CFAG12864A-TMI-VN Connection


New member

I know, using a CFAGx with a PC's parallel port is unsupported. Moving on.

I've wired up my CFAG12864a to a parallel port based on the pinout in the cfag_wintest cpp comments, but when I run the program, all I get are dark blue pixels, (ie all pixels off, no data). Here's how my pinout goes:
LCD Pin To
1 Ground
2 +5v
3 -4v
4 Parallel Pin 16
5 Not Connected
6 Parallel Pin 1
7 Parallel Pin 2
8 Parallel Pin 3
9 Parallel Pin 4
10 Parallel Pin 5
11 Parallel Pin 6
12 Parallel Pin 7
13 Parallel Pin 8
14 Parallel Pin 9
15 Parallel Pin 14
16 Parallel Pin 17
17 Not Connected
18 Not Connected
19 +5v
20 20ohm resistor + 100ohm Pot to G (as in

Does this all seem correct so far?

Looking for additional LCD resources? Check out our LCD blog for the latest developments in LCD technology.
You shouldn't leave pins 5 and 17 floating. Connect 5 to ground, and 17 to +5v or an RC network (10K to +5, 1uF to G), as noted in the code file.
Last edited:


New member
OK, with those pins connected, I now get the image below with the WinTest software in Logo mode. (sorry for the somewhat blurry photo)

Any idea where I should be troubleshooting?



Your cable should be short, not more than a few feet.

All the multiple ground wires in the cable should be connected to the CFAG ground pin.

Check to see whether your parallel port I/O puts out a full 5v logic level, or only 3.3 volts. If 3.3v, you may need pullup resistors.

If you're using a cell phone camera, shoot from further away, and then crop the image.


New member
My parallel cable is only 18" and there's less than 6" of wire between my project box-mounted db25 connector and the LCD.

All grounds are connected to a common bus on the pc board I'm using.

My parallel port *is* only putting out 3.3v. If I manually set the parallel port to low on RS, RW, D7, D6, and D0, and make D1-D5, E and CS1 or 2 high, I can blank the display. Given this, and that 2 rows appear to be displaying correctly, is it possible that the less-than-5 logic voltage isn't a problem? I ask because I'm not sure how to pull those values up. The 10k resistors in look like what you're talking about but they only apply them on the cs , e, and rs lines.

It's not a cellphone camera - my office just has a very old digital camera.
Last edited:
The 3.3v port output shouldn't be a problem, I'm just tossing out ideas to see if any of them help. If you had a poor ground, different data patterns may change the ground level and input level. But that doesn't seem likely, based on what you say. Do you know for sure that none of the I/O lines are shorted to another?

Since there is no obvious hookup problem, it might be useful to look at the data and signal lines with a scope, while running a continuous loop writing to the display. Are you able to do that? The most important lines are E and the data bus.


New member
I just put our scope on the display while running WinTest repeatedly. None of the data or signal lines appear shorted to each other or power or ground. They all showed logic highs and lows during the test. I also double checked all other pins and don't have any shorts.

Interestingly, today, before I tested with the scope and for shorts, I mounted my display's power and db25 connectors to the metal box I am using, and now the two rows of data that showed up before no longer show. The display remains completely white. The only difference I can see now is that the db25 outer shield is now connected to the same ground as the rest of the circuit. I'm not sure why that should be making a difference. All the pins appear to be getting proper data.
The next thing to check is the signal timing. This is a rather slow display, interface wise. Be sure the data is on the bus pins for more than 200 nS before the E strobe falls. Note that the E width minimum is 450 nS, and the E cycle time is 1 uS min. Refer to the timing chart in the data sheet.


New member
I've verified the line timings on my scope and they seem to be adequately long as well. With everything put together I'm still just getting those two rows of text/graphics from the wintest demo.