7i43 connectivity problem

Hello LinuxCNC Forum. Any help you can provide with the problem described below would be greatly appreciated.

I received a Mesa 7i43-U-4 in the mail earlier this week and I'm now trying to get it to work with emc 2.4.7. I have experience with emc using Pico Systems Universal PWM Servo Controller but this is my first Mesa board.

I'm using USB to power the 7i43 and have the jumpers set as follows:

1,2,3 UP
4,5 DOWN
6 UP
7 DOWN

The 7i43 is connected to my pc's mainboard parallel port at 0x0378 (address and EPP mode verified in system logs) and the 3 LED's (PWR, /DONE, /INIT) come on when I plug in the USB cable. I used pncconf to generate the emc configuration. There is nothing else attached to the 7i43 at this time.

When I start emc, I get the following error when the firmware tries to load:

[789141.731091] hm2: loading Mesa HostMot2 driver version 0.15
[789141.737175] hm2_7i43: loading HostMot2 Mesa 7i43 driver version 0.3
[789141.739350] hm2_7i43.0: firmware: requesting hm2/7i43-4/SV8.BIT
[789141.742772] hm2/hm2_7i43.0: board has FPGA '3s200tq144', but the firmware in hm2/7i43-4/SV8.BIT is for FPGA '3s400tq144'
[789141.742787] hm2_7i43.0: board at (ioaddr=0x0378, ioaddr_hi=0x0778, epp_wide ON) not found!

From what I have been able to determine, the firmware mismatch is a phantom message and the real problem is that the 7i43 board can't be found at 0x0378. I have verified that the board is a 400 and not a 200.

I've tried 3 different parallel cables, replaced the ribbon cable going to the 7i43, and triple checked that the ribbon cable is installed so that the pins are correctly matched.

This seems like it should be really simple but I can't make it work.

What motherboard are you using and what LinuxCNC version?

I'm using version 2.4.7 originally installed from an ISO downloaded from linuxcnc.org. I'll have to check later today on the motherboard specifics but right now I can tell you that it's an "older" (c2007) Intel board.

Well that should be OK. I ask because some newer motherboards (the Intel D525 for example) do not set the parallel port into EPP mode even when the BIOS says its in EPP mode and also do not set the port address properly.

I do think W7 should be up so that the 7I43 power is not dependent on USB enumeration

I've tried it with W7 up and down but it doesn't make any difference. As a simple test to see if the system is able to connect to the 7i43, I ran the Parallel Port Tester with and without the board connected. When the board is NOT connected (ribbon cable removed from 7i43), all of the Inputs are green. When the board IS connected, all of the Inputs are red except for pin 15. Unless I am mistaken, that tells me that hal is at least aware that there's something at 0x0378.

In any case W7 should be up to reduce the number of variables (you dont want USB re-enumeration to power down the 7I43)

I wonder if your parallel port is truly in EPP mode. Can you get the motherboard number part number?

Also the fact that the port tester reads all 0s only when the 7I43 is connected is suggestive of a cable problem as the 7I43 does not drive the parallel port bus lines (other than D0) until the FPGA is not yet programmed

I dont think it could be an issue as the CPLD disables the USB chips outputs when in parallel mode
but you might try powering from an external 5V supply (or make a power only USB cable by disconnecting the data leads in the cable)

I found some extra parallel cables that I'll try when I get home.

this usualy happens when the bios of the mother board is not in epp mode
it's more down to the 7i43 not being fully reconised , make sure the 3 jumpers near the parallell port are in the up position i.e to the parrell port , and that the usb jumper is disabled also
if the motherboard has bios settings of epp only then use that rather than spp/epp mode ( or similar dual mode use)

i notice your settings of the 7i43 do appear to be correct , so i'd say it's more of a cable or bios settings being incorrect
make sure your parallell port cable is using a moulded lead is in fact all pins , some are cut down for printer use , or made for serial port use and do not work in this application

also have the 7i43 powered up before the pc may help , narrow down the problem as a test

No luck on any of the suggestions. Here's what I've done since the last post:

- moved W7 to UP (and left it that way from this point forward)

- tried 2 new parallel cables. Both are fully pinned.

- powered the board using an external 5v power supply - pin W6 DOWN

- tried using each of the 4 possible parallel card settings in bios (see MB description below). Interestingly, in any mode other than EPP the error message changed slightly to contain "/DONE is not low after CPLD reset!". That's what I would expect but it also leads me to believe that setting EPP mode in bios is allowing some communication with the 7i43.

[82813.790484] hm2: loading Mesa HostMot2 driver version 0.15
[82813.796696] hm2_7i43: loading HostMot2 Mesa 7i43 driver version 0.3
[82813.798824] hm2_7i43: /DONE is not low after CPLD reset!
[82813.798865] hm2_7i43.0: firmware: requesting hm2/7i43-4/SV8.BIT
[82813.804055] hm2_7i43: /DONE is not low after CPLD reset!
[82813.804085] hm2/hm2_7i43.0: failed to reset fpga, aborting hm2_register
[82813.804095] hm2_7i43.0: board at (ioaddr=0x0378, ioaddr_hi=0x0778, epp_wide ON) not found!

The motherboard is an Intel D915GUX. It has 4 possible settings for the parallel port in bios:

Output Only
Bi-directioal
EPP
ECP

I'm beginning to believe that there may be a problem with the physical parallel port on the MB. HOWEVER, I've tried using a PCI addon parallel card and it gets the same or similar results (I don't have those results available but could reproduce them). When using the addon card, I've verified that the card is in EPP mode (using the lsmod command) and added ioaddr=0xbc00 (card address) to the hal configuration.

Did you check the DB25 to hdr 26 cable? thats another possible issue

Also note that most PCI parallel port cards are based on the NetMOS/MOSChip chips which do not implement EPP properly and thus will not work with EPP peripherals