PCI and PCI-E add on cards with parallel port that work out of the box

Excellent idea, Tom!

For testing convenience I have extended the Parallel Port Tester from BigJohnT to also test in "IN" and "X" modes.  

     

I have also polished the "showport" utility created by Kirk Wallace.

Now one can set and read all 8 parallel port modes from on-board and PCI/PCIe cards.

Instead of showing only the numeric (decimal) value of each register, now it shows the mode name and respective binary value.

Command line is back compatible with original version, but now it is also possible to reference the modes by index.

Compile "sudo gcc -O2 -o showport showport.c"run with "sudo ./showport (base1) (extended1) (mode) (base2) (extended2) (mode) ...

For example - "sudo ./showport 21D0 21C8 1 21C0 21B8 4"
Modes Indexes:    
    0 = Standard Mode
    1 = PS/2 Mode
    2 = Parallel Port FIFO Mode
    3 = ECP FIFO Mode
    4 = EPP Mode
    5 = Reserved
    6 = FIFO Test Mode
    7 = Configuration Mode
    s = PS/2 Mode
    e = EPP Mode
    n = No change

For terminology reference:  wearcam.org/seatsale/programs/www.beyondlogic.org/ecp/ecp.htm



It is not always like this, but for a Sunix SUN2212 these are the addresses I've got. You have to search for two 8 byte addresses. 

I know Tom has already tested this chipset, but I'll add some more details.

Model: PAR6408AXX100
Chipset: Sunix SUN2212
Interface: PCIe
Supports PS/2: YES
Supports EPP: NOT TESTED
Supports IN Mode: YES (In PS/2 Mode)
Supports X Mode: YES (In PS/2 Mode - Control port pins are open drain/collector)
Tested OS: linuxcnc-2.8.4 / Debian Buster
Tested Computer: Dell Optiplex 760





For IN and X Modes to work, mode has to be set to PS/2.

I couldn't figure out why, but a couple of times during my testing the mode changed itself to Standard Mode, whitch is not bi-directional. If one tries to use IN or X modes while in Standard Mode some pins won't work. 

Does anyone know what may be causing the ECR register to randomly reset to Standard Mode (0b000XXXXX) ? If I can't find a way to fix it I am thinking of setting a startup script to make sure the right mode is always in place.

Model: Naxos NX1PAR-PCI
Chipset: Moschip MCS9805CV
Interface: PCI
Supports PS/2: YES
Supports EPP: NOT TESTED
Supports IN Mode: YES (In PS/2 Mode)
Supports X Mode: YES (In PS/2 Mode - Control port pins are open drain/collector)
Tested OS: linuxcnc-2.8.4 / Debian Buster
Tested Computer: Dell Optiplex 760

What's the PCI ID of this card, that's the secret source.

With regards to the second port, you'll have to check how the second parallel port is enabled, some chipsets use strapping or use a eeprom for the config.
If you aren't getting the address of the second parallel port come up adding the connector and resistors may not enable it.
Find the data sheet first and work out how to config the chip first before getting out the soldering iron.

unknown wrote:

With regards to the second port, you'll have to check how the second parallel port is enabled, some chipsets use strapping or use a eeprom for the config.
If you aren't getting the address of the second parallel port come up adding the connector and resistors may not enable it.
Find the data sheet first and work out how to config the chip first before getting out the soldering iron.

You are right!

I tried to read the registers with the remaning adresses and it didn't work.  



I tried both Base/Extended ccd8/cce0 and cce0/cce8. Didn't work. They must be strapped, as you said.

 
 

 

unknown wrote:

What's the PCI ID of this card, that's the secret source.
 

9710:9805 - NetMos Technology - PCI 1 port parallel adapter

Thank you very much for the tip!

I forgot to mention you also look in /dev to see how many nodes are made for the parallel port, and also look at what dmesg reports .

sudo cat /proc/ioports

Will return addresses

unknown wrote:

I forgot to mention you also look in /dev to see how many nodes are made for the parallel port, and also look at what dmesg reports .

sudo cat /proc/ioports

Will return addresses
 

sudo cat /proc/ioports | grep parport returns only one set of addresses, (just to make sure)