other values...no visible effects.
If I put the hdd in the Multibay slot and "change" the actual DMA mode with hdparm to UDMA5, I got full reading speed of 80MB/s but slow writing speed of 30MB/s. On primary IDE channel and real UDMA5 I got 80MB/s (R) & 60MB/s (W).

I also tried several tools (HDClone, Clone-Maxx) of UltimateBootCD. All of these reported UDMA2(40pin-cable) on Multibay device, so I believe
this is a hardware issue. I'll try to get a pinout schematic of the Multibay connector to fix this, but I doubt to find anything.


I finally got two choices:

1. Kick out the old PATA hdd & only use the new hdd in the fixed ODD bay on primary channel/UDMA5

2. Keep everything & use the new drive in Multibay slot on secondary channel/UDMA2 (Thanks again to HP for your precious patronizing behaviour)


One last question: Is there any possibility to get the caddy led to work as it should? I googled for some kind of datasheet of Marvell 8xxxx chipset and found nothing.
Is there anybody out there who has detailed information on this chipset. Thanks a lot!

-sticcio

 

sticcio

could you please post your bar edit and hdparm -i screenshot.

I'm using GRUB2 to configure my IDE controller before loading Windows 7

as for 8710w it has no multibay slot, but there is a special caddy that allows connecting a hdd instead of CD/DVD.

Before configuring with bar edit, try to disable ide controller in windows

 

Here we go (all screenshots without notebooks internal PATA hdd)...

Seagate HDD in Multibay slot:



Seagate HDD in fixed optical drive slot:



-sticcio

 

Try writing 0x1033 instead of 0x3033 as explained to unlock the write speed.

 

Hello nando4,

I already read this thread (your experiences were the main reason for trying an upgrade) , but...

...and also 0x1033 don't has any effect. But I'll try to disable IDE controller before editing, tomorrow.

-sticcio

EDIT: Disabling the IDE controller or the primary/secondary channel isn't possible/allowed along as a drive is attached!

 

sticcio

as I understand you boot drive is attached to the same controller, so my trick with disabling wont work.

Are you familiar with GRUB2 ?
You can try to setup GRUB2 to start your windows, and then use it to alter PCI configuration.

I examined the schematics of your notebook ( http://www.4shared.com/document/Z497rDrq/HP_Compaq_NC8000_schematics.html)
on page 52 there is multibay connector pinout. Pin PDIAG that is used for 80pin cable/DMA66 detection is not connected, so the only way to enable DMA66 for the multiby will be direct controller configurtion.

 

You'd be able to effectively do the same thing by disabling the IDE controller in Windows, use BarEdit to change the PCI Configuration registers, then start the IDE driver. The process can be automated using devcon + BarEdit's 'startup' feature.

 

Unfortunately not...I looking for some tutorials right now. Is it necessary to install Linux to get GRUB2 to work?
Fortunately this notebook is not in use and not much-needed at the moment. I can test everything to find a solution.

This is awesome, thank you so much!! So I can now stop arguing with that damn HP support.

According to this information (p.42) by T13 technical committee, that's absolutely correct for 80pin-cable detection. I did also try to isolate JAE50-pin34, but that changed anything. I guess you're right and GRUB2 is my last option to get it to work with both hdds.

EDIT:

I can't disable the IDE controller, there's no option to. Is there a trick I don't know?

-sticcio

 

As I understand, it is not possible to disable the IDE controller if the boot disk is attached to it. I'm not able to confirm it because on my 8710w boot disk is attached to SATA controller.

 

Yes.. could be right. In that case an alternative to installing Linux to get grub2 would be to use DIY ViDock's Setup 1.x. Can install it on USB or as a disk image in Windows accessible via the Windows bootmenu. It can do the necessary setpci commands then chainload into windows. It can be done completely from ramdisk with no active disk accesses during the process.

If the PDiag pin is set for 80-pin mode then presumable the Windows driver would detect this and run the 66Mhz timings.

If you boot Linux and do a write test, eg "dd if=/dev/zero of=[file on slave HDD] bs=64k count=1000", does it get > 30MB/s writes? If so, then it would be an idea to evaluate how the IDE PCI Config space has been assigned by the Linux kernel.

 

sticcio

Could you please try do configure 54h register using BAR Edit and then simply restart windows without shutting down.

After restart check the DMA mode and the value of the 54h register.

 

No luck...

This is my baredit ini file:

Code:

[PCI]
Nb=1
Addr1=8000F954
Val1=00001033 <-- or 1011,3033,f0ff,...

...but after every restart the value of offset 54h is reset to default = 0x0000.

Uh, and I noticed Windows lists the hdd caddy as a removable drive. Is this a normal behaviour? It is the only disk installed and contains system partition.

-sticcio
----------------------
EDIT:

Ubuntu Syslog:

Code:

Feb 22 22:32:19 ubuntu kernel: [    3.504602] ata2.00: ATA-8: ST9250410AS, 0002SDM1, max UDMA/133
Feb 22 22:32:19 ubuntu kernel: [    3.504609] ata2.00: 488397168 sectors, multi 16: LBA48 NCQ (depth 0/32)
Feb 22 22:32:19 ubuntu kernel: [    3.504641] ata2.00: limited to UDMA/33 due to 40-wire cable
Feb 22 22:32:19 ubuntu kernel: [    3.536612] ata2.00: configured for UDMA/33

I guess, there's no chance to get it work @ UDMA5/80-wire cable?

 

sticcio and nando4

It seems that I missed one important thing that ICH4M has two IDE channels.

The HDD and fixed optical drive are connected to primary IDE channel and the multibay is connected to secondary IDE channel.
I looked into the ICH4M datasheet http://hackipedia.org/Platform/x86/Intel integrated chipsets/i852/252337.pdf

There is a description of register 54h on page 413.
To force slave channel to fast DMA mode we should set bits 15,14,7,6,2,3
Please check the datasheet, may be I missed something.

In linux world everything is simpler.
There is a discussion about the same problem https://bugs.launchpad.net/ubuntu/+source/linux/+bug/303429

 

Yes, you're right...please have a look at my first post in this thread.

I set bits 15-12 & 7-0 = 0xF0FF, but that changed anything.

Obviously not...

I tried some older distributions, same results:

Backtrack 3 (Slax, Kernel 2.6.21.5) -> UDMA2
Knoppix 3.8.1 (Debian, Kernel 2.6.11) -> UDMA2

It must be a hardware related issue.

-sticcio

 

ok i dug out my old girl and had a play with this, and the multibay is limited to a 40pin IDE connection like the ancient CD-roms used to have, so udma2 is your max.

i even followed some of the responses on the thread to see if i had any results.

but nothing at all.

not bad, my nc battery still lasts for 1.2hours lol

 

Hello Siorah,

and thanks for reanimating your old "beauty", even if it only confirmes our actual suspicions.

I've been thinking about it again and decided to connect the hdd caddy to primary IDE channel as a replace of the fixed optical drive.
So I still have the option to use an additional battery pack in Multibay slot.

This leaves me with the problem of Multiword-DMA2 on slave device (internal hdd) of primary channel. Any ideas to fix this issue or am I forced to remove the internal hdd due to too many read errors?

-sticcio

 

I know that it is too simple but maybe it will be usefull to use jumpers on HDDs?

 

hiya, sorry for not getting back to you sooner, been doing some research on this.

It seems the secondary ide channel (for CD-Rom) is limited in hardware to a multiword dma setup. it seemed that hp were shortsighted in the manufacture process by only thinking people would use CD-Rom drives in that port.

I've not found any workaround for this mate i'm afraid, but you can get ok speeds if you run the hdd as a usb hard drive, in fact i got quite a bit of an improvement over the internal option, but then it's not internal!

so am at the end of the tether, nothing really can be done

 

Me again...

Yesss...too simple Of course, all jumpers are set in the right way. I guess it is a missing connection/wire of hdd caddy JAE50 to MiniIDE connector (grounded or open pin 47 makes no difference).

I have the following configuration at the moment...

Code:

Primary Master - HDD caddy with Seagate 7200.4
Primary Slave  - empty

Secondary Master - Multibay caddy with DVD R/W

...and it runs very fast & stable. Sometimes I get some boot up failures, but that is not worth mentioning. The missing HDD activity LED (no signal from Marvell chipset) is more annoying and forces me to use a software solution.

I think of buying another caddy (one of the "Fenvi" labeled caddies)...maybe.

Anyways...I want to say thanks for all your help and good advices to relive an old lady's youth.

--
sticcio

 

This is how I got the system LED responding when my Marvell sata-to-pata caddy was accessing the HDD. Pin37 on the JAE50 connector is DASP (Drive Active / Slave Present) which will be asserted when the drive is being accessed. Assert it by wiring the LED +ve signal to that pin.

 

Thanks, I already knew this

Believe me, activity status is really dead. You get no signal from chipset even with a LED soldered to +/- on PCB.

BTW: I got a different PCB. It looks like this one (last pic before USB2.0)... Maybe, there's a failure in design.

--
sticcio

 

Hi again. Today i got 2nd HDD caddy and of course uDMA2 on it.
1. OP said he used HDParm to increase speed of reads. Was it on multibay? If it increased speed of reads then maybe it is not so desperate hardware issue/ maybe there is a chance?
2. You said about disabling controller. There are 2 ATA channels in Device manager not including single controller. Maybe disabling ATA channel 0 (primary) will fit your requires?
3. It is myonly computer at the moment so i can't risk. but perhaps disabling UDMA on ATA 0 channel and reboot will help to use UDMA5 on multibay?
Out of curiosity, you have 3 IDE outputs on nc-8xxx? i have nc6000.


EDIT\\\ HOW TO USE Hdparm ON WINDOWS TO SET UDMA5??

 

check this - DIY: Adding SSD or HDD storage using an optical bay caddy [ver 2.0]

this solution seems to be working fine.