And here are the benchmarks both without the chuck mod and the same crappy dying harddisk. As you can see there is not a big difference between both boards because the circuit should be the same.
V1: Write Speed : 213,99 KB/s Read Speed : 248,67 KB/s 8K random, 70% read : 21,20 IOPS Sector random read : 60,37 IOPS Average seek, including latency, is 17,17 ms.
V2: Write Speed : 214,06 KB/s Read Speed : 243,11 KB/s 8K random, 70% read : 21,15 IOPS Sector random read : 65,05 IOPS Average seek, including latency, is 15,67 ms.
I have cleaned the messy 8-16 bit mux so its easier to see how it works. [attachment=0] Its a bit later than expected because i had some trouble with the "Xilinx ISE Webpack" that crashed a couple of times and corrupting my working files.
First start-up nothing happened apart from the harddisk led. After going trough the ISE build settings i found that it didn't include the ROM decoder when building the programming file. After fixing that problem the PC successfully loaded the BIOS. Now its hanging on the 8 to 16 bit mux inside the CPLD so that's something i will try to fix tomorrow.
[quote author="ian"] I sent one to Pietja already because he is a super builder :) J1mbo I want to send you one. Anyone else? [/quote] This is a real "Free PCB" :)
I better start buying some parts now.
When i got home from work through all the snow the PCB was waiting for me :)
Global pins are designed to efficiently carry clock signals all over the cpld so there is a minimum of delay between the different logic cells. They can also be used as general I/O pins.
Auto routers are evil Never use any auto-router except for very specific parts like ddr3 data buses and things like that and only use them if you know how to properly setup your auto-router.
[attachment=0] Placing a pin header for the module is an option if sourcing the individual chips gives problems. My idea was to hack my XT-IDE adapter v1a to test this module first.
If that works we can always use a bigger rom so both the bios'es can fit in a single chip
That atmel part is only a usb slave device so we cant connect a usb harddrive to it.
The CH375B can be a usb host but only for usb harddrives however there is a ready made non opensourse bios for it, the bad part is, it seems to be only available on ebay.
other parts that can do usb host are the FTDI Vinculum-II, a PIC controller with usb on the go or many of the usb to spi controllers like the MAX3421E that needs to connect to a microcontroller. But with all those options we need to write our own firmware/bios.
Routing the CF and IDE connector dont have to be that critical since this controller is designed for older pcs that don't go very fast having long traces don't matter all that much.
I just uploaded my CPLD source to the SVN, i haven't tested it because i dont have the PCB. I have added the 8to16 bit mux but that is just a big mess so don't be surprised if noting works the first time.
Anyone is free to fix my crappy CPLD schematic or write something new in verilog or VHDl.
I the meantime i got something new from eBay to make in a new DP project. [attachment=0] Its the CH375B ISA to USB Host chip, the manufacturer has a download with schematics for a couple of demo boards, a dos driver and a BIOS so your old 80X86 pc can boot from a usb drive. http://wch-ic.com/download/list.asp?id=57
If we ever make a version 3 of the XT-IDE adapter we could include this USB chip for even more functionality.