Additional USB ports and Internal HD!

[parper]

Hi,
I want to share a mod that I did yesterday on the w500gp.

I wanted to add an internal 2,5 HD into the w500gp.
So I bought a USB/ide interface(already had a 2,5 80Gb disc) and decided to try to see how I might
insert the HD into the case.

The next challenge was to see If I could use the internal USB ports on the VIA VT6212L ( which is a PCI USB 2.0 Controller chip).
There are 4 USB ports on that chip and only two are connected.
So I got the Data sheet and found a couple of solder points to try out. USBP4+ and USBP4-.
See attach 1

Data sheet:
http://down.chinaeda.cn/upload/2007_...1218477472.pdf
I soldered the VCC and GND to one of the existing USB connections.
As long as the device uses 100ma i think it will be fine. Else I would think that the port will be closed(according to USB spec.).

See attach 2

I did the soldering and I needed to actually get external power to the HD from the 5v connection in the 500gp directly and that worked fine. Couldn't get it to work from the USB connection directly.

So this configuration seems to work and the final product looks like attach 3.

Cheers
/P

[tesseract]

Cool! just thinking about it. But i want to use SSD for better speed. Thanks for specs.

[Arise]

What about the speed of this HDD? Can you transfer at USB 2.0 speeds?

[parper]

Hi,
I have a couple of ideas on why there is a performace problem here.
Assume that the USB interface is capable of 480 Mbit/s.

1. The file system implementation would be one bottle neck.
Think about it, the USB is connected to the VIA chipset and from there the data is transported in to local memory. The cpu needs handle this and if you use the network it needs to pipe it through to a network hardware. Lots of bottle necks.

2. The main bottleneck would probably be the CPU when transfering lots of data. More memory wouldn't hurt either to have bigger buffers for handling the idle when transfering smaller files.

Anyone checked the load on the cpu when transfering much data?

/P

[calebz]

Im still waiting for the 480mbit screenshot.

I don't think its the CPU problem but its loaded 100% yes.
That doesn't really mean anything tho.
I believe it is the bus construction of the router causing this.
My theory is that all the stuff passes through all the buses possible before it gets out on Ethernet. In PC its simple DMA is used mostly so the transfer doesnt load your CPU. ( I think ^^)
I have tried several file systems to solve this and it didn't change much.

Attaching my record speed !

[jayajag]

@parper

nice work.

can you make some pictures for me where i can see of you fixed the hdd?
and can you say soemthing about the temps?.

best regards

[parper]

Hi,
I made my own pillars(had a couple of rubber cones of about 1cm) of some sort that is glued to the hd and the board with a glue gun.
You can put them anywhere as long as you see that the hd is supported.

You can start by glueing the stuff to the hd and then try to test fit it on to the board.
I can't take a photo of this because I don't have access to the device right now.

The heat is not a problem at all. I though it was going to be but as you notice the case has holes in it and the hd has about on 1cm of space between the board and it.

Cheers

/P

[parper]

The USB 2.0 standard is capable of running high speed (480MBit/s) and full speed (12MBit/s) so when a manufacture tells you that a device supports 2.0 it can be full speed if you are unlucky.
You always need to check if the device supports high speed.

Anyway to answer your question:
1. the usb connectors on the chip are USB 2.0 high speed ones, as the external ones.
2. So the limitation would be the usb to ide interface. But mine is a 2.0 high speed.
3. And the ide 2,5" disc is a couple of years old disc but it is probably working with 20-30 Mbyte/s (160 - 240 Mbit/s) so the interface should be holding up.

I haven't tried to do any benchmarking yet, but the general speeds would be as if you would connect an normal external usb drive and if there are any limitations it would be on the via chip it self.

/P

[al37919]

the real spead will be on the order of magnitude of 2-3Mbytes/s (which is definitely usb2 speed, but still far from the theoretical peak value)

[ABATAPA]

the real spead will be on the order of magnitude of 2-3Mbytes/s (which is definitely usb2 speed, but still far from the theoretical peak value)

In my tests - up to 5-6 MBps.

[Serpent]

Give us a proof, please!
In my benchmarks maximum was about 3,5 MBytes/s!

[ABATAPA]

Give us a proof, please!
In my benchmarks maximum was about 3,5 MBytes/s!

Use ext2/ext3 fs, large file blocks/buffers.

[parper]

What kind of test method are we talking about ?

The bestway would be to use some sort of standarized linux program that does this.
Else SAMBA is a good choice but that will depend on the network.

/P